Datasets:

---

ExponentialScience

/

DLT-Scientific-Literature

like 0

Follow

Exponential Science 4

ERROR: type should be string, got "https://doi.org/10.1007/s12063 021 00198 9\n\n# Managing disruptions and risks amidst COVID‑19 outbreaks: role of blockchain technology in developing resilient food supply chains\n\n**[Manu Sharma[1] · Sudhanshu Joshi[2] · Sunil Luthra[3] · Anil Kumar[4]](http://orcid.org/0000-0001-7571-1331)**\n\nReceived: 28 August 2020 / Revised: 5 April 2021 / Accepted: 17 June 2021 / Published online: 16 July 2021\n© The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2021\n\n**Abstract**\nThis study attempts to present a contribution of Blockchain Technology (BC-T) in managing disruptions, and risk caused by\nthe COVID-19 outbreaks and extending profound support in developing resilient Food Supply Chains (FSCs). The effects\nof the pandemic can be witnessed on global supply chains in their demand & supply side disruptions and in the changing patterns of consumer buying in the food industry. The assessment of the disruptive factors is required to explore the\npresent issues and promising resilient strategies to impart robustness to the FSCs for mitigating disruptions in the future.\nAn integrated approach of Fuzzy Analytic Hierarchy Process (FAHP) and Weighted Assessment Sum Product Assessment\n(WASPAS) is employed to assess the factors related to sourcing, lean, workforce, and flexibility, as well as evaluation of\nthe BC-T enabled FSCs resilient strategies that mitigate the effect of disruption during the pandemic. The findings exhibit\nthat ‘Sourcing related’ is the most affecting disruptive factor causing distress in the FSCs and ‘flexibility resilient strategy’\nis the most relevant resilient strategy for BC-T enabled FSCs. The BC-T acts as a catalyst in enhancing the flexibility, traceability, and shorter supply chain network structure that may help the FSCs to mitigate risk and disruption in the pandemic\nsituation. The BC-T helps the FSCs to control the demand and supply shocks, and supports the organization with real-time\nmonitoring and sharing information. This study provides insights to the decision-makers, managers, and other stakeholders\nto take significant decisions during an emergency.\n\n**Keywords Blockchain Technology (BC-T) · Disruption · Food Supply chains (FSCs) · Resilient strategies · FAHP ·**\nWASPAS\n\n### 1 Introduction\n\n\n\n- Sunil Luthra\nsunilluthra1977@gmail.com\n\nManu Sharma\n\nmanusharma@doonuniversity.ac.in\n\nSudhanshu Joshi\n\nsudhanshujoshi@doonuniversity.ac.in\n\nAnil Kumar\n\nanilror@gmail.com\n\n1 Guildhall School of Business and Law, London Metropolitan\nUniversity, London, United Kingdom\n\n2 Operations and Supply chain Management Area, School\nof Management, Doon University, Dehradun 248012,\nUttarakhand, India\n\n3 Ch. Ranbir Singh State Institute of Engineering &\nTechnology, Jhajjar 124103, Haryana, India\n\n4 Guildhall School of Business and Law, London Metropolitan\nUniversity, London, United Kingdom\n\n## 1 3\n\n\nDuring the crisis, the resiliency of the Food Supply Chains\n(FSCs) has become significant and needs consideration of\nthe decision-makers to adapt and adjust urgently to manage\ndisruptions and risk caused in the demand and supply side\nstreams (Hosseini et al. 2019). The SC have faced unprecedented challenges because of disruptions during Coronavirus outbreaks (de Sousa Jabbour et al. 2020). People all\naround had dealt with changing lifestyles from self-isolation\nto stockout situations. The pandemic has instigated panic\nbuying of essentials and triggered the consumer’s buying\nbehavior (Barrett 2020). And due to the severe virus spread\nall across the world, consumers feel safe and minimum risk\nin buying through online shopping. Several scholars have\ncategorised Supply Chain (SC) risk into operational and\ndisruption risks (Xu 2020). The risks in the context\nto the general disturbance in SC operations such as demand\nfluctuations are categorized as operational risks whereas the\n\n\n-----\n\nevents related to low frequency and high impacts are known\nto be disruption risks (Fahimnia et al. 2018; Ivanov et al.\n2018; Hosseini et al. 2019). One more category has been\nadded to the risks that is epidemic outbreaks which has the\ncharacteristics of high uncertainty and ripple effects (Ivanov\nand Dolgui 2020). The epidemic’s effect has highlighted\nseveral threats to firm’s viability during the last decade\n(Ivanov and Dolgui 2020; Karmaker et al. 2021).\nThe supply side of FSCs concentrated towards contactless\ndeliveries using real time ordering, just-in-time order fulfilment strategies. The pandemic has shown the significance\nof short supply chains and production at local level\n(Cappelli and Cini 2020). The core strategy of the FSCs in this\npandemic time is more reliant on digital technologies, real\ntime information sharing, collaboration, viable supplier networks, configurations, omni-channels platforms, and hybrid\nbusiness models (Moktadir et al. 2019). The industry 4.0\ntechnologies provide new opportunities to the supply chains\nto enhance their transparency and flexibility but it also needs\nto understand issues related adoption and implementation to\noptimize the profitability and efficiency in the SCs (Lohmer\net al. 2020)\nSeveral industries have been affected by the pandemic,\nFSCs is one of the most affected industry as it serves the\nessential needs of humans and also one of the fastest growing industries across the world. The Indian food industry\ncontributes more than 40% of India’s consumer packaged\ngoods (CPG) industry (Chowdhury et al. 2020). As the\nglobal FSCs are disturbed, the intermediaries; producers,\nsuppliers, logistics chains, buyers, and customers are supposed to be dealt directly without any middlemen (Filimonau\nand Naumova 2020).\nThe need of the hour is to adopt a structural change in the\n‘new normal’ era where high disruption will have to be handled through digital technology called Blockchain Technology (BC-T) that can perform even without intermediaries\n(Lakhani and Iansiti 2017). This BC-T is commonly called\nas a disruptive technology that obliterates processes and\ncan bring radical changes in business models (Leible\net al. 2019). The BC-T can contribute to achieve flexibility,\nstability, traceability, resiliency, minimizing risk, sustainability, and reducing cost (Hughes et al. 2019; Behnke and\nJanssen 2020). The ledgers used in the BC-T enabled systems is replicated and maintained by a number of identical\nhosts (Iansiti and Lakhani 2017). Once information is stored\nin one record, all the replicated copies are updated near realtime and are immutable. The updated records are verifiable\nthus eradicating the need for the middlemen verification and\ndevelops trust among the partners (Kshetri 2018). This adds\nto enhance efficiency and reduction in the cost by removing\nredundancy. BC-T has addressed Supply Chain (SC) pain\npoints across several industries, logistics, and counterfeit\nproduct identifications (Treiblmaier 2019).\n\n\nAlthough research on BC-T has gained attention in the\nlast decade, there is less focus on the factors enhancing the\neffects of disruption, and risks on FSCs and also the possible solutions to become resilient in the post pandemic era\n(Dolgui and Ivanov 2021). This study aims at developing\na holistic model for FSCs during the pandemic situation\nto enhance the organizational traceability, flexibility and\nresilience whereas previous research was limited to BC-T\napplications and measuring its effects. Thus, the research\nobjectives of the study are as follows \n- To assess the factors affecting FSCs in the disruptive\nenvironment during the pandemic.\n\n- To measure the contribution of BC-T in managing disruption, risk in FSC during the pandemic situation\n\n- To develop a holistic decision model for the FSCs considering disruption, risk, and efficient resilient strategy.\n\nThe current study attempts to develop a holistic model for\nFSCs using an integrated approach through Multi-criteria\ndecision-making methods (MCDM) - Fuzzy Analytic Hierarchy Process (FAHP) and Weighted Assessment Sum Product\nAssessment (WASPAS). The decision-making in the pandemic time is extremely complex and thus integrated methods\nare appropriate to be utilized for assessing key issues in BC-T\nenabled FSCs. The effects of the disruption risks are measured\nthrough FAHP, based on the experts’ judgment and evaluation\nof the resilient strategies through WASPAS method.\nThe organization of paper is in 6 sections. Section 2\nreviews literature on disruptions, risk and factors affecting FSCs. Section 3 imparts the research methodology\nundertaken followed by a proposed framework in section 4.\nSection 5 discusses the findings, research and managerial\nimplications. Section 6 summarizes he study and suggest\ndirections for future studies.\n\n### 2 \u0007Literature review\n\nThe Systematic Literature Review (SLR) is employed using\ntwo databases – The Web of Science (WoS) and Scopus\ndatabases. The search term (‘Disruption’ OR ‘Risk’ OR\n‘Resilience’) AND (‘Food Supply chain’s) AND (‘Blockchain’ OR ‘Blockchain technology’) had been used with\nthe term 2015-2020. This study has used the systematic\nliterature adapted from Garza-Reyes (2015) and Nadeem\net al. 2017. The databases were considered major sources\nof information to establish the understanding of BC-T application and FSCs scenario in last 5 years. The major criteria\nfor selection were to explore the BC-T adoption and FSCs\nrequirements for developing a holistic model for handling\ndisruption. The study has only included those articles that\nhave been published with direct focus and within the context\n\n## 1 3\n\n\n-----\n\nof BC-T and FSCs. The search retrieved 545 papers. After\neradicating the duplicates, 292 articles are shown. The\nabstracts were thoroughly read to identify the relation to\nthe research objectives. Finally, 44 papers were found to be\nappropriate and in the context to the objectives.\nThe flowchart of the SLR is shown in Fig. 1.\n\n#### 2.1 \u0007Disruptions and risk amidst COVID‑19 outbreak in food supply chains\n\nThe disruptions are rising extensively and extending\nthroughout the entire SC network since the corona outbreaks\noutbursts across the world. The perishable characteristics of\nthe FSCs make it more complex than the other SCs (Ali and\nNakade 2017). Since, last few years FSCs are facing challenges of price volatility, issues in food wastage, food security (Von Braun 2018; El Bilali 2019; El Bilali et al. 2019;\nQi et al. 2018), and governance (Gokarn and Kuthambalayan\n2017). The complexity of global SCs, low stock levels, fewer\nredundancies, and wide varieties of products are required to\nachieve operational efficiency. The operational efficiencies\nincrease the exposure to uncertainties related to risks and\ndisruptions (Chopra and Sodhi 2014). The redundancies and\nwide range of products together develop resilience in FSCs.\nThe corona outbreaks have profoundly disturbed the FSCs\nat each and every level. At least 265 million were at risk of\ngoing hungry during the lockdown. Due to the excessive\ndemand of food products, there was an increase in panic buying among the individuals and by traders also. The disruption\nwas more severe for perishable products (Coluccia et al. 2021).\nDuring COVID-19 pandemic, approximately 110 million\npeople were living in acute food insecurity (UN 2020). The\ndeveloping countries had bigger challenges where people\nwere facing acute hunger. People lost their job, employment\nand income and disruptive SCs and led towards double crisis food insecurity in developing countries. India’s food\ninsecurity was very poor even before the lockdown; India\nranked 102 in Global Hunger Index. India’s per capita GDP\nin purchasing-power-parity (PPP) terms being almost double\nof each of the neighboring countries such as Bangladesh,\nNepal and Pakistan (World Bank 2020).\nThe disruptions exist at several levels such as sourcing,\nflexibility, lead-time, and workforce safety concerns. The\nlack of workforce, transportation, and logistics has reduced\nthe level of production that consequently leading towards a\n\n**Fig. 1 A systematic literature**\nreview\n\nInitial\nliterature\nsearch n=545\n\n## 1 3\n\n\ndecrease in operational efficiency. There has been a question\nraised on the survivability of the SCs. To develop resilient\nSCs, firms have adopted lean and JIT practices that decrease\nin the inventory levels and help the firm to reduce their cost\nand manage disruptions. There has been a change in the SC\nconfigurations such as shifting towards SC networks to deal\nwith the disruptions (Ivanov and Dolgui 2020). The COVID19 effect on the SCs is severe and due to the change in the\nconsumption pattern inventory buffering is gaining hikes.\n\n#### 2.2 \u0007Blockchain‑enabled food supply chains amidst COVID‑19\n\nThe food chain initiates with farmers using farm supplies such\nas seeds, fertilizers, and machinery etc. The farmers transport the food through logistics providers directly or indirectly\nthrough storage or marketing. The farmer is limited to the\nprocessor and does not extend to the customer or even the\ndistributor. There is a lack of traceability & transparency in\nthis traditional SC (Garnett et al. 2020; de Sousa Jabbour\net al. 2018). In the Industry 4.0 era, digitization has enhanced\nthe performance level of the firms and technologies like additive manufacturing, Internet of Things (IoT), and BC-T has\nstrengthened the processes of the FSCs (Kamilaris et al. 2019;\nWong et al. 2020). BC-T is a decentralized platform that\nallows peer to peer direct transactions, eradicates the third\nparty and validates information. It has been observed that\nBC-T improves traceability and revolutionize the digitized\ncontemporary FSCs (Kouhizadeh and Sarkis 2018).\nIn BC-T, a list of transactions is recorded into a ledger\nover a given period and created ‘block’. Each transaction is\nkept into a block and each block is connected to the other\nblocks before and after it. These blocks are ‘chained’ together\nthough hashing function (Wang 2019). These chained blocks\nare immutable. The uniqueness of the BC-T is its ability\nto create a self-correcting system without any third party.\nInstead, the enforcement is executed through a consensus\nalgorithm (Min 2019). The Blockchains (BCs) may be in two\ncategories: public and permissioned BCs. The main benefits\nof the BC-T are disintermediation, transparency, security, and\nautomation (Tönnissen and Teuteberg 2020).\nBC-T enabled FSCs offers transparency to the partners\nwhich is essential to improve the traceability and authenticity of the food products (Leng et al., 2018). During\nthe pandemic, real-time tracking was very much required\n\n\n-----\n\nto know the source and tracking of products (Kim and\nLaskowski 2018). During the pandemic the collaboration\namong the partners is very crucial and thus transparency\nof BC helps in developing trust among the SC sellers,\nbuyers and manufacturers, and third party. The BC is also\nuseful in enhancing the efficiency of the organization by\ntaking preventive measures, reducing waste, operational\ncost, and better inventory management (Kharif 2016;\nKlimczuk-Kochańska 2018).\n\n#### 2.3 \u0007Blockchain technology and resilient food supply chains\n\nFSCs need collaboration and information sharing to\nenhance their resilience (Ambulkar et al. 2015; Bottani\net al. 2019, 2020). BC-T may act as an intermediary for\ninter-relationships among the SC actors (Crosby et al. 2016).\nDuring the pandemic time, FSCs may receive information faster with BC-T implementation. BC-T can connect\nto Industry 4.0 technologies and may help in optimization\nprocesses (Saberi et al. 2019).\nSC agility is the strategic approach to accept change with\nthe corresponding organization’s actions promptly. It is\ndetermined by visibility and velocity. In BC-T based enabled\nsystems agility can be enhanced by adding new partners,\ninformation sharing, and resources that mitigate risks in the\ndisruptive environment (Cole et al. 2019). Food products can\nbe tracked, and traced with the help of real-time information throughout the system that enhances the SC resilience\n(Tendall et al. 2015; Stone and Rahimifard (2018). Velocity is\nlinked to flexibility, as the pace of adaptation towards disruption is a key issue. BC-T can particularly influence the pace\nto discover from disruptions. Agility is highly influenced by\ncollaboration, integration, and communication through BC-T\n(Ivanov and Rozhkov 2019). The relationship between disruption and risks during pandemic and adoption of resilient\nFSCs facilitated by BC-T is exhibited in Fig. 2.\n\n#### 2.4 \u0007Research gaps\n\nThe BC-T contribution is multidisciplinary, such as tourism (Kwok and Koh 2019; Sigala 2020); consumer to consumer business model (Sigala 2017); distribution channels\nand trade (Önder and Treiblmaier 2018; Treiblmaier 2019);\nSmart hospitality (Buhalis and Leung 2018); sustainability (Gretzel et al. 2015); strategic management (Kewell and\nWard 2017); Healthcare (Sharma and Joshi 2021; Filimonau\nand Naumova 2020); SCs (Helo and Hao 2019; Chang\net al. 2020; Laskowski and Kim 2016). The resilience in\nSC strategies has been elaborated with high flexibility and\nagility (Christopher and Peck 2004; Francisco and Swanson\n2018). The significance of redundancy and flexibility in\nresilience is measured (Sheffi and Rice 2005).\n\n\nThe risk in SCs has also been discussed and proposed\nflexible strategies for mitigating risk, but still there is a need\nof an integrated approach for sustainable practices to manage\nrisk (Tang and Musa 2011; Choi 2020; Joshi et al. 2020).\nMonroe at al. (2014) discussed the SC vulnerabilities and\ndisruptions to understand the possible risk and developing\nrisk mitigation strategies. Sustainability issues need to be\naddressed for developing resilient SCs (Amui et al. 2017).\nThe resilience among FSCs is also evaluated (Zhao\net al. 2017; Gholami-Zanjani et al. 2021).\nThe resilient strategies have been discussed using several\nMCDM methods. Risk mitigation strategies have been analyzed using DEA and non- parametric statistical methods\n(Talluri et al. 2013). Despite the short span of time since\nthe COVID-19 outbursts, there are research studies that\nfocused on healthcare, consumer’s decision-making, GSCs,\nand FSCs (Petetin 2020). Though the BC-T contribution,\ndisruption in FSCs, and resilience among SCs have been\ndiscussed by the researchers in past but till date, no study has\nbeen conducted to evaluate the disruption and risk related\nfactors affecting FSCs under COVID-19 environment. Also,\nhow BC-T can enhance the resilience in FSCs is still unexplored. This study bridges this gap by assessing the factors\naffecting FSCs and also the BC-T contribution in developing\nresilience.\n\n### 3 \u0007Research methods\n\nFuzzy AHP (FAHP) and WASPAS methods are employed\nin the current study. With the help of the systematic literature review, the factors (main four criteria and fourteen\nsub-criteria) are assessed using. These factors are evaluated\non the extent of effects raised by the pandemic disruption\nin the environment. The BC-T enabled resilience strategies\nthat may enhance the efficiency and survivability of FSCs.\nA total of twelve alternatives are evaluated using WASPAS\nmethod. The following sections elaborates the methods\nundertaken in the study.\n\n#### 3.1 \u0007Fuzzy analytic hierarchy process (FAHP)\n\nIt is a complex task for the decision-makers to decide on\nthe multifaceted problem, as a number of uncertainties arise\nduring the analysis of the problem. In a complex situation,\nMCDM methods are significant for assessment and choosing\nthe best alternative. The FAHP is an effective assessment\nmethod that is used to collect the responses from the experts\n(Wang et al. 2019a, b, 2020).\nIn this study, FAHP examines the factors affecting FSCs\nin the disruptive environment. The pairwise comparisons are\nmade using Triangular Fuzzy Numbers (TFNs), employed\nto evaluate and obtaining the weights. Fuzzy set theory is a\ngeneral form of the crisp values, and fuzzy set numbers only\n\n## 1 3\n\n\n-----\n\n**Fig. 2 Disruption, blockchain technology and resilient food supply chains amidst COVID-19**\n\n\nconsider the values range from 0 and 1, where 0 signifies\nthe non-membership function and 1 denotes full membership function. The TFNs are very helpful in fuzzy situations.\nThe TFNs scale used for the current problem is exhibited\nin Table 1.\nThe steps of FAHP are as follows:\n\n**3.1.1 \u0007Establishing pairwise comparisons**\n\nThe responses from the experts are collected using linguistic\nscale; the pairwise comparisons for criteria and sub-criteria\nare developed. Each expert is asked to respond for each criterion and sub-criteria.\n\n\n⎤\n⎥, ...A[∼][k] =\n⎥\n⎥⎦\n\n\nFrom Eq. 1, the pairwise matrix is formed where k represents the experts who are requested to assess the factors.\nEach element aij-K of the pairwise comparison matrix A-K\nrepresents the fuzzy number corresponding to the linguistics\nscale.\n\n**Table 1 Linguistic variable and TFNs (Kaganski et al. 2018)**\n\nNo. Linguistic variable TFNs\n\n1 Equally significant (ES) (1,1,1)\n2 Equally to average significant (EAS) (1,2,3)\n3 Averagely significant (AS) (2,3,4)\n4 Averagely to strongly significant (ASS) (3,4,5)\n5 Strongly significant (SS) (4,5,6)\n6 Strongly to very strongly significant (SSS) (5,6,7)\n7 Very strongly significant (VSS) (6,7,8)\n8 Very strongly to extremely significant (VES) (7,8,9)\n9 Extremely significant (EXS) (9,9,9)\n\n\n1 _a[∼][1]12 ... a[∼][1]1n_\n_a[∼][1]21_ 1 _a[∼][1]2n_\n∶ ∶ ∶\n_a[∼][1]n1 a[∼][1]n2 ..._ 1\n\n\n1 _a[∼][k]12 ... a[∼][k]1n_\n\n⎤\n\n_a[∼][k]21_ 1 _a[∼][k]2n_ ⎥\n∶ ∶ ∶ ⎥\n_a[∼][k]n1 a[∼][k]n2 ..._ 1 ⎥⎦\n\n(1)\n\n\n⎡\n⎢\n⎢\n⎢⎣\n\n\n_A[∼][1]_ =\n\n\n⎡\n⎢\n⎢\n⎢⎣\n\n\n## 1 3\n\n\n-----\n\n**3.1.2 \u0007Developing aggregated fuzzy pairwise comparison**\n**matrix**\n\n_A[∼][1]_ = ⎡⎢⎢ _a[∼][1]21[⊕]..∶k1.[⊕][a][∼][k]_ 21 _a[∼][1]12_ _⊕...1k_ _⊕a[∼][k]_ 12 ... _aa[∼][∼][1][1]_ 12∶nn⊕⊕......kk _⊕⊕aa[∼][∼][k][k]_ 12nn ⎤⎥⎥ (2)\n\n⎢ ⎥\n\n_a[∼][1]n1[⊕]...[⊕][a][∼][k]_ _n1_ _a[∼][k]_ _n2[⊕]...[⊕][a][∼][k]_ _n2_ 1\n\n⎢⎣ _k_ _k_ ⎥⎦\n\nThe aggregated fuzzy pairwise matrix is developed as\nshown in Eq. 1.\n\n**3.1.3 \u0007Defuzzifying the pairwise comparison values**\n**and checking consistency ratio**\n\nDefuzzifying the values by graded mean integration approach\nchecks the consistency of the fuzzy aggregated pairwise\ncomparison matrix. The results need to be re-evaluated,\nin case the result is not consistent.\n\n**3.1.4 \u0007Computation of weights**\n\nThe fuzzy geometric mean is calculated according to Eq. 2.\nThe fuzzy geometric mean of the first parameter of the TFNs\nin each row is calculated as follows:\n\n\nthe alternatives (Mardani et al. 2017). A decision matrix\nis developed where n is the number of alternatives, m represents the evaluation criteria and ­Xij represents the performance of i[t][h] alternative with respect to ­j[th] criterion. The\nfollowing steps are undertaken for evaluating the alternatives\n(Table 7).\n\n**3.2.1 \u0007The category of criteria is defined**\n\na) If beneficial criteria,\n\n\n∼\n_xij =_\n\n\n_xij_\n\n_maxxij_\n\n\nb) If non-beneficial criteria\n\n\n(5)\n\n(6)\n\n\n∼\n_xij =_\n\n\nminxij\n\n_xij_\n\n\n**3.2.2 \u0007Computation of total relative importance of ­i[th]**\n**alternative of WSM**\n\n∑m ∼\n\n_Q[(]i[1][)]_ = _i=1_ _xijwj_ (7)\n\n\n(3)\n\n\n_ai1 = [1Xai12X...Xai1n]_\n\n_ai2 = [ai12X1X...Xai2n]_\n\n\n1\n_n_\n\n1\n_n_\n\n\n**3.2.3 \u0007Computation of total relative importance of ­i[th]**\n**alternative of WPM**\n\n∑m ∼\n\n_Q[(]i[2][)]_ = _i=1_ (xij)wj (8)\n\n\n_aii = [ain1Xain2...X1]_\n\n\n1\n_n_\n\n\nThe geometric mean of second and third parameters of\nTFNs of each row is calculated similarly using Eq. 2.\n\n**3.1.5 \u0007Computation of fuzzy weights**\n\nThe fuzzy criteria weights are calculated as Eq. 4\n\n\n∑n\n\n_j=1_\n\n\n∼\n_xijwj+0.5_\n\n\n(∼\n_xij_\n\n\n_Qi = 0.5Q[(][1][)]_ + 0.5Q[(][2][)] = 0.5\n\n\n∑n\n\n_j=1_\n\n\n**3.2.4 \u0007Final weights calculation**\n\n\n)wj\n\n(9)\n\n(10)\n\n\n)wj\n\n\n∑n ∑n\n\n_j=1_ _[x][ij][w][j][ + (][1][−)]_ _j=1_\n\n\n(̃xij\n\n\n_Qi = Q[(]i[1][)]_ + (1−)Q[(]i[2][)] =\n\n\n∼\n\n_W =_\n\n\n∼\n_W1_\n∼\n_W2_\n∶\n∼\n_W_ _n_\n\n\n=\n\n\n∼\n( al1\n\n_aus_ [,][ a]a[m]ms[1] [,][ a]a[u]ls[1]\n\n( al2\n\n_aus_ [,][ a]a[m]ms[2] [,][ a]a[u]ls[2]\n\n∶\n( aln\n\n_aus_ [,][ a]a[mn]ms [,][ a]a[un]ls\n\n\n)\n\n)\n\n)\n\n\n(4)\n\n\n#### 3.3 \u0007Selection of experts and data collection\n\nThe current study has undertaken 12 professionals from the\nFSCs in India. These experts are aware about the effects\nof disruptions, issues, and difficulties faced by the FSCs.\nThe experts are selected on the basis of their experience\nin food industry. These professionals are also competent to\nforesee the future of the food industry in the long run and\nin post pandemic situation. The panel of experts consists of\nsmall and long SCs to understand the different perspectives\nof the food industry in the current scenario. The panel also\nincluded IT professionals who have practiced BC-T implementation in the firms for developing more resilient SCs in\n\n## 1 3\n\n\nThe fuzzy weights are obtained and are undertaken for\nalternative evaluation using WASPAS method in section 3.2.\n\n#### 3.2 \u0007Weighted assessment sum product assessment (WASPAS)\n\nWASPAS includes two different models a) Weighted Sum\nModel (WSM) and b) Weighted Product Model (WPM).\nThis method is the most suitable method for evaluating\n\n\n-----\n\nthe last few years. The panel consists of 2 professionals in\nthe area of packed food, 2 purchase managers for supermarkets, 1 SC consultant in the food industry, 2 are risks and\ncrisis management consultants, 2 IT experts, 2 corporate\nstrategists and 1 academician in the area of agri-food SC.\nThese experts are asked to respond on the questionnaire on\nthe linguistic scale for performing the pairwise comparison of the factors and the resilient strategies (Appendix-A,\nTables 9, 10, and 11). The data collection has been done\nthrough telephonic communication during April-May, 2020.\n\n### 4 \u0007Proposed research framework\n\nIt consists of a sequential procedure for FAHP and WASPAS\nmethod implementation. It is presented in Fig. 3.\nThe three phases of the research study are: a) Defining\nthe problem b) Application of FAHP to calculate weights of\nfactors affecting FSCs c) Application of WASPAS for evaluation of BC-T enabled resilient SCs FSCs.\n\n#### 4.1 \u0007Defining the problem\n\nThere are number of factors that are disrupting the FSCs\nduring the pandemic. With the BC-T inclusion, the resilient\n\n**Fig. 3 Proposed research framework**\n\n## 1 3\n\n\nFSCs can be developed. The study aims to select the most\nappropriate resilient strategy that a FSC should adopt to\nmitigate the effect of disruption. The experts are undertaken\nfrom the area of SCs, digital technologies, and FSCs. These\nexperts are the key decision-makers in the FSC industry.\nDuring COVID-19, the disruption affecting FSCs through\nfactors such as sourcing related factors, lead time related\nfactors etc. The goal of the decision-making problem is to\nevaluate the most preferred resilient strategy of BC-T enabled FSCs during the pandemic is at the first level. From\nthe systematic literature review, 14 factors are identified\nand categorized into four main criteria. At the second level,\nfour main criteria are sourcing related factors, lead-time\nrelated factors, flexible system and, workforce related factors. Each criterion has sub-criteria. The linguistic scales\nare also selected to receive feedback from the experts. The\nlinguistic Table formed is exhibited in Table 2. Similarly, the\nlinguistic scales for the sub-criteria are established based of\nthe response from experts.\n\n#### 4.2 \u0007Application of FAHP to calculate weights of factors affecting FSCs\n\nEach expert response is undertaken and aggregated value for\nthe factors are obtained. The method is followed sequentially\n\n\n-----\n\n**Table 2 Linguistic scale values**\nfor main criteria C1 C2 C3 C4\n\nE1 E2 ... E12 E1 E2 ... E12 E1 E2 .... E12 E1 E2 ... E12\n\nC1 ES ES ... ES VES SS ... VSS ASS ASS ... SSS ES AS ... ES\nC2 AS AS ... EAS ES ES ... ES SS SS ... ASS AS ES ... ASS\nC3 VSS SS ... VSS SS SS ... SSS ES ES ... ES ASS ASS ... SSS\nC4 EAS SS ... AS AS ASS ... ASS AS SS ... AS ES ES ... ES\n\n\nin subsection 3. By using Eq. 1 aggregated fuzzy pairwise\nmatrix is obtained, presented in Table 3.\n\n**4.2.1 \u0007Calculation of geometric mean fuzzy values**\n**for criteria**\n\nFrom Eqs. 1, 2, the fuzzy weights for the main criteria are\ncalculated. Using Eq. 3, the average weight and normalized\nweight criterion are obtained. For the main criteria and sub\ncriteria, average weight (Mi) and normalized weight (Ni) are\nobtained (Table 4).\nThe average weight (Mi) and normalized weight (Ni) for\nthe sub-criteria are obtained using the similar steps.\n\n**4.2.2 \u0007Calculation of global weights and ranking**\n\nThe global weights were computed based on the weights of\ncriteria and sub-criteria. The ranking is performed on the\nobtained global weights. The global weights of the main\ncriteria and the sub-criteria are shown in Table 5.\n\n#### 4.3 \u0007Application of WASPAS for evaluating the resilient strategies of BC‑T enabled FSCs\n\nThe response from the experts has been taken on the scale\n1-9, where 9 represent the highest priority and 1 represents\nthe least priority. The resilient strategies ­(R1-R12) are rated\nby the experts based on the highly prioritized to mitigate\nrisk and manage disruption during the pandemic time. The\nbeneficial criterion is decided by the experts for all the alternatives (R1- R12).\n\n**4.3.1 \u0007Developing decision matrix**\n\nWith the help of equations, Eqs. 5, 6, the beneficial criteria\nare defined for evaluating the alternatives. Each alternative\nis evaluated by the experts and decision matrix is formed.\n\n\n**4.3.2 \u0007Total relative importance (WSM and WPM)**\n\nUsing Eqs. 7, 8 the calculations for WSM and WPM are\nperformed. From both the models WSM and WPM, the final\nweights of the resilient strategies are obtained.\n\n**4.3.3 \u0007Ranking of alternatives**\n\nUsing Eq. 10, the final values for the alternatives (resilient\nstrategies) are obtained. The final weights ( Qi ) of the resilient strategies are shown in Table 6. The ranking is also performed to identify the highest prioritized resilient strategy\nto be adopted through BC-T inclusion during the pandemic\ntime.\n\n### 5 \u0007Findings and discussions\n\nThe FAHP results from Table 5 signify that sourcing related\n­(C3) is the most affecting factor in the disruptive environment during the pandemic. Sourcing related factors ­(C3)\nhas obtained the maximum weight (0.3084) followed by the\nflexibility factors with a weightage of (0.2457). During the\ncrisis, SCs in the food industry is facing challenges in sourcing factors due to the border restrictions worldwide, and lack\nof transportation. The stringent regulations, increasing cost\nof transportation, and less availability of shipping crews are\nenhancing the difficulties for the FSCs. The severity of the\nsourcing factors affecting FSCs is based on alternative suppliers, reliance on JIT, and inventory stock of food supplies\n(Pinner et al. 2020). The inability to provide a wide range of\nproducts ­(C3-3) has obtained highest global weight (0.1043)\nand ranked as the most crucial factor affecting the FSCs in\nthe current crisis.\nThe results also revealed that the suppliers are unable to\noffer a wide range of the food supplies and the resources\nfor the production during this pandemic due to the\n\n\n**Table 3 Aggregated fuzzy**\npairwise matrix for main\ncriteria\n\n\nC1 C2 C3 C4\n\nC1 (1.000, 1.000,1.000) (2.500,3.500,4.500) (4.910,5.910,6.910) (2.000, 3.800, 4.000)\nC2 (2.160, 3.080,4.000) (1.000,1.000,1.000) (2.750,3.750,4.660) (1.750, 3.500,4.410)\nC3 (4.830,5.830, 6.830) (4.160,5.160,6.160) (1.000,1.000,1.000) (4.250, 5.250,6.250)\nC4 (2.580,3.580,6.580) (2.660,3.660,4.580) (2.660,3.660,4.660) (1.000,1.000,1.000)\n\n## 1 3\n\n\n-----\n\n**Table 4 Averaged weight criterion (Mi) and normalized weight cri-**\nterion (Ni)\n\nCRI Mi Ni\n\nC1 0.259287771 0.245736556\nC2 0.227821177 0.215914508\nC3 0.325442444 0.308433773\nC4 0.242593903 0.229915163\n\ntransportation, logistics, and wide network of the supplier\nbase (Garnett et al. 2020).\nThe flexibility related factors are disrupting the global\nresources and affecting the wide network of suppliers.\nThough complex SCs are optimized to maximize the flow\nof resources but failure at one point in the network may\npropagate through the network and expose the FSCs to ripple effects (Dolgui et al. 2018). Failure in the complex SCs\nmay cause a deadlock in the network. The FSCs are vertically and horizontally coordinated where some FSCs are\ndepending on the wider organizational network. These FSCs\nare more prone to failures due to lack of input and other\nsupport functions. The upstream and downstream FSCs are\nfacing demand and supply shocks and changing consumer\npurchase decisions. The other factors related to the workforce are also enhancing the deadlock situations.\nThe COVID-19 has revealed the preparedness or readiness of the GSCs and lack of contingency planning. The\nGSCs are complex and there is need to monitor and measure\nthe practices of suppliers and buyers, but due to lack of digitization and transparency it is unable to obtain transparency,\ncommunication and information exchange (Marques 2019).\nDue to the lack of the collaboration among suppliers, FSC\nvulnerability has also increased. Because of increasing price\nelasticity and consumers’ fear towards shopping is a major\nconcern for the FSCs.\n\n**Table 5 Global weights and ranking**\n\nMain Local Sub- Local Global Ranking\nCriteria weight criteria weights weight\n\nC1 0.2457 C1-1 0.1867 0.0459 14\nC1-2 0.1893 0.0465 13\nC1-3 0.3627 0.0891 3\nC1-4 0.2613 0.0642 8\nC2 0.2159 C2-1 0.2575 0.0556 12\nC2-2 0.3460 0.0747 6\nC2-3 0.3964 0.0856 4\nC3 0.3084 C3-1 0.1915 0.0591 11\nC3-2 0.1937 0.0597 10\nC3-3 0.3381 0.1043 1\nC3-4 0.2767 0.0853 5\nC4 0.2299 C4-1 0.4179 0.0961 2\nC4-2 0.2716 0.0624 9\nC4-3 0.3105 0.0714 7\n\n## 1 3\n\n\nThe WASPAS results for the resilient strategies of BC-T\nenabled FSCs are shown Table 6. From the Table 6, it is\nclear that improving flexibility (R7) is the most significant\nkey resilient. This strategy has obtained a weight of 0.875.\nFlexibility of the SCs is their ability to absorb any change\ncaused by the disruption. The main advantage of the flexibility is that it encompasses the redundancy and thus more\nnumber of suppliers for same products with different risk\nis essential (Sheffi 2015). For long run, organizations need\nto enhance flexibility to respond quickly to the disruptions.\nThis strategy is essential to become resilient and preparing\nthe FSCs for future disruptions.\nThe flexibility in BC-T enabled FSCs will be relying neither on the identity of the participants nor on whether the\nparticipant change over time making the network flexible\nnetwork an independent from central authority (Kamilaris\net al. 2019). Resilience refers to the redundance o_f the\nstored information that leads to robustness against malicious\nattacks as well as censorship it is the aforementioned characteristics of trust, shared availability, low friction due to the\ncut-out of trusted middle men, peer verification, underlying\ncryptography, immutability, decentralization, redundancy,\nversatility, and the potential for automation that all blockchains share (Casino et al. 2020).\nVisibility is another resilience strategy in the FSCs (Purvis\net al. 2016). The study has shown that visibility is the key\nresilient strategy and ranked second with a weightage of\n0.0848. This finding is line with the previous study where\nvisibility of the FSCs with the BC-T inclusion reduced the\ninformation audit cost, enhanced the information sharing\nthat consequently increases transparency at the customer’s side and diminished the volatility of demand (Choi\net al. 2020). This is the advantage of BC-T that enhances\nthe overall efficiency of the organization is enhanced. BC-T\nalso supports the SCs in customers identification affected\n\n**Table 6 Ranking of Alternatives**\n\n\nAlternatives\n(Resilient strategies)\n\n\n**_Q[(]i[1][)]_** **_Q[(]i[2][)]_** **_Qi_** Ranking\n\n\nR1 0.206 0.169 0.375 12\nR2 0.359 0.345 0.704 6\nR3 0.346 0.323 0.669 7\nR4 0.433 0.415 0.848 2\nR5 0.226 0.219 0.445 11\nR6 0.304 0.291 0.595 9\nR7 0.439 0.436 0.875 1\nR8 0.409 0.383 0.792 4\nR9 0.412 0.405 0.818 3\nR10 0.342 0.307 0.649 8\nR11 0.400 0.355 0.754 5\nR12 0.310 0.281 0.591 10\n\n\n-----\n\nby the disruption. Thus, with enhanced visibility and real\ntime information of the SCs supported by BC-T positively\nimpacts the ripple effect that further helps the organization\nto remain effective and efficient during the crisis.\nThe third most significant strategy is the change in the SC\nnetwork structure. This resilient strategy (R9) has obtained a\nweight of 0.0818. Due to the pandemic disruption, the complex GSCs are not able to manage their operations due to the\nwide network structure and thus, with the help of BC-T and\nother digital technologies the SC will become shorter and\nmore efficient during any disruption. The shortening may\nboost the resilience in the FSCs. The velocity of the SCs is\nalso significant.\nIn the disruptive environment, there is also a need of\ntracking and tracing the system for enhancing the efficiency\nof the operations (Saberi et al. 2019). Thus, BC-T can support the organization to enhance the velocity for making the\nFSCs more efficient. The disruption has not only affected the\nsupplier networks but the consumer’s purchase decision has\nalso been changed. The shift has been visible in the buying\npatterns of the customers and thus new models need to be\ndeveloped (Galanakis 2020). The pandemic has shown an\nopportunity for the FSCs to serve the consumers with online\nbusiness models that may help them to be safe and secure\nwhile shopping.\nIt is clear that FSCs need to have innovative strategic\nplan to redesign the FSCs to become more resilient and\nsustainable in future. This pandemic has exposed the vulnerability of the SCs in terms of sourcing, labor, lack of\npreparedness etc. Hence, there is need for digital transformation, integrated decision-making by stakeholders, collaboration for developing more resilient FSCs Montecchi\net al. (2019).\n\n#### 5.1 \u0007Theoretical implications\n\nThe main objective was to assess the factors affecting\nFSCs in the disruptive environment during the pandemic\nand measure the role of BC-T in managing disruption,\nrisk in FSC during the pandemic situation. The current\nstudy contributes in many ways. First, the research has\ncompiled all the relevant studies on BC-T and extracted\nthe literature to identify the factor for managing the disruption in FSCs. The current study is the pioneer research\ntype that has identified factors affecting FSCs in the disruptive environment during the COVID-19 pandemic\nusing an integrated approach of FAHP and WASPAS\nin context to India. The study provided insights to the\nresearchers and practitioners to manage the factors in\ndeveloping BC-T enables FSCs. Finally, the developed\nframework may help the researchers to consider the factors and create some research models.\n\n\n#### 5.2 \u0007Managerial implications\n\nThere are many issues arisen due to the current crisis but\nthe main problem is the high level of uncertainty and complexity. The FSCs need to be quick, digitalized, shorter, and\ncollaborated efforts of all the stakeholders. Regardless of the\nsize of their organizations, there is a need to enhance flexibility, visibility, traceability and shorter SCs. This study has\nwide-reaching implications for the FSCs as it has enhanced\nthe understanding of factors and resilience strategies that is\nmanaging the disruptions in the environment. The decisionmakers need to redesign their business models now as the\ndigital platforms have taken a lead in the consumer’s preference list.\nThe sale of the products through e-commerce has gained\nmomentum significantly in the last year able to satisfy the\nconsumer needs. The digitization not only helps the organizations to sell but also provides an opportunity to manage\nthe inventory more efficiently. The BC-T enabled FSCs may\naccess real-time information that facilitates decision making. The changed environment where people are compelled\nto remain indoors, there has been a marked shift in the ecommerce transactions. The use of BC-T controls and\nenhances the traceability of the products and processes\nwith optimized opera\\1tions. The traceability has enabled\nthe FSCs to track their product sourcing and transportation,\nhelping the decision makers to reduce risk, uncertainty, and\nenhances the coordination among the SC partners. There\nis reduction in the unnecessary production & wastage of\nfood that helps in achieving the sustainable outcomes for\nthe organizations. In the current circumstances, the FSCs\nneed to become flexible for adopting new modes of buying\npatterns.\nThe firms should focus on developing efficient and\nadvanced risk identification measures. The current\nstudy shows that the suppliers are limited with food\nsupplies and the resources for the production during this pandemic due to the transportation, logistics,\nand wide network of the supplier base. This justifies\nthe need to develop BC-T enabled systems for developing resilient FSCs. The firms can develop SC risk\nmanagement approach to enhance their efficiency and\nface high-frequency-low-impact events. The disruption\nduring COVID highlighted the significance of collaboration and inter-collaboration resource sharing. This study\ndemonstrates the need for the SCs and the focus area\nfor the SC practitioners and decision-makers to focus\non flexibility, visibility, volatility, network structure of\nSCs, and new business models for managing disruptions\ncaused by the pandemic.\nThe BC-T has the potential to replace some workflows\nthat are currently captured by ERP systems. For BC-T\nimplementation in FSCs, an integration is required with\n\n## 1 3\n\n\n-----\n\nERP systems. The main concern is the integration of\nuser interface and integrating data structures that is generated and stored in blockchain. Moreover, standardisation becomes more important. When several members of\na supply network decide to implement their own proprietary blockchain and promote their use along their own\nsupply chain, it is conceivable that administrative complexity for upstream suppliers first drives costs before it\neven becomes unmanageable. Participatory governance\nmechanisms could facilitate the development and acceptance of a standard. Given the prospect of cost reductions\nand efficiency gains for all members of the supply network, the collective action initiative could be extended to\na multi-stakeholder initiative in order to gain moral legitimacy. A successful implementation of blockchain technology clearly needs a collaboration of all supply network\nmembers alongside the provision of support services such\nas infrastructure and training. It must be ensured that the\ntechnology serves all stakeholder interests, including data\nportability to conventional databases, standardization and\nparticipatory governance.\n\n#### 5.3 \u0007Unique contribution of the research\n\nThe study has assessed the factors affecting the FSCs in\nthe disruptive environment during COVID-19. This study\nis the first attempt to analyze the factors and resilient strategies for the BC-T enabled FSCs. The study has highlighted\nthe insights for the food industry professionals, consultants,\nand strategists and government organizations to manage\ntheir FSCs to become more resilient and prepared for disruptions. The study has also used an integrated approach\nFAHP-WAPAS for examining these factors and strategies. \n### 6 \u0007Conclusion\n\nThe impact of the pandemic is witnessed across all the\nindustries but the major impact is seen in the food industry.\nThe severity of the effects is visible across all the companies,\nbased on factors such as network structure, flexibility, agility, etc. FSCs suffer from uncertain dangerous disruptions\nleading towards huge financial losses. The perishability of\nproduct is to be considered while developing the network\nstructure of the FSCs.\nFSCs need to be supported through technology such\nas BC-T. The study aimed to explore the factors affecting\nFSCs in the disruptive environment during COVID-19 and\nthe resilient strategy to mitigate the risk, uncertainty, and\ncontrolling disruptions. Accordingly, the robust and resilient\nFSCs have emerged as the most significant characteristic\nsince last year.\n\n## 1 3\n\n\nThe pandemic has exposed the unprepared state of\nFSCs and inefficiency to control disruption during the\ncrisis. The FSCs are shifting towards shortening of their\nSCs, shifting towards local SCs, impending digital technologies, enhancing flexibility, traceability to manage\nthe disruption, enhances their efficiency, and developing\nrobust SCs for the future. This study has highlighted the\nbenefits of resilience during the crisis. The shorter SCs\nwith higher traceability of products and processes will\ndevelop stronger contingency plans for the future that may\nmanage GSCs efficiently. The changing needs of consumers have also created an opportunity for innovation in business models.\nThe study has a few limitations. Firstly, the inability to\nhave a face-to-face discussion with the experts as the data\nhas been collected remotely. The technical faults and connection issues that caused the disturbance in the process\ninterrupted the discussion. Secondly, the pairwise comparisons are based on expert judgments and hence the results\nmay be biased. Thirdly, the study may be extended for\nempirical analysis of the events in real networks for testing\nthe results of the study.\n\n### Appendix‑A\n\n**Table 7 List of Factors affecting FSCs in disruptive environment**\n\nMain Criteria Sub-criteria\n\nC1 Flexibility C1-1 Change in yields\nrelated C1-2 Change in production\nactivities\nC1-3 Complex network of\norganization\nC1-4 Wider supply chain\nnetwork\nC2 Lead time C2-1 Reduction in supply\nrelated base\nC2-2 Delay in lead time\nC2-3 Price elasticity\nC3 Sourcing C3-1 Souring efficiency\nrelated C3-2 Demand and inventory\nimbalance\nC3-3 Inability to provide\nrange of resources/\nproducts\nC3-4 Accessibility to inputs\nC4 Workforce C4-1 Workforce safety\nrelated concerns\nC4-2 Safety norms\nC4-3 Lack of workforce\n\n\n-----\n\n**Table 8 Resilient strategies for BC-T enabled FSCs**\n\nR1 Enhancing security\n\nR2 Business model innovation\nR3 Coo petition\nR4 Enhancing visibility and Traceability\nR5 Social capital competency\nR6 Contingency planning\nR7 Increasing flexibility\nR8 Enhancing velocity\nR9 Supply chain network structure\nR10 Developing logistics capabilities\nR11 Information sharing\nR12 Smart contracts\n\n\n**Table 9 Criteria pairwise**\ncomparison for Fuzzy AHP\ncalculation (Please provide\nresponse in the linguistic scale)\n\n\nC1 C2 C3 C4\n\nC1 ES\nC2 ES\nC3 ES\nC4 ES\n\n\n**Table 10 Sub-criteria pairwise comparison for Fuzzy AHP calcula-**\ntion\n\nC1-1 C1-2 C1-3 C1-4 .... C4-3\n\nC1-1 ES\nC1-2 ES\nC1-3 ES\nC1-4 ES\n... ....\n...\nC4-4 ES\n\n**Table 11 Matrix for WASPAS calculation (Please rate on the scale of**\n1-9, R1-R9 are resilient strategies mentioned in Table 8)\n\nResilient Max/Min Max/Min Max/Min Max/Min\nStrategies\n\nC1-1 C1-2 ------- C4-3\nR1\nR2\n...\n...\nR12\n\n### References\n\nAli SM, Nakade K (2017) Optimal ordering policies in a multi-sourcing\nsupply chain with supply and demand disruptions-a CVaR approach.\nInt J Logist Syst Manag 28(2):180–199\n\n\nAmbulkar S, Blackhurst J, Grawe S (2015) Firm’s resilience to supply\nchain disruptions: Scale development and empirical examination.\nJ Oper Manag 33–34:111–122\nAmui LBL, Jabbour CJC, de Sousa Jabbour ABL, Kannan D (2017)\nSustainability as a dynamic organizational capability: a systematic\nreview and a future agenda toward a sustainable transition. J Clean\nProd 142:308–322\nBarrett CB (2020) Actions now can curb food systems fallout from\nCOVID-19. Nat Food 1–2\nBehnke K, Janssen MFWHA (2020) Boundary conditions for traceability in food supply chains using blockchain technology. Int J\nInform Manag 52:101969\nBottani E, Murino T, Schiavo M, Akkerman R (2019) Resilient food\nsupply chain design: Modelling framework and metaheuristic\nsolution approach. Comput Ind Eng 135:177–198\nBottani E, Tebaldi L, Lazzari I, Casella G (2020) Economic and environmental sustainability dimensions of a fashion supply chain: A\nquantitative model. Prod 30\nBuhalis D, Leung R (2018) Smart hospitality—Interconnectivity and\ninteroperability towards an ecosystem. Int J Hosp Manag 71:41–50\nCappelli A, Cini E (2020) Will the COVID-19 pandemic make us\nreconsider the relevance of short food supply chains and local\nproductions? Trends Food Sci Technol 99:566\nCasino F, Kanakaris V, Dasaklis TK, Moschuris S, Stachtiaris S,\nPagoni M, Rachaniotis NP (2020) Blockchain-based food supply chain traceability: a case study in the dairy sector. Int J Prod\nRes 1–13\nChang Y, Iakovou E, Shi W (2020) Blockchain in global supply chains\nand cross border trade: a critical synthesis of the state-of-the-art,\nchallenges and opportunities. Int J Prod Res 58(7):2082–2099\nChoi T-M, Guo S, Liu N, Shi X (2020) Optimal pricing in ondemand-service-platform-operations with hired agents and\nrisk-sensitive customers in the blockchain era. Eur J Oper Res\n284(3):1031–1042\nChoi TM (2020) Supply chain financing using blockchain: impacts on\nsupply chains selling fashionable products. Ann Oper Res 1–23\nChopra S, Sodhi MS (2014) Reducing the risk of supply chain disruptions. MIT Sloan Manag Rev 55(3):73–80\nChowdhury MT, Sarkar A, Paul SK, Moktadir MA (2020) A case study\non strategies to deal with the impacts of COVID-19 pandemic in\nthe food and beverage industry. Oper Manag Res 1–13\nChristopher M, Peck H (2004) Building the resilient supply chain. Int\nJ Logist Manag 15(2):1–14\nCole R, Stevenson M, Aitken J (2019) Blockchain technology: implications for operations and supply chain management. Supply Chain\nManag 24(4):469–483\nColuccia B, Agnusdei GP, Miglietta PP, De Leo F (2021) Effects of\nCOVID-19 on the Italian agri-food supply and value chains. Food\nControl 107839\nCrosby M, Pattanayak P, Verma S, Kalyanaraman V et al (2016) Blockchain technology: Beyond bitcoin. Appl Innov 2(6–10):71\nde Sousa Jabbour ABL, Jabbour CJC, Hingley M, Vilalta-Perdomo EL,\nRamsden G, Twigg D (2020) Sustainability of supply chains in\nthe wake of the coronavirus (COVID-19/SARS-CoV-2) pandemic:\nlessons and trends. Modern Supply Chain Res Appl\nde Sousa Jabbour ABL, Jabbour CJC, Foropon C, Godinho Filho M\n(2018) When titans meet–Can industry 4.0 revolutionise the environmentally-sustainable manufacturing wave? The role of critical\nsuccess factors. Technol Forecast Soc Change 132:18–25\nDolgui A, Ivanov D, Sokolov B (2018) Ripple effect in the supply\nchain: an analysis and recent literature. Int J Prod Res 56:414–430\nDolgui A, Ivanov D (2021) Ripple effect and supply chain disruption\nmanagement: New trends and research directions. Int J Prod Res\n59(1)\nEl Bilali H (2019) Research on agro-food sustainability transitions:\nwhere are food security and nutrition? Food Sec 1–19\n\n## 1 3\n\n\n-----\n\nEl Bilali H, Callenius C, Strassner C, Probst L (2019) Food and nutrition security and sustainability transitions in food systems. Food\nEnergy Sec 8(2)\nFahimnia B, Jabbarzadeh A, Sarkis J (2018) Greening versus resilience:\nA supply chain design perspective. Transportation Research Part\nE:Logist Transpo Rev 119:129-148\nFilimonau V, Naumova E (2020) The blockchain technology and\nthe scope of its application in hospitality operations. Int J Hosp\nManag 87:102383\nFrancisco K, Swanson D (2018) The supply chain has no clothes:\nTechnology adoption of blockchain for supply chain transparency. Logistics 2(1):2\nGalanakis CM (2020) The Food Systems in the Era of the Coronavirus\n(COVID-19) Pandemic Crisis. Foods 9(4):523\nGarnett P, Doherty B, Heron T (2020) Vulnerability of the United\nKingdom’s food supply chains exposed by COVID-19. Nat\nFood 1–4\nGarza-Reyes JA (2015) Lean and green–a systematic review of the state\nof the art literature. J Cleaner Prod 102:18-29\nGokarn S, Kuthambalayan TS (2017) Analysis of challenges inhibiting the reduction of waste in food supply chain. J Clean Prod\n168:595–604\nGholami-Zanjani SM, Klibi W, Jabalameli MS, Pishvaee MS (2021)\nThe design of resilient food supply chain networks prone to epidemic disruptions. Int J Prod Econ 233:108001\nGretzel U, Werthner H, Koo C, Lamsfus C (2015) Conceptual foundations for understanding smart tourism ecosystems. Comput Hum\nBehav 50:558–563\nHelo P, Hao Y (2019) Blockchains in operations and supply chains:\nA model and reference implementation. Comput Ind Eng\n136:242–251\nHosseini S, Morshedlou N, Ivanov D, Sarder MD, Barker K, Khaled A.\nAl. (2019) Resilient supplier selection and optimal order allocation under disruption risks. Int J Prod Econ 213:124–137\nHughes L, Dwivedi YK, Misra SK, Rana NP, Raghavan V, Akella V\n(2019) Blockchain research, practice and policy: Applications,\nbenefits, limitations, emerging research themes and research\nagenda. Int J Inform Manag 49:114–129\nIansiti M, Lakhani KR (2017) Do Not Copy or Post., HBR, R1701J,\nJan-Feb 2017\nIvanov D, Dolgui A (2020) A digital supply chain twin for managing\nthe disruption risks and resilience in the era of Industry 4.0. Prod\nPlan Control 1–14\nIvanov D, Rozhkov M (2019) Disruption tails and post-disruption\ninstability mitigation in the supply chain. IFAC-Papers Online\n52(13):343–348\nIvanov D, Sethi S, Dolgui A, Sokolov B (2018) A survey on control\ntheory applications to operational systems, supply chain management, and Industry 4.0. Annu Rev Control 46:134-147\nJoshi S, Singh RK, Sharma M (2020) Sustainable agri-food supply\nchain practices: Few empirical evidences from a developing economy. Glob Bus Rev 0972150920907014\nKaganski S, Majak J, Karjust K (2018) Fuzzy AHP as a tool for\nprioritization of key performance indicators. Procedia Cirp\n72:1227–1232\nKamilaris A, Fonts A, Prenafeta-Boldύ FX (2019) The rise of blockchain technology in agriculture and food supply chains. Trends\nFood Sci Technol 91:640–652\nKarmaker CL, Ahmed T, Ahmed S, Ali SM, Moktadir MA, Kabir G\n(2021) Improving supply chain sustainability in the context of\nCOVID-19 pandemic in an emerging economy: Exploring drivers using an integrated model. Sustain Prod Consum 26:411–427\nKewell B, Michael Ward P (2017) Blockchain futures: With or without\nBitcoin? Strateg Chang 26(5):491–498\nKharif O (2016) Wal-Mart tackles food safety with trial of blockchain. https://​www.​bloom​berg.​com/​News/​Artic​les/​2016-​11-​18/​\n\n## 1 3\n\n\nWal-​Mart-​Tackl​es-​Foods​afety-​with-​Test-​of-​Block​chain-​\nTechn​ology (Visited on 16th November 2017)\nKim HM, Laskowski M (2018) Toward an ontology-driven blockchain design for supply-chain provenance. Intelligent Systems in\nAccounting, Financ Manag 25(1):18–27\nKlimczuk-Kochańska M (2018) Startups as a Source of Innovation in\nthe Agri-Food Industry. Mark i Rynek 2:21–30\nKouhizadeh M, Sarkis J (2018) Blockchain practices, potentials, and\nperspectives in greening supply chains. Sustainability 10(10):3652\nKshetri N (2018) 1 Blockchain’s roles in meeting key supply chain\nmanagement objectives. Int J Inform Manag 39:80–89\nKwok AO, Koh SG (2019) Is blockchain technology a watershed for\ntourism development? Curr Issues Tour 22(20):2447–2452\nLakhani KR, Iansiti M (2017) The truth about blockchain. Harv Bus\nRev 95(1):119–127\nLaskowski M, Kim HM (2016, July) Rapid prototyping of a text mining\napplication for cryptocurrency market intelligence. In 2016 IEEE\n17th International Conference on Information Reuse and Integration (IRI) (pp. 448-453). IEEE\nLeible S, Schlager S, Schubotz M, Gipp B (2019) A review on Blockchain Technology and Blockchain projects fostering open science.\nFront Blockchain 2:16\nLeng K, Bi Y, Jing L, Fu HC, Van Nieuwenhuyse I (2018) Research on\nagricultural supply chain system with double chain architecture based\non blockchain technology. Future Gener Comput Syst 86:641–649\nLohmer J, Bugert N, Lasch R (2020) Analysis of resilience strategies\nand ripple effect in blockchain-coordinated supply chains: An\nagent-based simulation study. Int J Prod Econ 107882\nMardani A, Nilashi M, Zakuan N, Loganathan N, Soheilirad S, Saman\nMZM, Ibrahim O (2017) A systematic review and meta-Analysis\nof SWARA and WASPAS methods: Theory and applications with\nrecent fuzzy developments. Appl Soft Comput 57:265–292\nMarques L (2019) Sustainable supply network management: A systematic literature review from a knowledge perspective. Int J Prod\nPerform Manag 68(6):1164–1190\nMin H (2019) Blockchain technology for enhancing supply chain resilience. Bus Horiz 62(1):35–45\nMoktadir MA, Ali SM, Paul SK, Shukla N (2019) Barriers to big data\nanalytics in manufacturing supply chains: A case study from\nBangladesh. Comput Ind Eng 128:1063–1075\nMontecchi M, Plangger K, Etter M (2019) It’s real, trust me! Establishing supply chain provenance using blockchain. Bus Horiz\n62(3):283–293\nMonroe RW, Teets JM, Martin PR (2014) Supply chain risk management: an analysis of sources of risk and mitigation strategies. Int\nJ Appl Manag Sci 6(1):4–21\nNadeem SP, Garza-Reyes JA, Anosike T, Kumar V (2017) \"Spectrum\nof Circular Economy and its prospects in Logistics\", Proceedings\nof the 2017 Symposium on Industrial Engineering and Operations\nManagement (IEOM), Bristol, UK 440–451\nÖnder I, Treiblmaier H (2018) Blockchain and tourism: Three research\npropositions. Ann of Tour Res 72(C):180–182\nPetetin L (2020) The COVID-19 Crisis: An Opportunity to Integrate Food\nDemocracy into Post-Pandemic Food Systems. Eur J Risk Reg 1–11\nPinner D, Rogers M, Samandari H (2020) Addressing climate change\nin a post-pandemic world. McKinsey Quarterly April\nPurvis L, Spall S, Naim M, Spiegler V (2016) Developing a resilient\nsupply chain strategy during ‘boom’and ‘bust.’ Prod Plan Control\n27(7–8):579–590\nQi X, Fu Y, Wang RY, Ng CN, Dang H, He Y (2018) Improving the\nsustainability of agricultural land use: an integrated framework for\nthe conflict between food security and environmental deterioration. Appl Geogr 90:214–223\nSaberi S, Kouhizadeh M, Sarkis J, Shen L (2019) Blockchain technology and its relationships to sustainable supply chain management.\nInt J Prod Res 57(7):2117–2135\n\n\n-----\n\nSharma M, Joshi S (2021) Barriers to blockchain adoption in healthcare industry: an Indian perspective. J Glob Oper Strateg Sourc\n14(1):134–169\nSheffi Y (2015) The power of resilience: How the best companies manage the unexpected. mit Press\nSheffi Y, Rice JB Jr (2005) A supply chain view of the resilient enterprise. MIT Sloan Manag Rev 47(1):41\nSigala M (2017) Collaborative commerce in tourism: implications for\nresearch and industry. Curr Issues Tour 20(4):346–355\nSigala, M (2020) Tourism and COVID-19: Impacts and implications\nfor advancing and resetting industry and research. J Bus Res\n117:312-321\nStone J, Rahimifard S (2018) Resilience in agri-food supply chains: a\ncritical analysis of the literature and synthesis of a novel framework. Supply Chain Manag Int J 23(3):207–238\nTalluri S, (Sri), Kull TJ, Yildiz H, Yoon J (2013) Assessing the Efficiency of Risk Mitigation Strategies in Supply Chains. Journal Of\nBusiness Logistics 34(4):253–269\nTang O, Musa SN (2011) Identifying risk issues and research\nadvancements in supply chain risk management. Int J Prod Econ\n133(1):25-34\nTendall DM, Joerin J, Kopainsky B, Edwards P, Shreck A, Le QB,\nSix J (2015) Food system resilience: defining the concept. Glob\nFood Sec 6:17–23\nTönnissen S, Teuteberg F (2020) Analysing the impact of blockchaintechnology for operations and supply chain management: An\nexplanatory model drawn from multiple case studies. Int J Inform\nManag 52:101953\nTreiblmaier H (2019) Combining blockchain technology and the physical internet to achieve triple bottom line sustainability: a comprehensive research agenda for modern logistics and supply chain\nmanagement. Logistics 3(1):10\nUN (2020) World Food Programme. Retrieved from un.org: https://​\n\nunite.​un.​org/​sites/​unite.​un.​org/​files/​techn​ovati​on/2_​wfp_\n​build​ingbl​ocks_​robert_​opp.​pdf\n\n\nVon Braun J (2018) Bioeconomy-The global trend and its implications\nfor sustainability and food security. Glob Food Sec 19:81–83\nWang S, Ouyang L, Yuan Y, Ni X, Han X, Wang F-Y (2019)\nBlockchain-enabled smart contracts: architecture, applications, and future trends. IEEE Trans Syst Man Cybern Syst\n49(11):2266–2277\nWang YS (2019) The challenges and strategies of food security under\nrapid urbanization in China. Sustainability 11(2):542\nWang Y, Singgih M, Wang J, Rit M (2019) Making sense of blockchain\ntechnology: How will it transform supply chains? Int J Prod Econ\n211:221–236\nWang Y, Xu L, Solangi YA (2020) Strategic renewable energy\nresources selection for Pakistan: Based on SWOT-Fuzzy AHP\napproach. Sustain Cities Soc 52:101861\nWong L-W, Leong L-Y, Hew J-J, Tan GW-H, Ooi K-B (2020) Time\nto seize the digital evolution: Adoption of blockchain in operations and supply chain management among Malaysian SMEs. Int\nJ Inform Manag 52:101997\nWorldbank (2020) Worldbank.org. Retrieved from Blockchain & Distributed Ledger Technology (DLT): https://​www.​world​bank.​org/​\nen/​topic/​finan​cials​ector/​brief/​block​chain-​dlt\n\nXu H (2020) Minimizing the ripple effect caused by operational risks\nin a make-to-order supply chain. Int J Phys Distrib Logist Manag\n50(4):381-402\nZhao G, Liu S, Lopez C (2017) A literature review on risk sources and\nresilience factors in agri-food supply chains. Work Conf Virtual\nEnt 739–752. Springer, Cham\n\n**Publisher’s Note Springer Nature remains neutral with regard to**\njurisdictional claims in published maps and institutional affiliations.\n\n## 1 3\n\n\n-----\n\n"

ERROR: type should be string, got "https://doi.org/10.1186/s42400 020 00068 0\n\n\n### RESEARCH Open Access\n\n\n# A PLS blockchain for IoT applications: protocols and architecture\n\n#### Alex Shafarenko\n\n\n**Abstract**\nThis paper proposes an architecture and a protocol suite for a permissioned blockchain for a local IoT network. The\narchitecture is based on a sealed Sequencer and a Fog Server running (post-quantum) Guy Fawkes protocols. The\nblocks of the blockchain are stored in networked Content Addressable Storage alongside any user data and validity\nproofs. We maintain that a typical IoT device can, despite its resource limitations, use our blockchain protocols directly,\nwithout a trusted intermediary. This includes posting and monitoring transactions as well as off-chain (post-quantum)\nemergency communications without an explicit public key.\n\n**Keywords: Blockchain, Guy Fawkes protocol, Post-quantum, HORS-OTS, LoRa, Concurrent transmission**\n\n\n**Introduction**\nThis article presents a Block Chain construction based\non the well-known Guy Fawkes Protocol (GFP)\n(Anderson et al. 1998) for digital signature, which we\nextend and bring to bear on Block Chain (BC) technology\nintended for a swarm of low-power devices (IoT things).\nThe primary purpose of this blockchain is to support an\nimmutable distributed ledger that ensures the authenticity and sequencing of user records posted on it. Financial\ntransactions for IoT are not our intention, but they should\nbe compatible with our approach.\n\n**Motivation We set ourselves the following design con-**\nstraints on behalf of the participating things:\n\n1. Post Quantum restriction, in particular no public key\ncrypto.\n2. Low power, low energy. Notice that the avoidance of\npublic key crypto is synergetic with this constraint.\n3. Low local storage. A thing may have a flash card\nembedded in it, but the use of the flash card eats into\nthe energy budget.\n4. Local communications with low bandwidth and\nshort messages. The security protocol should be\n\n[Correspondence: A.Shafarenko@herts.ac.uk](mailto: A.Shafarenko@herts.ac.uk)\nDepartment of Computer Science, University of Hertfordshire, Hatfield\nAL10 9AB, UK\n\n\nconducted via UHF broadcasts over the target area.\nAn effective adversary should have to radiate\nsignificant power (reliably over the legal limit) and\nexpose itself to triangulation.\n5. It must be possible for all things to authenticate the\nledger under realistic assumptions without relying on\na trusted intermediary.\n\nThe above constraints have not, to the best of our\nknowledge, been considered together, but are characteristic of low-end IoT devices. The Post Quantum restriction\nis currently a bonus (due to insufficient power of available\nquantum computing), but will doubtless become important at some point in not so distant future.\n\n**Challenge We believe the main security risk with our**\nscheme is the Denial of Service (DoS) attack. Due to the\nuse of local communications (constraint 4 above), DoS\nattacks only need to be impeded but not totally suppressed. The latter is impossible due to the possibility\nof an attacker’s physically jamming the communication\ninfrastructure.\nDoS is our only concern for the following reasons. Our\nproposed protocol does not permit an imposter to impersonate an IoT device to the blockchain manager provided\nthat the device was correctly enrolled on the blockchain\nand is somehow made physically secure.\n\n\n© The Author(s). 2021 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which\npermits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit\nto the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The\nimages or other third party material in this article are included in the article’s Creative Commons licence, unless indicated\notherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended\nuse is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the\n[copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.](http://creativecommons.org/licenses/by/4.0/)\n\n\n-----\n\nAnother possible threat is a weaker version of manin-the-middle whereby the man in the middle does not\nattempt to convince the legitimate recipient that he is the\nlegitimate sender; he instead receives the sender’s broadcast (i.e. radio) message while simultaneously jamming the\nsender’s signal so as to convince the legitimate recipient\nthat no message has been sent yet. The man in the middle then uses the received message to defeat the protocol\nand impersonate (spoof) the legitimate sender to the legitimate recipient. We call such attacks jam/spoof attacks\n(more detail given in the section “Original GFP”) and propose a countermeasure (see the section “Posting on the\nblockchain”).\n\n**Contributions of the paper:**\n\n - We have proposed a Guy-Fawkes type protocol as a\nbasis for a blockchain construction. We show that the\nprotocol achieves authentication, non-repudiation\nand secure sequencing of the blockchain blocks by\nitself (not relying on block content). The protocol\ndoes not require public key crypto or high-volume\ncommunication, both important for an IoT device\nthat is limited in its compute power and\ncommunication duty cycle.\n\n - We use the principles of GFP to construct a protocol\nfor an IoT device to post on the blockchain, again\nwithout public key crypto, but with guaranteed\nauthenticity and non-repudiation and similarly low\ncommunication costs.\n\n - Since the protocol by itself does not protect against\nDoS attacks we further proposed a network\narchitecture which reduces DoS exposure by\nexploiting the physical properties of a mainstream\ncommunication format: LoRa.\n\n - Finally, since our focus is on the IoT applications, we\nhave constructed an off-chain emergency\ncommunication mode that provides authentication\nand non-repudiation without communicating or\nstoring additional authenticated key material and\nagain, without public key crypto. This is achieved by\nre-using nonces from the main protocol. As a result\neach thing is able to send an emergency message\nbetween two blocks of the blockchain while enabling\nany observer of the blockchain to prove the\nauthenticity and provenance of the emergency\nmessage based solely on the already published blocks.\n\n**Structure The section “Basic protocols” introduces the**\nblockchain protocol. The section “System architecture”\ndefines the system architecture. The protocol for posting content on the blockchain is presented and discussed\nin the section “Posting on the blockchain”. In the subsequent section, the section “Enrolment and optimisations”,\n\n\nwe show how a user can be enrolled on the chain at a point\nother than the beginning. The next section puts forward\na solution for emergency communications, when the originator cannot wait for the next block of the blockchain\nto be published. The section “Related work” cites some\nrelated work and finally there are conclusions.\n\n**Basic protocols**\nThe Guy Fawkes Protocol family was first proposed by\na Cambridge group (Anderson et al. 1998) 10 years\nahead of Nakamoto. The original formulation is very clear\nand can be practically useful, but it was not specifically\nintended as a blockchain protocol. We will summarise\nit here for ease of reference, but we still recommend\n(Anderson et al. 1998) as it contains some important\nbackground and a useful discussion.\n\n**Original GFP**\nThe original GFP is a protocol intended for signing messages Mi, i = 1, 2, . . . using a series of secret codewords\n_Xi = 0, 1, . . ., which are revealed one after another as the_\nprotocol progresses. The initial codeword, X0, is authenticated out of band when it is revealed, e.g. by digital\nsignature.\nThe GFP is defined inductively as an unlimited series of\nrounds. At any round i ≥ 0:\n\n1. Select a random codeword Xi+1 and keep it\nconfidential.\n2. Form its hash h (Xi+1)\n3. Publish Zi+1 = h (Mi+1, h (Xi+1), Xi)\n4. Reveal the hash pre-image: Mi+1, h (Xi+1), and Xi\n\nHere h(·) is a cryptographic hash. The security properties of the GFP are based on the computational hardness\nof fitting a pre-image to a known hash, which is also the\nbasis of, for example, the well-known proof-of-work for\nthe Bitcoin blockchain, and is believed to be robust for a\nstrong enough hash function h(·) even Post Quantum.\nHow does it work? As this is a signature protocol, it\nis supposed to ensure that only the originator can run it\nsuccessfully to sign their messages. The originator is differentiated by their knowledge of X0 and, by induction,\nof every other Xi, since the introduction of Xi+1 in step\n1 of any round is accompanied by the disclosure of Xi in\nstep 4 of the same round. The verifier at step 4 is able\nto verify that the originator is who they claim they are\nbecause it is cryptographically hard to fit the pre-image\n_(Mi+1, h (Xi+1), Xi) to its previously published hash Zi+1,_\nwhich means that it is the same pre-image that was used\nby the actor that published Zi+1. Since a part of that preimage is Xi, which is the pre-image of the hash h (Xi)\npublished as step 4 of the previous round, i − 1, we conclude by the same logic, that the actor is the same as one\npresent at round i − 1. We now claim by induction that the\n\n\n-----\n\nactor revealing the pre-image in step 4 round i is the same\nas that in round 0. However, to participate in round 0, the\nactor has to know the signed value of X0 before that round,\nand only the legitimate originator does. This proves the\nprovenance of all messages Mi and excludes repudiation.\nHowever, the protocol is phrased in terms of “publish”\nand “reveal”, and as it often happens, the definition of\nthese terms can leave loopholes that nullify the security\nproperties.\n\n**Jam/spoof attack. The definitive paper (Anderson et al.**\n1998) treats step 3 non-specifically and gives newspaper advert publishing as an example of what is meant by\n“publish”. The tacit assumption here is that whatever the\noriginator publishes will be seen by verifiers with certainty\n**before the step-4 message is revealed. Therein lies the**\nprotocol’s main vulnerability.\nImagine a situation when an adversary is able to prevent\nthe verifier from receiving the newspaper in question, but\nthe adversary receives it. It then sends the counterfeit copy\nof the newspaper, in which the message is not included,\nto a verifier. The verifiers will be unaware that the step3 message has been published. When the step-4 message\nis revealed by the originator, the adversary learns Xi and\nis able to publish their own step-3 message impersonating the originator and using the value of Xi learned from\nthe originator when the latter revealed it. Then the adversary is able to send a step-4 message consistent with that\nstep-3 message and from this time on the current and all\nsubsequent messages are compromised.[1]\n\nWe call this the jam/spoof attack. Its existence highlights the fact that the verb “publish” in the protocol\ndescription means not only to send a message to the\nworld, but also to guarantee that it has been received.\nWhich is how “publish” differs there from the verb “reveal”,\nwhich assumes no such guarantee. The verifier expects to\nsee a reveal message after it has received a publish message. If the reveal message is not received before the next\n_publish message, the verifier will simply request one and_\nwill keep requesting it until it receives the correct content\n(the pre-image of the publish message).\nThe metaphor of newspaper is ideal for this kind of\n“publish” (provided that the verifier can at least access\na library copy of each newspaper issue in time for the\n_reveal message). In reality “publishing” is done by net-_\nwork broadcasting, and in the IoT case that is quite often\nradio broadcasting, where both jamming and spoofing are\nquite feasible. Finally, it is interesting to observe that the\njam/spoof attack is of the same kind as the classic manin-the-middle attack, since the adversary impersonates\nthe originator to the verifier by placing itself in between,\n\n1 Notice that neither step 3, nor step 4 require an authenticated channel; in\nother words, anybody can send these messages and the protocol is supposed\nto be able to determine which of them come from the genuine counterparty.\n\n\nhowever the behaviour is different from the classical\nunauthenticated public-key exchange.\n\n**DoS attack. An adversary can exhaust the verifiers’**\nresources without jamming the originator. If M step-3\nmessages and N step-4 messages are sent impersonating the originator, the verifier will have to perform up to\n_M × N verification actions, whereby_\n\ni the value Xi of the step-4 message is hashed to match\nwith the value of h (Xi+1) from the previous round. If\nthey don’t match, this step-4 message is invalid,\notherwise\nii the whole step-4 message is hashed to match with at\nleast one step-3 message.\n\nThis takes into account the fact that the true step-4\nmessage can be delayed as it is forwarded through the network, and that the adversary can receive it early. Then a\nlarge number of fake step-4 messages with the correct Xi\ncan be produced that pass check (i), making the amount\nof work closer to the limit M × N.\nWe will address both attacks in our version of GFP\nwhich we present next.\n\n**PLS protocol**\nWe will now introduce a similar construction optimised\nfor our purposes.\n\n**Goal Assume that a single transmitter is to broadcast a**\nstream of public, authenticated messages to an unspecified number of receivers. The following conditions must\nbe satisfied:\n\n1. It should be possible (ideally at low cost) for each\nreceiver to prove, without trusting any intermediary,\nthat the message was sent by the transmitter.\n2. It should be cryptographically hard for an attacker to\nmodify any message or to change the order of\nmessages without the receivers noticing\n3. The broadcaster should be able to send an unlimited\nnumber of messages without weakening the security\nof the previous two constraints (this is common with\nthe original GFP).\n\nNow to the protocol. It uses a standard cryptographic\nhash H(x) (e.g. SHA-256). All values except Mk are binary\nstrings of the same length as H. Additionally the protocol\nuses symmetric encryption Eq(p) which encrypts plaintext p under the key q producing a ciphertext, and its dual\ndecryption: Dq(Eq(p)) = p. This could be any standard\ncipher, e.g. AES128, suitably adapted to the key and text\nsize using one of the standard methods.\nThe protocol operates in steps according to the wallclock time. All receivers and the transmitter synchronise\n\n\n-----\n\ntheir clocks so that when the transmitter’s clock registers\na time tT, any receiver’s clock tR is no more than ϵ away:\n\n|tT − _tR| < ϵ ._\n\nThe transmitter broadcasts at regular intervals, t0, t0 +\n_τ_, t0 + 2τ, . . ., where τ ≫ _ϵ. Each broadcast consists of_\nthree messages of the same length as H: P, L and S; they\nare a proof, link and signature message, respectively. It is\nconvenient to think of them as being broadcast on three\ndifferent channels, or in three different time slots, or with\na tag that tells the receiver which message it is. The messages are not explicitly indexed, but it is convenient to\nthink of them as being indexed with the Broadcast Interval Number (BIN): BIN 0 corresponds to the interval [ t0 +\n_ϵ, t0+τ_ −ϵ], BIN 1 to the interval [ t0+τ +ϵ, t0+2τ −ϵ], etc.\nNote that the P, L and S broadcasts in the same interval\nare not mutually ordered.\nThe protocol is presented in Table 1. In or before the\nfirst interval, receivers obtain independent authentication\nof P1. In each interval k the transmitter creates a fresh random nonce, Nk+1, and keeps it secret until the end of the\nnext interval k + 1. One such nonce, N1, is created by the\ntransmitter before launching the protocol.\nAt every step the transmitter transmit the three messages mentioned above and each receiver attempts to\nreceive them. The link message is saved for the next step\nand the signature and proof messages are used in the\ncurrent one. Next the receiver executes the validation procedure detailed in the middle column, which consists of\none bitwise XOR calculation and one hash calculation.\nThe calculations involve the current proof message Pk\nand the link message Lk−1 saved at the previous step. The\nreceiver may have more than one candidate value for P\n\n**Table 1 PLS Protocol**\n\n\nand L which will be obtained by peer communication\nbetween receivers. The receiver will subject each (P, L)\npair to the validation procedure until it obtains the pair\nthat satisfies it. If no pair passes validation, the protocol\nfails due to denial of service.\nOtherwise the valid (P, L) pair is combined with the current signature message Sk to obtain a signed message hash\n_σk = H(Mk). From that time on, the receiver will treat a_\nmessage whose hash equals σk as having being signed by\nthe transmitter.\n\n**Security analysis of PLS**\n**Threat model.**\n\ni An attacker can force one or more receivers to\nreceive the attacker’s arbitrary message instead of the\none being transmitted by the transmitter, or prevent\na receiver from receiving the message at all.\nii However, the attacker cannot thus disrupt all\nreceivers and it cannot make any receiver conclude\nthat the broadcast did not take place. The former can\nbe achieved by delivering at least one copy of the\nbroadcast message by an alternative physical channel,\nand the latter by broadcasting messages on a public\nwall-clock schedule.\n\nPoint (ii) makes it possible for a receiver that the protocol determines has received an invalid message to solicit\nunauthenticated candidate messages from peer receivers.\nWe will show that using the protocol each receiver will\nbe able to select the genuine message out of a set of candidates. This makes threat (i) a DoS threat rather than a\ndata-integrity one.\nWe continue with Table 1.\n\n\n**BIN** **Transmit/Receive** **Verify** **Obtain**\n\n1 _L1 = H(N2) ⊕_ _N1_\n\n_S1 = EN1_ _(H(M1) ⊕_ _H(N2))_ _P1 out of band_\n\n_P1 = H(N1)_\n\n2 _L2 = H(N3) ⊕_ _N2_\n\n_S2 = EN2_ _(H(M2) ⊕_ _H(N3))_ _H(L1 ⊕_ _P2) = P1_ _H(M1) = P2 ⊕_ **DL1⊕P2** _S1_\n\n_P2 = H(N2)_\n\n3 _L3 = H(N4) ⊕_ _N3_\n\n_S3 = EN3_ _(H(M3) ⊕_ _H(N4))_ _H(L2 ⊕_ _P3) = P2_ _H(M2) = P3 ⊕_ **DL2⊕P3** _S2_\n\n_P3 = H(N3)_\n\n_. . ._ _. . ._ _. . ._ _. . ._\n\n_k_ _Lk = H(Nk+1) ⊕_ _Nk_\n\n_Sk = ENk_ _(H(Mk) ⊕_ _H(Nk+1))_ _H(Lk−1 ⊕_ _Pk) = Pk−1_ _H(Mk−1) = Pk ⊕_ **DLk−1⊕Pk** _Sk−1_\n\n_Pk = H(Nk)_\n\n\n-----\n\nAt BIN=1, the transmitter sends out the link, signature\nand proof messages, saves N2 and keeps it secret till the\nend of the next interval. The receiver receives and saves\nthe received messages. It uses the remaining time in the\ninterval to poll its peers to learn any alternative values of\n_L, S, and P should they be received (which may be due_\nto signal propagation problems, deliberate jamming or a\ncyber attack).\nAt BIN=2, the receiver receives P2 (possibly more than\none candidate value)and verifies that for some candidates\n_L1 and P2, H(L1 ⊕_ _P2) = P1. It means that these values_\nof L1 and P2 are genuine. Indeed, an attacker wishing to\nconvince the receiver that an alternative link message\n\n_Lˆ_ 1 = H( ˆN2) ⊕ _N1 ̸= L1_\n\nis genuine in order to make a forged signature for its own\nmessage _M[ˆ]_ 1\n_Sˆ1 = EN1_ _(H( ˆM1) ⊕_ _H( ˆN2))_\n\nwould have to have sent these messages at BIN=1, when\nonly the transmitter knows the value of N2, so the attacker\nwould have to use its own nonce, _N[ˆ]2 ̸= N2, and then force_\nthe receiver to receive _P[ˆ]_ 2 = H(N[ˆ] 2). To succeed at that, the\nattacker must be able to obtain N1 within interval 1, to use\nit in the _L[ˆ]_ 1 message. But all that is publicly known about\n_N1 then is its hash, H(N1) = P1._\nThat is the linchpin of the security of any Guy Fawkes\nprotocol, our version or the classic (Anderson et al. 1998)\nalike. The attacker has to find a pre-image of a public hash\nvalue in order to mount a successful attack. The chances\nof finding a pre-image are slim, 2[−][(][l][−][1][)]; for l = 256\nthe attacker would have to do around 10[77] hash calculations to find the pre-image on a classical computer. This\ncan be improved upon by quantum computing, reducing\nthe number to 2[−][l][/][2] ≈ 10[38], still reliably unfeasible and\ncertainly good enough for the application domain of the\nIoT.\nLeaving the DoS scenario aside, we assume that the\nreceivers succeed. Next the receiver uses the formula in\nthe last column of Table 1 to calculate H(M1) on all candidate S-messages recorded at BIN=1. All these values are\nconsidered equally valid, but only one of them, namely the\ngenuine value of H(M1), where M1 is a message known to\nthe transmitter at BIN=1, can ever be used. An attacker\ncorrupting message S1 can only achieve denial of service:\nto fit a message to an arbitrary hash value of it (for example, by extending the message with a non-data-bearing\ntail) is computationally as difficult as it is to fit N1 to a\nknown H(N1).\nThe reader may wonder about the purpose of encryption. In the original Guy Fawkes protocol only hashes\nare used, but at least four hash-length items need to\nbe communicated in each signing round, whereas PLS\nrequires only three, a 25% saving in communication costs.\n\n\nCommunication is important for the IoT world, where\ntypically the radio duty cycle of a thing is limited to a\nfraction of 1%. Also we argue that the computational cost\nof a hardware-accelerated symmetric encryption (which\ntends to be AES) is several times cheaper than that of\na hash for at least some popular IoT platforms, see the\nsection “Related work” for a specific example and possible\nreasons.\nAnother question is whether the public availability\nof H(M1) gives the attacker an alternative method for\nobtaining N1 at BIN=1: by brute-forcing the encryption\nkey. The answer is that this would require the plaintext\n(the cipher text is publicly available as S1 at this point),\nand to obtain the plaintext the attacker needs to know\n_H(N2) = P2, which is not revealed at BIN=1. To obtain_\n_H(N2) from the public value of L1, the attacker requires_\nthe value of N1 which is what the attacker is trying to fit.\nAt the end of interval 2 the receiver will have received,\nand collected from peers all alternative versions of, L2 and\n_S2 and is now prepared for the next, third interval, etc._\nThe protocol is run periodically as long as the transmitter\nstays in commission. Since any secrets have a short lifetime 2τ after which they are published rather than merely\nnot used, there is no accumulation of confidential material at the transmitter site; consequently the transmitter\nhas no security motivated expiration time.\n\n**System architecture**\nThe PLS protocol described above is well suited to serve\nas a basis for a blockchain system. The key property that\nmakes it so suitable is provable, time-referenced forward\n_chaining, the fact that L-messages establish a cryptograph-_\nically protected, unsplittable temporal chain of {Mi}, with\neach Mi being defined by its hash H(Mi). With the head of\nthe chain independently authenticated for all actors before\nthe protocol launch, the set of link/proof pairs and the\nhardness of the hash pre-image problem guarantee that\nthe chain can be validated in isolation by any observer who\n_is present and able to receive messages at broadcast times._\nNo additional source of trust is required to validate the\nchain although a trusted third party may well be useful as\ndefence against a DoS attack, bearing in mind that light\ntouch security would be sufficient for that purpose.\nWe propose an architecture of an IoT system with\nblockchain services, see Fig. 1.\n\n**Sequencer At the core of it is placed a physically**\nsecure, firmware based, multi-radio connected blockchain\nSequencer. The job of the Sequencer is, as the name suggests, building the sequence of chained blocks. The blocks\nthemselves are prepared for the Sequencer by the Fog\nServer (see below). The Fog Server communicates with\nthe Sequencer in one way (transmits), over a separate\nauthenticated radio channel (Bluetooth, for example), the\n\n\n-----\n\n**Fig. 1 Architecture of a PLS-blockchain system**\n\n\ndetails of which are private to them. What differentiates\nthe Sequencer from the Fog Server is the fact that the\nSequencer is not connected to any general purpose networks. Its security role is to run an active PLS protocol on\nschedule and to keep each nonce Nk confidential until the\nnext broadcast period. The Sequencer’s embodiment as a\nseparate air-gapped unit with enhanced physical security\nserves only one purpose: prevention of a blockchain split,\nwhich is effective if and only if the confidentiality of Nk\nover a short period of time can be assured. Split avoidance\nis important for defeating DoS attacks, but the validity of\nthe blocks is not in jeopardy due to the second aspect of\nthe Sequencer, namely the fact that it performs a radio\nbroadcast using a wall-clock synchronised, long-distance\nsignal. As a result, if the defences fail and an attacker compromises the Sequencer, the only non-DoS way to profit\nfrom this is to have different broadcasts directed to different groups of things. In this case the inconsistencies will\neventually be detected by radio monitoring, but it could be\ntoo late, especially if things perform critical work, hence\nthe importance of air-gapping and physical protection for\nthe Sequencer.\n\n**CAS. For the purposes of sequencing and signing we are**\ngoing to use the PLS protocol (and a similar SLVP protocol\nfor things, see the section “Posting on the blockchain”), but\nwhat is going to be sequenced and signed are in fact the\nhashes of the actual blocks and blockchain users’ content\nmessages of variable size. To store these items and retrieve\nthem at a user’s request, we place a Content-Addressable\n**Storage (CAS) unit on a local TCP/IP subnet and organ-**\nise a private channel (behind the organisation’s firewall)\nbetween the unit and the Fog Server. CAS operates in\n\n\nWORM mode: a file is stored once under the hash of its\ncontent as its file name, which means that it cannot be\nchanged without the file name changing, that it is easy\nto verify that it hasn’t been changed and that no trust\nbetween a user and the CAS is required. Due to the irreversible nature of hashing, it is prohibitively expensive to\nstore a different content under the same hash, so all the\ncontent recipient needs to do to ascertain the integrity of\nthe content is check that its hash is correct.\nEvery time a new block is formed to go on the block\nchain, the block content is stored in CAS and the hash of\nit is used in the protocol. All things have access to CAS via\nthe Server or its Proxy and can retrieve files whenever the\ncontent hash is known.\nNote that the hash functions used by CAS and our\nprotocols do not have to be the same. By extending the\nencryption in protocol S-messages to accommodate a\nlonger plaintext (which would require expanding H(Nk+1)\nby replication and concatenation), one could use longer\nhashes for storage than those used in the protocol. It may\nbe profitable to do so given that the second pre-image\nresistance of a hash in the PLS protocol need only to withstand an attack over 2τ, while the resistance of the hash\nfunction used to compute CAS file names is potentially\nrequired to be much greater (years if not decades before\nbroken). This is not an argument for a longer hash for\nCAS (a standard SHA-256 is quite adequate even against\nquantum attacks) but perhaps a shorter hash for PLS (128\nbits) could be sufficient. This can be achieved easily, by\njust taking the first 128 bits from the 256 bit output of\nthe SHA-256 hash and ignoring the rest. The shortening\nis possible due to the fact that security properties of a\ncorrectly designed hash function are assured on a per-bit\n\n\n-----\n\nbasis for the output value; so a subset of the output bits\nhas the same properties as the full set except the collision\nfrequency[2] (vulnerability to a pre-image attack).\n\n**Fog Server. An IoT system’s security cannot be com-**\npletely decentralised since, ultimately, the network of\n_things is owned by an organisation, and since that organ-_\nisation should have authority to add and remove things,\nconfigure them, assign jobs to them, etc. at any time and\nwithout waiting for a slow validation process characteristic of most blockchains. This does not necessarily mean\nthat everything should be centralised; in fact very little\ncentralisation is required: mainly the issues of reconfiguration (adding/removing) and conflict resolution (proof\nservice) and possibly running smart contracts on the\nblockchain, which we will not discuss here.\nThe presence of the infrastructure above the IoT can\nbe captured by introducing into our architecture an actor\nthat we call the Fog Server (FS), which is a local datacenter that has direct connection to things via a suitable radio\nnetwork (e.g. LoRa), as well as a sufficient compute power,\nstorage and connection to Cloud. The blockchain users\nshould trust the FS for:\n\n - enrolment of new things to the block chain, and their\nremoval. At the point of enrolment, the Cloud\nsupplies to the FS confidential identity material for\nthe new thing which allows the FS to establish initial\ncredentials of the thing on the blockchain. It also\nshares with the thing a symmetric key. Also note that\na modern IoT device is equipped with hardware\nencryption facilities and can hold the key in\nsoftware-unreadable, immutable persistent memory\ninside its Hardware Security Module (HSM).\n\n - as part of the previous, supplying to a thing that joins\nthe blockchain late the authenticated hash of the\nrelevant blockchain history.\n\n - withdrawal of a thing from the blockchain.\n\nThe FS is the actor that forms blocks for the blockchain\nby gathering and validating SLVP protocol messages (see\nthe section “Posting on the blockchain”) from things and\noutside agents (via Cloud) and collecting them into blocks.\nNo trust is required for this, since the FS processes messages received from things and agents on the basis of the\nblockchain content, which is public and available to all\nusers. Any violation of the validation rules will be noticed\nby the parties affected and any well equipped witness, i.e.\na (possibly non-enrolled) radio listening post.\n\n2Whether the increased vulnerability to a pre-image attack is significant\ndepends on the application’s specific threat model. In most cases it is not,\nsince the attacker would only have a few minutes between the consecutive\nblocks for the attack. For the avoidance of doubt, we do not demand the hash\nsize reduction, only suggesting it for resource optimisation. As a matter of\nfact, our solution would accommodate a double-length hash, 512 bits, just as\neasily should it become necessary.\n\n\nWhat cannot be protected 100% reliably by our methods is progress. It is possible in principle for the FS to deny\nservice by refusing to react to valid, legitimate messages\nsent by things, or by maliciously modifying the content of\nthose messages, if the FS is compromised. However, server\nprotection is not an issue of IoT security, but a general\ncybersecurity concern, which is beyond the scope of this\npaper.\n\n**Proxies This is yet another kind of zero-trust agent. It has**\ntwo functions:\n\n1. Amplify the Sequencer’s radio broadcasts (by\nconstructive interference, if LoRa communications\nare used, or by re-broadcasting broadcast messages).\nAll Proxies are sent PLS messages by the Sequencer\nvia the FS just before they are due to be broadcast.\n2. Pick up messages from things directed to the\nblockchain. This is useful because direct connection\nto the FS over the air may not be possible given a low\npower budget and compromise antennae that things\nhave to work with. All Proxies are connected to the\nintranet and can forward messages to the server and\nCAS. The blockchain posting protocol (SLVP), which\nwe will introduce in the next section, is robust:\nProxies can forward messages of unclear origin and\nauthority; at best they will be filtered out by the FS, at\nworst they will find their way to the blockchain but\nwill not be properly signed by a legitimate actor and\nhence will have no effect other than waste of\nresources.\n\nSince all Proxies are connected to the intranet, they can\nperform function 1 above.\n\n**Communications The reader will have noticed that the**\nproposed architecture uses the combination of a dedicated radio channel for security-related data and a\ngeneral-purpose communication infrastructure, whether\nwired or wireless, on which things may establish an\n_auxiliary channel. This is a deliberate choice for the fol-_\nlowing reasons.\nThe purpose of the dedicated radio channel is to ensure\nby physical means the reliability of broadcast. The only\ncredible threat to the PLS protocol is a DoS attack whereby\none or more users of the network are prevented from getting uncorrupted P, L or S messages sent by the Sequencer.\nThe architecture enables a volume transmission of these\nmessages, with the Sequencer radiating them at maximum legal power and the Proxies joining in by concurrent\ntransmission. We propose that the security radio channel is implemented via LoRa (Sentech Corporation 2019),\na Long Range spread-spectrum technology with many\nremarkable properties.\n\n\n-----\n\nWe exploit the fact that as a spread spectrum format,\nLoRa benefits from concurrent transmissions, where two\nor more signals carrying the same content under the same\nmodulation regime can be broadcast simultaneously from\ndifferent locations. Any receiver at whose location the signal of one transmitter exceeds the rest only by a factor of 2\n(3dB) will not sense the others ((Liao et al. 2017) p.21436,\nunder “Capture Effect”, (Zhu et al. 2018)). This is due to\nthe Frequency Modulated (chirp) nature of LoRa, and is\nwell known as the capture effect in both communication\nand FM broadcast industry (Leentvaar and Flint 1976).\nThe idea is to endow the Proxies with maximum legal\npower transmission capabilities and to place them in such\na way that the communication range of each Proxy captures a certain structural unit on the premises, e.g. a floor\n(or a building if it is small enough), to form a communication locus. Different loci are separated by distances\n(as in the case of loci as individual buildings) or obstacles (e.g. construction elements supporting a floor if the\nloci belong to different floors). This was studied at length\nin (Liao et al. 2017) (see p.21443 under “The Robustness\nof CT-LoRa”). The distance is by itself quite an effective\ndampener, as radio signals fade by 6db when the distance between transmitter and receiver is doubled, which\nwould ensure a more than sufficient power contrast for\nthe capture effect.\nIf properly deployed, the proposed architecture ensures\nthat an attacker can only deny LoRa communications to an\nIoT node by radiating power far exceeding legal limits, or\nby installing additional equipment on the premises in violation of physical security. In both cases a large proportion\nof the IoT devices will remain unaffected by the attack.\nThey will be able to collect all versions of the security messages (including genuine ones and those coming from the\nattacker) and exchange them between each other using an\nunprotected network. Security protocols will then quickly\nestablish which versions are genuine.\n\n**Posting on the blockchain**\nIt is tempting to use a protocol similar to what we\ndescribed in the section “Basic protocols” in order for an\nIoT device to post transactions directly on the blockchain.\nThe advantage of a Guy Fawkes type protocol (which we\nwill call a GF protocol for short) is that any secrets are\nshort-lived, the calculations basic and post-quantum, and\ncommunications modest. However the principle vulnerability of such a protocol is its inherent reliance on the\nprecedence of events. With PLS we used wall-clock time\nto separate intervals; wall-clock time with a reasonable\naccuracy (less than tens of seconds drift over a year) is\ncheap and available to even the tiniest of IoT platforms.\nHowever, a thing does not generally need to post a transaction on each block of the blockchain, and it is often\nquite expensive for it to do so from the energy perspec\n\ntive. Precedence can easily be established if the protocol\npublishes messages on the ledger using an ideal communication environment, where all messages reach their\ndestination and all can be published in the interval that\nthey were emitted. Which is far from reality in the IoT\nworld.\n\n**Jam-spoof attack. With Sequencer broadcasting mes-**\nsages for all users without exception, having maximum\nlegal transmission power and being further supported by\nProxies, one can guarantee message delivery (possibly several unauthenticated versions, but that is no problem)\neither directly, or via subsequent exchanges between users\nbefore the interval is over. When a thing sends its content\nto the FS, the content is of no interest to the peers and\nthe resources available for delivering it are quite limited. If\nthe thing posts on the chain infrequently, at unpredictable\ntimes, any GF protocol based on imperfect communication is potentially vulnerable to the jam-spoof attack (see\nthe section “Original GFP”) as follows:\n(below T is a thing and M is an attacker)\n\n1. T runs the protocol to the point where it is about to\nreveal to a verifier the so far confidential pre-image to\nprove a signature (a key feature of any GF protocol)\n2. M suppresses the verifier’s receiver by jamming the\nbroadcast channel. At the same time M uses a highgain directional antenna and sophisticated signal\nreconstruction techniques beyond the capabilities of\nthe verifier to reliably receive the message from T. As\na result M learns the secret, but the verifier is left\nbelieving that the secret has not been revealed yet.\n3. M masquerading as T proceeds to publish its linkand proof-records based on the knowledge of the\nsecret pre-image. The link-message will use a\ndifferent next nonce than the genuine link-message\nfrom T published on the blockchain earlier, but the\nsame current nonce, thus forking T’s GF sequence.\nAlso the knowledge of both nonces enables M to\npost its own signature message on the blockchain to\nsign an arbitrary hash on behalf of T and to continue\nto do so indefinitely.\n4. T sees the split of its sequence on the blockchain and\nalerts the FS, but now the FS (or any other arbitrator)\nis unable, based solely on the blockchain content, to\ndetermine whether it is T or M that is the genuine\noriginator of the latest messages.\n\nTwo remedies are available. One is to require authentication of all messages from a thing to the FS. This immediately destroys the zero trust environment we have built,\nwhere the only aspect that all users, including the FS itself,\nhave to trust is the correct operation of the Sequencer. We\nwish to implement a GF protocol that does not require\nadditional trust and which survives a jam-spoof attack. It\n\n\n-----\n\n**Table 2 SLVP Protocol**\n\n**Block** **Transmit** **Verification** **BC Action**\n\n_i0_ _P1 = H(N1)_ Out of Band (Enrolment)\n\n_i1_ _S1 = EN1_ _(H(M1) ⊕_ _H(N2))_\n\n_i2_ _L1||V1 = H(N2) ⊕_ _N1 || H(H(N2)||N1)_\n\n... ... ... ...\n\n_in_ _Pk = H(Nk)_ ... post Pk\n\n_in+1_ _Sk = ENk_ _(H(Mk) ⊕_ _H(Nk+1))_\n\n_in+2_ _Lk||Vk = H(Nk+1) ⊕_ _Nk || H(H(Nk+1)||Nk)_\n\n_in+3_ _Pk+1 = H(Nk+1)_ Fetch latest P = Pk from block B = in\n\nSet failed=true\n\nfor LV: B < #(LV) < in+3\n\nif H(L ⊕ _Pk+1) ̸= P, continue_\n\nset N = L ⊕ _Pk+1_\n\nif H(Pk+1||N) = V\n\n**failed=false; break**\n\nif failed, exit ignore Pk+1\n\nfor L[′]V [′]: B < #(L[′]V [′]) < #(LV)\n\nif V [′] = H(L[′] ⊕ _N||N), exit_ ignore Pk+1\n\nfor all S: B < #(S) < #(LV)\n\ndetermine HM = DN S ⊕ _Pk+1_\n\nsend to CAS: (k, UID, HM, α(LV), α(S)) post Pk+1\n\nexit\n\nAssume bitwise exclusive-or ⊕ to have a higher priority than concatenation ||. #(x) is the number of the block in which record x is located, and α(x) is the record’s unique ID\n(address). The sequence of block numbers is in strictly increasing order: i0 < i1 < i2 < ...\n\n\nturns out that a small modification is sufficient to solve the\nproblem, which brings us to the following SLVP protocol,\nsee Table 2.\n\n**SLVP protocol**\nA new blockchain user is enrolled by the Fog Server[3]\n\nby authenticating the user’s first P-record out of band.\nAssume that the user sends that record to the blockchain\n(i.e. to the FS in the first place) and it appears in some\nblock i0.\nAll records on the blockchain carry a UID, i.e. the User\nIDentification, stated by the message originator. We propose that the user is identified by the first two bytes of its\n_P1 record. The server will not accept the P1 message if it_\ndetermines that the first two bytes of the hash clash with\nan already established user. On the other hand, 2 bytes are\nsufficient for an excess of 64 thousand users, while a typical IoT swarm does not exceed[4] 1000. In the sequel we\nwill not distinguish between a UID and the user with that\nUID, if the context is clear enough to see which of the two\nwe mean.\n\n3this makes it a permissioned blockchain\n4There is some evidence that 1000 nodes could saturate the IoT long-range\njoint channel capacity (Haxhibeqiri et al. 2017)\n\n\nThe table in Table 2 presents the protocol from the\npoint of view of a single UID and the FS. Other UIDs\nwill conduct themselves in the same way. Like the PLS\nprotocol introduced earlier, the SLVP protocol is invoked\nperiodically, but now at some arbitrary times, which in\nterms of the blockchain schedule correspond to block\nintervals. The transmitting thing does not need a specific\nconfirmation that the message has been received; it simply\nchecks newly formed blocks to find the record that it has\ntransmitted. When this happens, the thing sends the next\nmessage according to the protocol. If a legitimate transmitter’s correct message is not posted on the blockchain\nafter it has been transmitted, it either wasn’t received at\nall, or it was received distorted. Either way the thing will\nre-send the message until it is posted on the blockchain\nintact.\nThe protocol proceeds in rounds, each consisting in\nthree steps:\n\n_S →_ _LV →_ _P_\n\nwhere S is, as before, the signature record, LV is an\nextended link-verify record, and P is the proof record as\nbefore. The LV record consists of the L-record, similar to\nthat of the PLS protocol, and a V -record for thwarting\n\n\n-----\n\nthe jam-spoof attack. Since the UID is not authenticated\nand the channel generally lacks integrity, any messages\nreceived by the blockchain can be arbitrarily distorted.\nThe SLVP protocol depends on things’ ability to receive\nblocks on the blockchain successfully via the PLS protocol.\nUnlike the Sequencer engaging in timed broadcasts, an\nSLVP user can be quiescent for many block periods. The\nnew round k starts after the user’s Pk message has been\nreceived by the FS, validated and posted on the blockchain\nin some block in. Which block this is going to be depends\non the timing of the Pk message. The user, having satisfied\nitself that its message Pk was received and posted under its\nUID, sends the S-message Sk and waits for it to get posted,\ntoo, say in block in+1 > in.\nHaving satisfied itself that the Sk has been posted on the\nblockchain, the user transmits its LV -message.\nThe message Pk+1 is sent in some later block period in+3\nto finish the current round. The FS verifies the message\nusing the verification procedure in the middle column of\nthe table in Table 2. If verification succeeds, the new Precord is posted.\nNext the FS will decipher all S-records posted after\nblock in under the user’s UID by computing\n\n_HM = Pk+1 ⊕_ **DPk+1⊕LS .**\n\nEach record r on the blockchain has its address α(r) =\n_(ir, lr), where ir is the block number in which r is located_\nand lr is the sequential number of r among the records of\nthe same type and under the same UID in block ir.\nFor each record S in the current round k, the FS collects\nproof data in the following form:\n\n_WS = (k, UID, HM, α(LV_ _), α(S))_\n\nand instructs the CAS unit to store WS under H(WS) as\nusual. The CAS unit will use HM as a trigger. When/if the\nuser UID stores content c in CAS, such that H(c) = HM,\nthe CAS manager will post a special C-record, on the\nblockchain on behalf of UID as follows:\n\n_C = UID : (HM, H(WS))_\n\nwhich serves as blockchain confirmation that CAS has\ntaken charge of the content file as well as the proof data\nfor it for any witness to verify. Triggers that are not triggered by the user over a certain number of blocks (large\nenough to conclude that the original S-message was counterfeit/distorted) are removed from CAS and entered into\nthe FS security log. A C-message will be ignored (and the\ncorresponding C-record not posted) if a trigger for it has\nnot been provided by the FS at the time of submission.\n\n\n**Security analysis of SLVP**\nAn S-message does not expose a single bit of the nonces\n_Nk and Nk+1 since the value Sk depends on yet undis-_\nclosed H(Mk) and since there does not exist an attack on\nthe cipher E where neither the key nor the plaintext is\nknown. Sometimes it is convenient for the user to post\nmore than one S-record, for example when several documents are to be signed by the same user but they are not\notherwise related. The user is allowed to send as many\ndifferent S-messages as necessary.\nFor an LV -message, the link part, L is the same as that\nin PLS, and it serves the same purpose: its value links the\ncurrent nonce Nk with the new one, Nk+1. The verify part\n_V is there to make sure that an attacker who learns Nk_\nlater cannot combine it with its own _N[ˆ]_ _k+1 and post_\n\n_Lˆ_ _k = H( ˆNk+1) ⊕_ _Nk_\n\non the blockchain. In such a case the FS would be unable\nto decide between Lk and _L[ˆ]_ _k due to the fact that Lk can be_\na distorted version of _L[ˆ]_ _k, and the message_ _L[ˆ]_ _k an attempt_\nto correct the distortion. With the V message in place, for\nany pair of LV -records:\n\n_Lk||Vk = H(Nk+1) ⊕_ _Nk || H(H(Nk+1)||Nk)_\n\nand\n\n_Lˆ_ _k|| ˆVk = H( ˆNk+1) ⊕_ _Nk || H(H( ˆNk+1)||Nk),_\n\nwhere\n\n_H(Lk ⊕_ _Pk+1) = H(L[ˆ]_ _k ⊕_ _Pk+1) = Pk,_\n\nthe one posted in an earlier block wins: the protocolcompliant user does not disclose the genuine Pk+1 =\n_H(Nk+1) in the same block as Lk||Vk and so the fact that_\n\n_Vk = H(Pk+1||Lk ⊕_ _Pk+1)_\n\nproves that the originator knew H(Nk+1) before it was\nposted. The only actor that knows H(Nk+1) before it is\nposted is the genuine user.\nNotice that no matter how many counterfeit LV messages have been posted by attackers since the last verified\n_P and no matter how many counterfeit P-records are sent_\nafter them, only one P record will be accepted and posted\nby the FS in any round of the protocol on behalf of any\ngiven UID. Also the FS will find only one LV -record to\nbe valid, which is the earliest LV -record compatible with\nboth the previous and the newly validated P-record.\nCounterfeit S-messages pose no threat. They will be\ndeciphered to an unpredictable HM, and an attacker would\nnot be able to provide content that matches a given hash\nvalue anymore than it is able to find a nonce N given\n_H(N), the latter being the main security assumption for_\nany GF protocol.\n\n\n-----\n\n**Enrolment and optimisations**\nIt has been mentioned earlier that the very first hash P1 of\nthe protocols is validated out of band. For a user to be able\nto start SLVP there are two requirements:\n\n - access to the blockchain which includes out of band\nvalidation of the latest Pk and all previous blocks\nfrom 1 to k − 1, inclusively\n\n - registration of the user’s P1 for out of band validation.\n\nEnrolment of new equipment normally requires a\nhuman administrator as it involves physical placement,\nconfiguration and initialisation of the item according to\nthe business objectives. We propose the following enrolment protocol:\n\n1. The administrator’s workstation establishes secure\nconfidential communication with the FS using Cloud\nand state of the art security. The FS shares a fresh key\nK with the administrator.\n2. The new thing that the administrator has ascertained\nto be genuine\n\n(a) receives K and the Sequencer’s latest Pk using\nnear-field communications (NFC) or similar,\n(b) generates N1 and another random nonce N [∗]\n\n(c) computes P1 = H(N1)\n(d) sends Q = P1||EK _(P1 ⊕_ _N_ [∗]) back to the server\nvia the administrator’s NFC port acting as a\nrelay.\n\n3. The FS examines a short prefix of P1, π(P1), e.g. 2\nbytes, and checks that no UID with this value has\nbeen enrolled. If that is the case, the server computes\n_N_ [∗] from Q and responds with ACK = H(N [∗])\notherwise the response is FAIL.\n4. If the response is FAIL, the thing generates a new\npair N1 and N [∗] and repeats steps 2 and 3. Otherwise\n\n(a) the thing verifies that ACK = H(N [∗]) and\nnotes its new UID, i.e. π(P1)\n(b) confirms completion to the administrator.\n\nIf ACK ̸= H(N [∗]) the protocol fails; a notification to\nthis effect quoting P1, N [∗] and ACK is sent to the\nadministrator for subsequent analysis.\n\nNow the new IoT device is ready to receive the minimum data necessary to access the blockchain. The amount\nof trust required for it is exactly the same as it is for\nany other user: it needs to authenticate the latest Pk, the\nonly difference being that for the devices that have been\npresent from the start the index k = 1. But how is it going\nto authenticate the blocks that were formed before block\n_k?_\n\n**Merkle tree. To describe our proposed solution we need**\nto remind the reader the idea behind the Merkle tree first.\n\n\nThe Merkle tree is a tree in which every leaf node is\nlabelled with the cryptographic hash of a data block, and\nevery non-leaf node is labelled with the cryptographic\nhash of the labels of its child nodes. See Fig. 2 for an\nexample.\nBased on the same fundamental hardness of finding a\nsecond pre-image, it can be concluded that if the top-level\nlabel is authenticated reliably, then no further node of the\ntree needs to be. Indeed, it is cryptographically hard to\nchange, for example, the data block a without changing\nthe label A, and if A changes, the value of X and consequently T will become invalid. So one hash value is\nsufficient to authenticate the whole data structure. At first\nglance, the tree seems to be an unnecessary complication,\nsince the hash of a, b, and c alone, h(a, b, c) would be\nsufficient to assure the integrity of all three items. However, the advantage of the Merkle tree is that it allows\none to access just the items that one wants. For instance,\nto access a, one does not have to read b and c; all that’s\nrequired is to read nodes T, X and A. The amount of data\nretrieved will be |a| + 5|h|, where the vertical bars denote\nthe item size[5]. Without the Merkle tree, in the case of\na single hash controlling a, b and c, the retriever would\nhave to read b and c and compute the hash to assure the\nintegrity of a, even when the contents of data blocks b and\n_c are of no interest. Since items can be of an unlimited_\nsize and the hash size is limited and small, the Merkle tree\noffers a clear advantage in both communication and hash\ncalculation costs. Now let us proceed to our proposed data\nstructure.\n\n**Merkle forest. We propose to communicate the Merkle**\nminimal forest roots of the current state in each\nblockchain block. The roots are a compact collection of\nroot hashes that can be followed on CAS to securely access\nany block from B1 to current stored in CAS. To illustrate\nthe concept, let us imagine a block sequence from block 1\n(initial) to block 7 (current), see Fig. 3. For simplicity we\nuse a binary Merkle tree, in which every non-leaf node\n\n5the label of a node is not stored at the node (except for the top node). The\nretriever will read T, X, C (and verify that T = h(X, C)) from the top node, A\nand B from node X and finally a from node A, which works out as 5 hash\nlengths plus the length of a.\n\n\n-----\n\nis composed of the hashes of the two child nodes’ contents. Some of the nodes are already formed and will never\nchange (shaded in the figure), and some are still being\nformed pending the future blocks. It is easy to see that the\nMerkle proof of any block up to and including 7 requires\nonly hashes of node 4321, 65, and 7 as roots, the paths of\nevery leaf from 1 to 7 is rooted at one of them. Notice that\nthe binary representation of 7 is 111 which corresponds\nto one node each at levels 0,1 and 2. For block 5=1012\nwe would have a level-2 block and a level-0 block, which\nagrees with the diagram.\nA binary Merkle tree is unjustifiably deep. Focusing on\nthe world of the IoT, we recognise that communications\nare typically limited to messages no longer than 200–250\nbytes, so given a typical hash size of 32 bytes, it is convenient to use a quad-tree, which will be much less deep.\nFor a quad tree in a blockchain of say, 1 mln blocks (at one\nblock per 15 min, this gives us more than 10 years’ running), we get log4 10[6] ≈ 10, which means that the server\nonly needs to authenticate at most 10 hashes to give a new\nIoT thing a secure start. For every k the record consisting\nof the minimal root set\n\n_�(k) = (r1, ..., rp)_\n\nis computed and stored by the FS under γk = H(�(k)) in\nCAS, where k is the current block number, and r1, . . ., rp\nare the Merkle tree hashes that correspond to the nonzero\ndigits in the base-4 representation of the number k.\nFor each k, the FS will put γk at the beginning of block\n_k + 1. Any user that wishes to trade storage for CAS_\ncommunication, or a new user who has missed an initial\nsegment of the block chain, but who trusts the latest Pk\ncan use the γ -record on the block to securely retrieve any\npreceding block(s) via CAS, if they choose to trust the FS.\nWe would like to remark that the FS in this particular case\n\n\n**is trustworthy, since the γ record can be computed by any**\nfull witness of the blockchain (i.e. any user that has been\npresent since block 1) and if the FS is compromised, the\nproof of that will be constructed immediately.\nRegarding the storage requirement in CAS, they are\nminimal. Summing up the geometric series for the radix4 Merkle tree with depth 10, we get circa 350,000 hashes\nto store for 10[6] blocks, about 10Mb, a trivial amount of\nstorage.\n\n**Countermeasures against DoS**\nThe acquisition of a shared secret between the thing and\nthe FS at the point of enrolment does not make the\nblockchain any less useful. Indeed, in our threat model\nthe FS is not trusted by any enrolled user any more than\nany other user of the blockchain, so the shared secret\ncannot be used to replace the security protocols that\nmake the blocks of the blockchain an immutable, ordered,\nauthenticated sequence of records. Nor is it any good\nfor non-repudiation. However, just as the Sequencer is\ntrusted to keep its secret for the avoidance of blockchain\nsplit so is the FS trusted to be interested in reducing the\namount of noise on the blockchain, i.e. records sent in by\nan attacker on behalf of a genuine UID, which will eventually be caught out and eliminated by the SLVP protocol.\nAfter all, as the FS is solely responsible for what does and\nwhat does not get posted, the proposed blockchain concept only works on the assumption that the FS itself is\n_not and can never be behind a DoS attack. The assump-_\ntion that the FS will be a willing party to an additional\nnoise-reduction protocol does not add much to that.\nWith this in mind we propose that each thing uses a\nvery short Message Authentication Code (MAC) based\non symmetric encryption and the shared key K received\nupon enrolment. The MAC need not be longer than\n\n\n**Fig. 3 Merkle forest**\n\n\n-----\n\n2 bytes (possibly even 1 byte) and can be computed\nusing standard techniques by the thing’s hardware security module or crypto accelerator. The MAC is computed\nfor each message of the SLVP protocol sent to the server\nas well as the content messages sent via the FS to CAS.\nDue to the shortness of the MAC, the exposure of the\nshared key is minimal, obviating session keys. If the MAC\ndoes not match, the FS ignores the message. With a 2-byte\nMAC, an attacker would have to send tens of thousands\nof messages to get through to the FS in the first instance;\nsuch a volume on behalf of a single IoT user will surely\nraise the alarm, resulting in the intruder’s triangulation\nand suppression.\nRecall that the Sequencer’s messages may arrive distorted or not arrive at all, and the users, especially things,\nmust talk to each other to collect a set of versions for\neach PLS message to ensure that the set contains the\noriginal. To facilitate this, a short authenticator can be\nsent by a Proxy on an auxiliary channel to each thing by\ntransmitting\n\n_u = UID∥cat∥π(H(M))_\n\nwhere UID is its User ID, π(H(M)) is a short hash of message M from the category cat (one of P, L, or S). Message\n_u is extended with MACK_ _(u), where K is the key agreed_\nwith UID at enrolment. The message u is prepared by the\nserver and is forwarded by one or more of the Proxies on\nthe auxiliary channel. User UID, having received u and\nchecked the MAC, recalculates π(H(M)) based on the latest message in category cat received (if it did at all) and\ncompares it with the value contained in u. If they match,\nthe device joins a re-broadcast concurrent-transmission\ngroup on a pre-arranged channel (frequency and time relative to the start of the Sequencer broadcast interval) to\nhelp nearby nodes with PLS reception. Given that PLS\nmessages are short (not much longer than 32 bytes if SHA256 is used for H(·)) and infrequent (3 messages typically\n2–5 times per hour, 0.5 KB/hour), a blockchain supported\n_thing can afford to transmit as much to help other things_\n(which in turn will help it) to survive a DoS attack.\nWhen it comes to the SLVP protocol, the user is the\nactive transmitter, and the roles are reversed. Now as a\nDoS resilience measure, the user UID adds a MACK _(x) to_\nevery message x that it sends to the server (possibly via a\nProxy) for posting on the blockchain. The FS checks the\nMAC based on the received UID and the shared key K and\nif the MAC does not match, it ignores x. Again, we must\nstress that if the MAC does match, this means nothing in\nterms of the SLVP protocol, since the FS does not trust the\n_thing any more than the thing trusts the FS. Reduction of_\nnoise is their common concern: the FS acts on behalf of\nthe owner of the IoT network and is interested in suppression of a DoS attacker, and the thing will keep its shared K\nsecret to avoid an attacker’s spoofing it and preventing its\n\n\nlegitimate messages from reaching the blockchain. Commonality of concern is the only reason why the additional\nauthentication will be effective.\n\n**Emergency mode**\nA distinguishing feature of IoT is its multiplicity of\ntime scales. Most things require only infrequent interaction with the outside world, reporting sensor readings,\nreceiving parameter updates and possibly code upgrades.\nAll these activities are easily accommodated by the\nblockchain mechanism and are protected by its inherent\nsecurity properties. A major downside of a blockchain\nis its latency. No matter how frequently new blocks are\nadded to the chain (and in our case they are not even\nmined), a thing may find itself in a situation when it must\nraise the alarm with its master sooner than a new block\ncan be published, especially since in the case of the PLS\nblockchain, blocks are published on a fixed wall-clock\nschedule. Even if a new block is to emerge soon, there is\nno guarantee that any given thing will be able to post its\nmessage in it rather than in a later block.\nThis problem is quite practical: a hospital monitor\ndetecting a catastrophic change in a patient’s condition\nand a nuclear plant’s sensor detecting a reactor malfunctioning are cases in point to name but two. We emphasise\nthat emergency messages are not an alternative of posting records on the blockchain. The latter is more powerful\nin that things are able to securely interact with each other\ndirectly via their signed blockchain messages, whereas\nemergency communications are processed solely at the FS\nfor off-chain delivery outside the IoT network.\nOne might think that emergency communications can\nbe supported by the shared key K that the originating\n_thing agreed at enrolment. Indeed the FS can request a_\nfull MAC and satisfy itself that the message is authentic.\nHowever, this is not enough. Emergency communications\ninvolve rapid response and that can only be provided\nif an independent arbitrator can establish that the message was sent by no-one but the claimed originator. In\nother words, a signature rather than mere authentication is required. In the absence of signature, the response\nagent would be running the risk of the originator repudiating the message: after all, the symmetric key K is\nshared with the FS, and so either the FS or an agent to\nwhich the FS has leaked the key (willingly or not) might\nhave sent the emergency message instead of the legitimate\nuser.\nNon-repudiation is not a concern with blockchain communications, they cannot be repudiated thanks to the\nproperties of the SLVP protocol. However, post hoc validation by blockchain is only useful for confirmation\nof valid messages, rather than proving a message to be\ninvalid, since the rapid response must come into effect\n_before blockchain validation may take place. The other_\n\n\n-----\n\nway[6] of ensuring non-repudiation is by One–Time Signature (OTS), which we will consider next.\n\n**OTS**\nOTSs are known to have a very large “public key”, i.e.\nauthenticated public data used for validation of a signature. In the original OTS proposed by Lamport (Lamport\n1979), the originator shares with the verifier k pairs\n\n_(H(n1), H(N1)), (H(n2), H(N2)), . . ., (H(nk), H(Nk)),_\n\nwhere all ni, Ni are random nonces. To sign a k-bit message {xi}, the originator additionally supplies k values\n{si}:\n\n�\n_ni if xi = 0_\n_si =_ _Ni otherwise_\n\nOTSs solve the problem of emergency non-repudiation\nif the public key is signed and posted on the blockchain in\nadvance (using SLVP), but the price for an IoT device using\nit is prohibitive. A straightforward application of OTS to\nsigning a full hash of an emergency message would require\n256 × 256 × 2 = 128K bits of public key, or 16KB. It is\neasy to see that the public key can only be used once if\nwe want the security of hash pre-image to work for every\nbit of a signed message. Even if using the large key once\nwere acceptable (think of catastrophic circumstances that\ndo not present themselves often), the signature size would\nalso be prohibitive: half the key size, or 8KB in this case.\nThis would take some time to communicate over a low\nbit-rate channel, especially in the presence of transmission\nerrors necessitating a re-transmission.\n\n**Public key**\nLet us start with the public key problem. We propose to\nbring the SLVP protocol messages to bear on the emergency mode to eliminate transmission and authentication\nof the public key. Recall that a thing running the protocol\nsends messages that depend on nonces Nk which are chosen by it at random, see Table 2. At the validation step, the\nFS computes _N[ˆ]_ _k = Lk ⊕_ _Pk+1 and verifies that_ _N[ˆ]_ _k = Nk_\nby applying H(·) to both sides and checking the equality.\nWe propose that every thing engaging in SLVP must\ncompute random nonces Nk by building a hash chain:\n\n_Nk = Nk[[][α][]]_\n_Nk[[][i][]]_ [=][ H][(][N]k[[][i][−][1]]) where i = 1, . . ., α,\n\nand where Nk[[0]] is completely random and is kept secret by\nthe thing for at least α rounds of the protocol.\nIn other words, every nonce is an image of a random\nnumber under α applications of H(·), which is known as\nthe Winternitz chain. When the server has posted the\n\n6Our design constraint 1, Post Quantum, prevents standard public-key\ncryptography, which would provide an effective signature if the originator’s\npublic key is validated in advance on the blockchain\n\n\nvalue Pk it received from the thing, it has access to, and has\nverified,\n\n_Ni[[][α][]]_ for all i = 0, . . ., k − 1\n\nFor a given UID, the private key for the period between\nthe postings of Pk and Pk+1 (i.e. when nonces up to and\nincluding Nk−1 have been revealed) is as follows:\n\n{Nk[[][α]−[−]i _[i][−][1]]} for i = 1, . . ., α −_ 1 .\n\nIf Lamport’s OTS is used, the thing sends to the FS a\nselection of the values Nk[[][α]−[−]i _[i][−][1]]. The FS verifies each value_\nby applying H(·) to it i + 1 times and comparing the result\nwith Nk[[][α]−[]]i [=][ N][k][−][i][ that it has received from the same UID]\nby SLVP protocol. Notice that as k advances to k +1 with a\nnew round of SLVP, the same chain Nk−i is used for public\nkey with an earlier pre-image:\n\n_Nk[[][α]−[−]i_ _[i][]]_ → _N([[]k[α]+[−]1[(])[i][+]−[1](i[)]+[]]_ 1)\nuntil the protocol is α rounds ahead of the chain at which\npoint the chain will have been fully used and is no longer\nrequired for OTS purposes. An example of α = 3 is displayed in Fig. 4, showing two consecutive rounds. Notice\nthat the values used in a later round are always lower on\ntheir Winternitz chain than those in earlier rounds, making them secure within the second pre-image hardness\nassumption.\nWe conclude that the “private key” for an SLVP round,\ni.e. the set of potential pre-images to be used for OTS, is\nunknown to the FS in that round and that the FS has access\nto the authenticated public key to verify the signature.\nThis arrangement of chains and pre-images makes it\npossible for a thing running SLVP not to share any public\n_key at all and at the same time be able to sign messages in_\nemergency mode without waiting for blocks to appear on\nthe blockchain. It is quite useful for the IoT world, and the\n\n\n-----\n\nprice that we pay is the need to pre-hash a random string\n_α times at every round of the SLVP protocol rather than_\nuse it directly as a nonce. Taking a popular ESP32 systemon-chip as a specific example we learn from (Espressif\nSystems 2020) that it takes ∼ 1μs at full power to process one AES256 hash block, perhaps 0.1ms for α = 100.\nThe energy spent is a fraction of the (LoRa) communication cost for the same. It is completely justified if the thing\npotentially requires emergency communications in this or\nany of the future α rounds, assuming that it is sufficient to\nsign a certain number of bits L of the emergency message\n(or its digest) to reassure the responder of nonrepudiation.\nFor Lamport’s classical OTS signature, L = α/2. We will\nimprove on this next.\n\n**GF-HORS**\nThe excessive size of OTS signatures have been recognised\nby many authors, and several proposals have been made\nto improve on it. We follow the methodology presented in\n(Reyzin and Reyzin 2002), where an original idea, Hash to\nObtain a Random Subset (HORS), was first put forward.\nAssume that α is a power of 2. Compute a length-L\ndigest of the message to be signed, and partition its binary\nrepresentation into slices log2 α bits long. Interpret these\nslices as unsigned numbers\n\n_L_\n_σj, j = 0, . . .,_\n\nlog2 α [−] [1 .]\n\n\nwhere all σj < α. Now for each j the thing supplies the\n\n[α−σj−1]\nvalue Nk−σj to form a signature. The FS validates the\nsignature by recomputing the digest of the message, then\nrecomputing {σj} from the digest, and then for each j ver\nifying Nk[α−−σjσj−1] by applying H(·) to it σj + 1 times and\nchecking that the result equals the previously obtained\nnonce Nk−σj .\nThe idea to use the digest of the message to be signed\nrather than the actual bits of it by partitioning the string\nwas first proposed in (Reyzin and Reyzin 2002), and the\nsecurity of this method is slightly less than that of the\nsecond pre-image hardness, since here the attacker only\nneeds to find a message whose digest partitioned into suitable chunks gives the same set or even a subset of {σj}\nin any order. However, the authors of (Reyzin and Reyzin\n2002) remark that finding a (useful) message that has the\nsame set or a subset of digest chunks as a given one is still\ncomputationally hard for a good digest.\nWe propose a GF-HORS signature (HORS signature\nwith SLVP-derived public key) based on a keyed MAC with\n_the shared key K as the digest. The MAC protects the mes-_\nsage being signed from an outside forgery, and the HORS\nsignature protects it from an insider job. Let us take a look\nat some example numbers to illustrate the efficiency of the\nscheme.\n\n\nIf we assume α = 64 and use AES-128 for the digest\nMAC (shortening it down to 126), we get up to 21 sigmas. Assuming for estimation purposes that the digest\nis a random bit string, the probability that an attacker’s\ndigest gives a subset of the sigmas, is less than (21/64)[21],\naround 10[−][10], a pretty good result for an IoT device nonrepudiation. The communication cost of the signature is\n256 × 21/8 = 672 bytes, about three messages on LoRa.\nRecall that we require three shorter messages (around 128\nbytes all together) for an SLVP round, which is in the same\norder of magnitude.\nA final remark. When a thing is first enrolled by the\nserver, there is not enough nonces in its history (in fact\nthere aren’t any initially) for the formation of the public key. One remedy could be to produce α nonce chains\nat enrolment and share α chain-ends with the FS at that\npoint. Another solution is to consider the first α SLVP\nrounds of a new thing a probationary period, when it is\nbeing tested and adapted to its environment and when it is\nnot allowed to participate in emergency communications.\n\n**Related work**\nThe advantages of blockchain technology in the case of\nIoT are not clearly articulated in literature. Recent surveys (Kouicem et al. 2018), (Wang et al. 2019) recognise\nblockchain as a disruptive technology for the IoT, and list\nthe benefits in generic terms:\n\n - Decentralisation: Distributed Ledger Technology is\nsupposed to be more robust and secure against a\nsingle point of failure.\n\n - Pseudonymity: the ability to enrol a new actor by\nregistering its public key (or public hash, in our case)\n\n - Security of Transactions. This boils down to the\nimmutability of the ledger.\n\nThis is matched with a plethora of use cases mentioned\nin (Kouicem et al. 2018), see pp. 212–214. However, none\nof the bullet points is specific for the IoT.\n\nWe find our objectives to be close to those of (Dorri\net al. 2017), and that paper is a good illustration of how\ndifferent our approach is from the direction inspired by\nthe typical assumptions. The authors of (Dorri et al. 2017)\nassume, like others (see, for example, (Danzi et al. 2019)),\nthat an individual IoT device is likely to be underpowered\nfor managing blockchain transactions directly, as it does\nnot have the storage space, communication bandwidth or\nprocessing power for such a task.\n\nAs far as communications are concerned, article (Dorri\net al. 2017) correctly posits that low bit-rate radio channels, such as LoRa will be used. However it pays to\ndifferentiate between communication of a small amount\nof security-related data and unsecured, bulk public data\ntransfer.\n\n\n-----\n\nStorage-wise, to the best of our knowledge published\nresearch assumes that the blockchain either has to be\nstored at the IoT device itself (which is indeed expensive), or else trust must exist between the device and any\nstorage server. The latter assumption is not necessarily\njustified due to the availability of Content-Addressable\nStorage (CAS), which is, by construction, self-certified\nnot requiring trust or secure communications. The idea\nof CAS goes back to the late 1990’s paper (Crespo and\nGarcia-Molina 1998) where it was proposed to use a file’s\nCRC as its name, which is not quite satisfactory due to\nmassive aliasing, but a few years later paper (Quinlan and\nDorward 2002) suggested the cryptographic hashes of files\nshould serve as file names. In the last five years the leading general-purpose CAS project has been one known as\nInterPlanetary File System (IPFS) (Benet 2014) and it is\nwidely used.\nThe original Guy-Fawkes protocol on which PLS is\nbased ((Anderson et al. 1998), p.12) requires four items\nto be published in every round of the protocol, while PLS\nonly publishes three. Also, verification in a round of Guy\nFawkes requires a calculation that involves three items\nto be hashed together, whereas PLS computes a hash of\none item of a minimum size, a factor of three saving on\nthe receive side. PLS performs a symmetric decryption to\nobtain and confirm the message (or, to be precise, the message hash), which Guy Fawkes does not need. However,\ntaking an example of ESP32 (Espressif Systems 2020) as\na popular system-on-chip for IoT with a crypto accelerator, the AES-256 decryption calculation costs at most 22\nclock cycles, while computing SHA-256 requires at least\n60 clock cycles to process one block plus a minimum of\n8 cycles to produce the digest. This means that a hash is\nat least three times as expensive as the standard encryption. This is not surprising since the security of the hash\nfunction depends solely on the diffusion properties of an\niterated mapping; to achieve good diffusion as many as 80\niterations are used (64 for a shorter hash). Whereas AES\nencryption involves a key, which injects entropy in the\nprocess consequently reducing the need for iterations (or\nrounds as they are called in the area of symmetric ciphers)\nfrom the diffusion point of view: only 10 or 12 rounds are\nused. Modern accelerators have enough resources to perform data-independent calculations in parallel, so it is the\nnumber of strictly sequential rounds (where the input of\none requires full valid output of another) that determines\nthe speed.\nWe conclude that PLS is both faster and less\ncommunication-intensive at the receiver end. At the\ntransmitter end performance matters little, since the FS\nand Sequencer are not on a tight energy budget.\nWe are aware of one prior attempt at using a GF protocol in conjunction with a blockchain: (Bonneau and Miller\n2014). In that paper the blockchain itself is assumed to\n\n\nbe Bitcoin and a GF protocol is used only for signing\nvalue transfer messages (i.e. transactions). The authors of\n(Bonneau and Miller 2014) were aware of the jam-spoof\nattack (which they call race-condition theft), but their\nsolution is partial, based on a time-out whereas the V messages in our SLVP protocol capture both pre-images,\nthe current and the next ones, to defeat the jam-spoof\nattack without needing a time-out facility (but we still\nrequire the “earlier LV message wins” analysis similar to\n(Bonneau and Miller 2014)).\nRepeated application of the hash function in order to\nreduce the size of the public key was first suggested by\nWinternitz according to Merkle (Merkle 1989). We are\nnot aware of any prior work on our proposed sliding window across Winternitz chains. We use the original HORS\n(Reyzin and Reyzin 2002) procedure, but this has been\nimproved to HORST (Bernstein et al. 2015), which we can\nalso accommodate. HORST differs from HORS by the fact\nthat the public key is stored in a Merkle tree, whose root is\nauthenticated in advance (at the cost of one round of the\nSLVP protocol in our case). We prefer the original, HORS,\nas it allows us to piggy-back the public key on the SLVP\nnonce sequence by producing each nonce off the top of\nan individual Winternitz chain with the bottom kept confidential; as a result the user does not need to publish its\npublic key at all. Not only does it save us a round of SLVP,\nit obviates communication of a large public key to CAS as\nwell. However, we recognise that HORST may be useful\nfor emergency messages if the required long-term security\nnecessitates a much longer signature, in which case the\nuser must store a sufficiently large HORST Merkle tree in\nCAS in advance and bear the risk of exposing the private\nkey, held inside an IoT device, to a physical intruder.\n\n**Conclusions**\nWe have presented the architecture and protocol suite\nfor a permissioned blockchain construction based on\nthe Guy Fawkes family of protocols. Our construction\nrequires limited trust for one sealed, air-gapped unit we\ncall Sequencer, which is not internet-connected and which\nis responsible for keeping a short-term secret. If the shortterm secret is kept, we show that this type of blockchain\nwill not split and will maintain immutability. The rest of\nthe network is untrusted.\nWe have proposed a protocol for posting signed messages on the blockchain without using public-key cryptography and discussed its security. Finally, we have shown\nhow emergency (zero-latency) communications can coexist with a PLS blockchain without requiring a public key,\nyet maintaining nonrepudiation.\nThe main threat to the PLS blockchain is DoS attacks.\nWhile those cannot be fully eliminated for a radio network susceptible to jamming, we suggested the use of a\nshared secret for non-physical DoS defence: reduction of\n\n\n-----\n\nthe number of counterfeit regular and emergency messages accepted for analysis. The outcome of that analysis\ndoes not depend on the security of the shared secret,\nbut the efficiency does. Our threat model assumes that\nsustained physical and non-physical attacks will trigger\ndirection-finding and triangulation of the signal source,\neventually eliminating the threat. The advantages of our\nproposed method are the following:\n\n1. GF protocols are as convenient as public-key crypto,\nwithout having to manage keys, perform costly\nlarge-number computations on underpowered IoT\ndevices, or be exposed to quantum attacks.\n2. things can validate each other’s transactions without\ntrusting third parties.\n3. even though the blocks are not mined, passive\nreceivers of the authenticated block content (which\nwe call witnesses) that monitor radio\ncommunications are able to detect wrongly accepted\nrecords and raise an alarm without the owner (Fog\nServer) being aware of the monitoring process.\n4. validation proofs become objects in their own right,\nstored in the same CAS structure as all other\nblockchain records; they can be re-validated by any\nwitness of the blockchain at low cost based solely on\ntheir content.\n\nFuture work will define mechanisms and protocols for\nmanaging trust whereby a thing may delegate verification\nof transactions to a blockchain witness.\n\n**Acknowledgement**\nDiscussions with Bruce Christianson and his feedback are gratefully\nacknowledged.\n\n**Authors’ contributions**\nThe single author contributed 100%. The author read and approved the final\nmanuscript.\n\n**Funding**\nThis work was supported in part by IMC corporation, Slovakia, under EU\nproject BRAINE (Grant 876967).\n\n**Availability of data and materials**\nNone produced, none available.\n\n**Competing interests**\nThe author declares that he has no competing interests\n\nReceived: 20 August 2020 Accepted: 16 December 2020\n\n**References**\nAnderson R, Bergadano F, Crispo B, Lee J-H, Manifavas C, Needham R (1998) A\nnew family of authentication protocols. SIGOPS Oper Syst Rev 32(4):9–20.\n[https://doi.org/10.1145/302350.302353](https://doi.org/10.1145/302350.302353)\n\nBenet J (2014) IPFS - Content Addressed, Versioned, P2P File System. arXiv\n[1407.3561. http://arxiv.org/abs/1407.3561. Accessed 15 Aug 2020](http://arxiv.org/abs/1407.3561)\nBernstein DJ, Hopwood D, Hülsing A, Lange T, Niederhagen R,\nPapachristodoulou L, Schneider M, Schwabe P, Wilcox-O’Hearn Z (2015)\nSphincs: Practical stateless hash-based signatures. In: Oswald E, Fischlin M\n\n\n(eds). Advances in Cryptology – EUROCRYPT 2015. Springer, Berlin,\nHeidelberg. pp 368–397\nBonneau J, Miller A (2014) Fawkescoin. In: Christianson B, Malcolm J, Matyáš V,\nŠvenda P, Stajano F, Anderson J (eds). Security Protocols XXII. Springer,\nCham. pp 350–358\nCrespo A, Garcia-Molina H (1998) Archival storage for digital libraries. In:\nProceedings of the Third ACM Conference on Digital Libraries. Association\n[for Computing Machinery, New York. pp 69–78. https://doi.org/10.1145/](https://doi.org/10.1145/276675.276683)\n[276675.276683](https://doi.org/10.1145/276675.276683)\n\nDanzi P, Kalør AE, Stefanovic C, Popovski P (2019) Delay and communication\ntradeoffs for blockchain systems with lightweight iot clients. IEEE Internet\nThings J 6(2):2354–2365\nDorri A, Kanhere SS, Jurdak R, Gauravaram P (2017) Blockchain for iot security\nand privacy: The case study of a smart home. In: 2017 IEEE International\nConference on Pervasive Computing and Communications Workshops\n(PerCom Workshops). IEEE, Piscataway. pp 618–623\nEspressif Systems (2020) ESP32 Technical Reference Manual. Available as\n\n[https://www.espressif.com/sites/default/files/documentation/](https://www.espressif.com/sites/default/files/documentation/esp32_technical_reference_manual_en.pdf)\n[esp32_technical_reference_manual_en.pdf. Accessed 15 Aug 2020](https://www.espressif.com/sites/default/files/documentation/esp32_technical_reference_manual_en.pdf)\nHaxhibeqiri J, Van den Abeele F, Moerman I, Hoebeke J (2017) Lora scalability:\nA simulation model based on interference measurements. Sensors\n[2017:1193. https://doi.org/10.3390/s17061193](https://doi.org/10.3390/s17061193)\n\nKouicem DE, Bouabdallah A, Lakhlef H (2018) Internet of things security: A\ntop-down survey. Comput Netw 141:199–221\nLamport L (1979) Constructing digital signatures from a one-way function. Vol.\n238. Technical Report CSL-98. Technical report\nLeentvaar K, Flint J (1976) The capture effect in fm receivers. EEE Trans\nCommun 24(5):531–539\nLiao C, Zhu G, Kuwabara D, Suzuki M, Morikawa H (2017) Multi-Hop LoRa\nNetworks Enabled by Concurrent Transmission. IEEE Access 5:21430–21446\nMerkle RC (1989) A certified digital signature. In: Proceedings on Advances in\nCryptology (CRYPTO ’89). Springer, Berlin, Heidelberg. pp 218–238\nQuinlan S, Dorward S (2002) Venti: A new approach to archival storage. In:\nFAST Vol. 2. pp 89–101\nReyzin L, Reyzin N (2002) Better than biba: Short one-time signatures with fast\nsigning and verifying. In: Batten L, Seberry J (eds). Information Security and\nPrivacy. Springer, Berlin, Heidelberg. pp 144–153\nSentech Corporation (2019) LoRa and LoRaWAN: A Technical Overview.\nTechnical report\nWang X, Zha X, Ni W, Liu RP, Guo YJ, Niu X, Zheng K (2019) Survey on\nblockchain for internet of things. Comput Commun 136:10–29\nZhu G, Liao C, Suzuki M, Narusue Y, Morikawa H (2018) Evaluation of LoRa\nreceiver performance under co-technology interference. In: 2018 15th IEEE\nAnnual Consumer Communications Networking Conference (CCNC). IEEE,\nPiscataway. pp 21430–21446\n\n**Publisher’s Note**\nSpringer Nature remains neutral with regard to jurisdictional claims in\npublished maps and institutional affiliations.\n\n\n-----\n\n"

ERROR: type should be string, got "https://www.emerald.com/insight/2632-0487.htm\n\n# Research on financial platform of railway freight supply chain based on blockchain\n\n## Peng Xie\n\n### Institute of Computing Technology, China Academy of Railway Sciences Corporation Limited, Beijing, China\n\n## Qiang Chen and Ping Qu\n\n### Haoji Railway Corporation Limited, Beijing, China, and\n\n## Jianping Fan and Zhijun Tang\n\n### Beijing Eigpay Information Technology Corporation Limited, Beijing, China\n\nAbstract\nPurpose – This paper aims to systematically expound the theory and development background of supply\nchain finance and blockchain, design a railway freight supply chain financial platform based on blockchain,\ndetermine the risk management system and business support system of supply chain finance business and\nanalyze the value generated by the combination of supply chain finance business and blockchain.\nDesign/methodology/approach – Investigation and research method; Prototype method; Model\nmethod; Value analysis.\nFindings – The business model integrating supply chain finance and blockchain technology will bring great\nchanges to freight industry. The development of supply chain finance is beneficial to the healthy development of\nthe core participants of railway freight transport business and its upstream and downstream ecosystems. It links\ncommerce, logistics, warehousing and financial services together and builds an industry-integrated ecological\nservice platform through information technology platform and supporting system, taking data as the basis and\ncombining information technology such as blockchain as innovative means.\nOriginality/value – This paper will provide important reference value for related research. This paper\ninnovatively designs the supply chain financial platform of freight transportation industry-integrating\nblockchain technology and analyzes its business model, technical system, risk management and control system\nand value system in detail, which will provide technical support for the innovative reform of freight information\ntechnology and realize the stable and high-speed development of freight logistics informationization.\n\nKeywords Blockchain, Supply chain finance, Freight transportation, Railway logistics\n\nPaper type Research paper\n\nRailway freight transportation urgently needs the vision of transforming into modern\nlogistics and the government’s policy support for the development of supply chain finance,\n\n© Peng Xie, Qiang Chen, Ping Qu, Jianping Fan and Zhijun Tang. Published in Smart and Resilient\nTransportation. Published by Emerald Publishing Limited. This article is published under the\nCreative Commons Attribution (CC BY 4.0) licence. Anyone may reproduce, distribute, translate and\ncreate derivative works of this article (for both commercial and non-commercial purposes), subject to\nfull attribution to the original publication and authors. The full terms of this licence maybe seen at\nhttp://creativecommons.org/licences/by/4.0/legalcode\nThis paper forms part of special section “Intelligent Technology and Application of Freight\nRailway”, guest edited by Jiansheng Zhu, Ruimin Wang and Peng Xie.\n\n\n## Railway freight supply chain\n\n 69\n\nReceived 28 September 2020\nRevised 14 October 2020\nAccepted 16 October 2020\n\nSmart and Resilient\nTransportation\nVol. 2 No. 2, 2020\n\npp. 69-84\n\nEmeraldPublishingLimited\ne-ISSN: 2632-0495\np-ISSN: 2632-0487\n[DOI 10.1108/SRT-09-2020-0007](http://dx.doi.org/10.1108/SRT-09-2020-0007)\n\n\n-----\n\n## SRT 2,2\n\n 70\n\n\nwhich has become a powerful driving force for railway enterprises to carry out supply chain\nfinancial services. Under the mode of supply chain management, enterprises gradually turn\nto cross-border integration, which makes the maintenance and management of customer\nrelations become more and more important. Railway logistics enterprises can support\nprofessional financial operation through modern logistics services and make the\nrelationship between upstream and downstream enterprises closer, thus being in a\ndominant position in the competition.\nAs the railway has the basic conditions for supply chain finance, such as\nwarehouse, information system, logistics monitoring and personnel, developing\nsupply chain finance business can improve the utilization rate of assets of railway\nenterprises, reduce idle assets and explore new profit sources, and developing supply\nchain finance can become a new profit growth point for railway enterprises.\nInnovative integration of logistics industry and financial industry can reduce not only\ncustomer transaction costs but also the risks caused by information asymmetry, and\nbecome an important business model of logistics enterprises. Cooperation in the\nfreight supply chain will improve the efficiency of the whole supply chain, improve\nthe economic and social benefits significantly and introduce new channels for\nenterprise financing. In addition, freight supply chain finance can bring lower\nproduction cost and efficient supply speed to the supply chain, improve the\neffectiveness of the whole supply chain process management, reduce the financing\ncost, accelerate the overall capital turnover rate of freight supply chain and provide a\nbrand-new operation mechanism and method for capital flow and seamless\nconnection, so as to truly realize the multi-flow integration of freight logistics,\ninformation flow, capital flow, people flow and business flow.\n\n1. Development status of supply chain finance and blockchain\n1.1 Supply chain finance theory and development status\nSupply chain finance is a new direction of the development of supply chain\nmanagement and financial theory in recent years. It is the latest developed effective\nmeans to solve the financing difficulties of small and medium-sized enterprises. From\nthe bank level, supply chain finance is a professional field for commercial banks to\ncarry out credit business, while from the enterprise level, it is a channel for enterprises,\nespecially small and medium-sized enterprises, to carry out financing. Specifically,\nbanks provide financing and other settlement and financial management services to\ncore enterprises, and at the same time provide suppliers of these core enterprises with\nthe convenience of timely receipt of loans, or provide advance payment and inventory\nfinancing services to their distributors. Simply put, supply chain finance is to speed up\nthe cash flow of each enterprise in the supply chain through structured operation, so\nthat enterprises can develop their business better and faster on the basis of reasonable\ncontrol of financial costs. However, the definition of supply chain finance is very close\nto the traditional factoring business and consignment business, but there are obvious\ndifferences. Factoring and consignment business are just a simple trade financing\nproduct, whereas supply chain finance is reached between core enterprises and banks.\nIt is a systematic financing arrangement for all member enterprises in the supply chain\n(Hua and Qiang, 2017; Gu Jing et al., 2017).\n\nIn the 21st century, supply chain finance has developed rapidly all over the world, and\nfinancing models, methods and products for different industries and different levels are\nconstantly appearing. All kinds of enterprises, especially small- and medium-sized\nenterprises, actively participate in it, but it is only a new thing in practice and theory. At\n\n\n-----\n\npresent, China’s supply chain finance has developed rapidly, which can be summarized into\ntwo categories: bank operation and entity operation. First, bank supply chain finance,\nrepresented by Ping An Bank, China CITIC Bank, Minsheng Bank, Industrial and\nCommercial Bank, etc., intervenes and conducts business with traditional credit advantages\nsuch as customer resources and capital flow; second, e-commerce companies such as\nAlibaba, JD.COM Mall and Suning Tesco, as representatives, set up professional companies\nto provide financing for distributors by virtue of their advantages in commodity flow and\ninformation flow (Shiqing and Bin, 2013).\n\nThe development of supply chain finance has gone through the process of trade\nfinancing, logistics finance, 1 N supply chain finance, M 1 N supply chain finance. In\nþ þ þ\nrecent years, with the help of digital means, it has evolved more diversified. From the\nperspective of the nature of pledge and the sequence of transaction scenarios of supply chain\nfinance, the current models of supply chain finance are generally divided into three\ncategories: accounts receivable for upstream enterprise financing, inventory based on\nlogistics and transportation industry, and prepayment for downstream sales enterprises,\nfrom which dozens of specific application modes are derived. From the perspective of supply\nchain finance business model and operation platform subject, it can be divided into core\nenterprise mode, e-commerce mode, logistics company mode and third-party platform mode.\nThe business process, participants, risk control points and other related contents of supply\nchain finance are different under different modes, reflecting the differentiation and\ndiversification characteristics of supply chain financial business.\nIn recent years, the central and local governments have successively issued policies to\nsupport the development of Supply Chain Finance: in October 2017, the general office of the\nState Council issued the guiding opinions on actively promoting the innovation and\napplication of supply chain finance; in July 2019, the banking and Insurance Regulatory\nCommission of the State Council issued the guiding opinions on promoting the supply chain\nfinancial services to the real economy, etc. The successive promulgation of these policies\nputs forward new requirements for the development of supply chain finance business and\npoints out the direction for the development of business.\n\n1.2 Development status of blockchain technology\nAs an integrated application of distributed data storage, point-to-point transmission,\nconsensus mechanism, encryption algorithm and other technologies, blockchain is\nconsidered as a subversive innovation of computing mode after mainframe, personal\ncomputer and internet, which is likely to cause a new technological innovation and\nindustrial change in the global scope. Blockchain technology originated from the groundbreaking paper Bitcoin: a Peer-to-Peer Electronic Cash System published by NakamotoS in\n2008. In a narrow sense, blockchain is a kind of chain data structure that combines data\nblocks in a sequential way according to time sequence and is guaranteed by cryptography\nas an unalterable and unforgeable distributed ledger. Broadly speaking, blockchain\ntechnology is a new distributed infrastructure and computing paradigm, which uses block\nchain data structure to verify and store data, uses distributed node consensus algorithm to\ngenerate and update data, uses cryptography to ensure the security of data transmission\nand access, and uses intelligent contracts composed of automated script codes to program\nand operate data. Blockchain is a distributed data application system, which is not easy to\ntamper with, difficult to forge and traceable. The blockchain records all the information of\ntransactions. Once the data enters the blockchain, it is difficult for even internal staff to\nmake any changes without being discovered. This feature determines that it is inseparable\nfrom internet applications. The larger and richer the application scenarios are, the faster the\n\n\n## Railway freight supply chain\n\n 71\n\n\n-----\n\n## SRT 2,2\n\n 72\n\n\ndevelopment of blockchain technology and industry will be. At present, blockchain\ntechnology is called by many organizations as a major breakthrough technology to\ncompletely change the business and even the operation mode of institutions. At the same\ntime, such as cloud computing, big data, Internet of Things, artificial intelligence and other\nemerging information technologies, blockchain technology is not a single information\ntechnology, but relying on existing technology, with original combination and innovation,\nso as to realize the functions that have not been realized before.\nAccording to the current development of blockchain technology, blockchain technology\nwill experience the blockchain 1.0 mode characterized by programmable digital\ncryptocurrency system, the blockchain 2.0 mode characterized by programmable financial\nsystem and the blockchain 3.0 mode characterized by programmable society.\nAt present, it is generally believed that blockchain technology is in the early stage of 2.0\nmode. Bitcoin is one of the earliest and most successful applications of blockchain 1.0 mode,\nwhich not only provides distributed public account books for digital cash transactions, but\nalso supports programmable scripts with limited capacity. Blockchain 2.0 mode aims at\nproviding programmable infrastructure and supports saving calculation results in public\nledger. The introduction of intelligent contract (or on-chain code) which can run\nautomatically in blockchain network is a highlight. Intelligent contract is a program\ndeployed on blockchain and can run automatically, which covers programming languages,\ncompilers, virtual machines, events, state machines, fault-tolerant mechanisms, etc.\nEthereum is a representative application of blockchain 2.0 mode (Jiehui et al., 2020).\n\n2. Analysis on the development of supply chain finance business in railway\nfreight industry\n2.1 Railway freight supply chain financial business development\n2.1.1 Financial development status of railway freight supply chain. In August, 2011, China\nRailway Modern Logistics Technology Co., Ltd. and Shenzhen Development Bank reached a\ncooperation intention, realizing the formal connection between the online supply chain\nfinancial system of Shenzhen Development Bank and China Railway Modern Logistics\nManagement System, representing the arrival of the era of real-time data exchange between\nthe supply chain financial system and logistics supervision system of domestic commercial\nbanks. In 2018, the total freight volume of China’s railways will be 4.026 billion tons. The\nintelligent railway supply chain will also promote the integration of trade and supply chain\nfinancial service systems directly related to the logistics service industry chain, and will\nobtain cross-border benefits through the integration of transportation, trade and finance.\nAccording to the long-term prospect in the Medium-and Long-Term Railway Network\nPlanning, by 2030, internal and external interconnection, smooth inter-regional multichannels, high-speed railway connectivity in provincial capitals, rapid access to cities and\ncounties will be basically realized, and the largest railway logistics network in China will be\nformed. Therefore, the railway logistics and freight transportation industry will inevitably\nplay a great role in supporting the factor endowment in the future railway supply chain\necology (Fengmao, 2020; Tian, 2020).\n\nThe “integration of transportation, trade and finance” led by the railway network aims at\nthe upstream and downstream supply chain systems of all kinds of production/commerce\nindustry chains, and uses the railway network, railway logistics base and controllable pickup and delivery to “door-to-door” road freight enterprises, etc., forming an integrated system\ncarrier of “transportation, storage and distribution”, as a “1” core enterprise and a closed\ncontrol platform system in the “1 N X” of supply chain finance, through systematic\nþ þ\nfinancial arrangements, we are deeply involved in various commercial enterprises and even\n\n\n-----\n\nproduction enterprises, forming an industrial chain ecology of “five flows in one” of business\nflow, information flow, capital flow, logistics and people flow, and using the blockchain and\nsupply chain double-chain integration model. TOKEN is used to control the transportation/\ntrade cycle within the tax cycle, helping upstream and downstream enterprises in the chain\nto obtain more reasonable value-added profits and further enhance the supply chain\nstability of the industrial chain ecological closed loop (Sixin and Jun, 2017; Geng, 2014).\n\n2.1.2 Analysis of business entities and business activities of railway freight supply chain.\nBased on the analysis of business entities and business activities in the supply chain of\nrailway freight industry, we take railway coal transportation as an example to analyze\nbusiness entities and business activities in the business chain:\n\n� Background analysis of coal railway freight transportation business.\n\nConstruct the main business chain through three core enterprises: upstream coal production,\nmidstream transportation and downstream coal consumption.\nCore enterprises at different stages gather a number of small- and medium-sized\nenterprises with supply chains around them:\n\n� Supply chain finance business scenario in coal railway freight transportation.\n\nSmall- and medium-sized enterprises around the supply chain of core enterprises at\ndifferent stages include production, loading and unloading, transportation, warehousing\nand service.\nFor different types of small- and medium-sized enterprises, accounts receivable which\ncan be used for financing are formed in the business dealings with core enterprises.\nThe supply chain finance business of railway freight industry will closely focus on all\nkinds of small-and medium-sized enterprises in the whole freight business chain and\ndesign corresponding business models to meet the financing needs of all kinds of\nenterprises.\n\n2.2 Supply chain financial business needs of various enterprises in freight chain\nBuilding a supply chain financial platform around the railway freight transport industry\nand forming integrated financial services such as centralized and unified logistics,\ninformation flow and capital flow will make it possible for all kinds of cooperative financial\ninstitutions (including factoring companies, banks, insurance companies, guarantee\ncompanies, etc.) and upstream and downstream enterprises in logistics and transportation to\ncooperate closely, and achieve a win-win result for all partners .It provides upstream and\ndownstream enterprises with financial services such as procurement, distribution,\ncollection, financing, settlement and insurance, effectively mobilizes the advantages of\nlogistics resources and realizes the effective mobilization and monitoring of logistics, from\ncontrolling goods and transactions to controlling data and funds, realizing the\ncomprehensive control of supply chain risks, and realizing the efficient and low-risk supply\nchain financial service capability.\nCommodities in railway freight transportation industry are generally bulk commodities.\nTherefore, we take railway coal transportation as an example to elaborate various supply\nchain financial business models. The chain of supply chain financial value-added service\nsystem of railway coal freight industry is shown Figure 1.\n\nAs shown in Figure 1, there are many large, medium and small enterprises and many\nbusiness scenarios derived from the upstream and downstream areas of the core railway\nfreight enterprises. For the upstream coal production enterprises, to expand the coal sales\nbusiness, a series of subsidiary enterprises such as loading and unloading, transportation\n\n\n## Railway freight supply chain\n\n 73\n\n\n-----\n\n## SRT 2,2\n\n 74\n\nFigure 1.\nChain diagram of\nsupply chain\nfinancial value-added\nservice system of\nrailway coal freight\nindustry\n\n\nand storage are needed to provide services for them; for the downstream coal enterprises, to\nexpand production and purchase coal, they need a series of subsidiary enterprises such as\nloading and unloading, transportation, storage and so on to provide services for them. In the\nvarious production activities of many enterprises, a series of basic trade has been produced.\nJust like the small- and medium-sized enterprises in various industries in China, the smallscale and insufficient working capital also exist in the small-scale and insufficient working\ncapital of the small-sized and medium-sized enterprises in the railway coal freight industry\nchain, which urgently need to solve the problems of capital shortage and high financing\ncost.\n\nFocusing on the financing needs of various enterprises in the upstream and downstream\nof the railway coal freight industry chain, several types of supply chain financial business\nmodels are designed specifically: accounts receivable, financing on behalf of mining, and\nfinancing on consignment. In the supply chain financial management, the supply chain\nfinancial platform will provide goods warehousing and logistics management for all\nparticipants in the business ecosystem, provide financial intermediation for upstream and\ndownstream enterprises, give full play to the advantages of all parties in the ecosystem and\ncreate value for all parties.\n2.2.1 Accounts receivable financing. Taking the accounts receivable vouchers held by\ncoal-producing enterprises as the subject matter (pledge or transfer), it provides enterprises\nwith short-term financing business with a term not exceeding the aging of accounts\nreceivable. In this business model, the responsibilities of the supply chain financial platform\ninclude: providing logistics and warehousing information services for banks and other\nfunders and providing financing services for small- and medium-sized enterprises. For the\ncredit line provided by banks and other funders, the supply chain financial platform\nprovides online functions such as business management, process supervision and\ncomprehensive information services.\n2.2.2 Financing on behalf of procurement. Supply chain financial platform provides\nfinancing services for coal consuming enterprises with purchasing needs and also provides\nlogistics services such as purchasing, transportation, warehousing supervision and\ndistribution.\n2.2.3 Generation of sales financing. Supply chain financial platform provides financing\nservices for coal-producing enterprises with sales demand and also provides logistics\nservices such as warehousing supervision, transportation and sales.\n\n\n-----\n\n2.3 Supply chain financial business development problem analysis and new technology\nintegration\n2.3.1 Analysis of the problems existing in the supply chain finance business of the railway\nfreight industry. To promote the financial industry to improve its service capacity and\nsupport the industrial transformation and upgrading, in recent years, various ministries and\ncommissions have successively formulated a series of relevant policies to encourage the\ndevelopment of supply chain finance business in various industries. However, there are still\nmany problems and challenges in the development of traditional supply chain finance\nbusiness, which are summarized as follows:\n\n� There is an information island in the supply chain: ERP systems among enterprises\nin the same supply chain are not interoperable, which leads to information\nfragmentation among enterprises, and it is difficult to integrate information in the\nwhole chain. For financial institutions such as banks, the opaque information of\nenterprises means that risk control is more difficult, which is a huge obstacle to\ncorporate financing and penetration of financial institutions.\n\n� Core enterprise credit cannot be transferred: the information island problem leads to\nthe indirect trade information between upstream suppliers and core enterprises\ncannot be proved, while the traditional supply chain financial instruments have\nlimited ability to transfer core enterprise credit, which leads to the core enterprise\ncredit can only be transferred to the first-level supplier level, but cannot be\ntransferred across the whole supply chain.\n\n� The performance risk cannot be effectively controlled: the payment and agreed\nsettlement between the supplier and the buyer, the financier and the financial\ninstitution are limited by the contractual spirit and the willingness to perform, and\nthere are many uncertain factors, which may lead to misappropriation of funds,\nmalicious default or operational risks.\n\n2.3.2 Integration needs of blockchain technology and supply chain finance. The business\npain point in the traditional supply chain finance scenario is the advantage of blockchain, a\nnew technology. Blockchain is a fusion technology in many fields, such as point-to-point\ncommunication, digital encryption, distributed ledger, multi-party collaborative consensus\nalgorithm, etc. It has the characteristics of unchangeable and traceable data on the chain,\nand is very suitable for multi-party supply chain financial business scenarios. Through\nblockchain technology, it can ensure data credibility and mutual recognition, transfer core\nenterprise credit, prevent performance risks, improve operational efficiency and reduce\nbusiness costs. The application of blockchain in the field of supply chain finance can finally\nsatisfy the mutual confirmation and matching of multiple information sources in the supply\nchain, and solve the pain point that the funder distrusts the transaction data (Xingxiong\net al., 2018). Specific instructions are as follows:\n\n� Transaction confirmation based on encrypted data.\n\nIntangible assets that are difficult to supervise and protect under the network environment,\nblockchain is based on timestamp technology and difficult to tamper with, which has\nbecome a new method of intellectual property protection under the virtual environment. In\nrecent years, it has begun to show important application value in the field of asset\nmanagement, which promotes the real-time confirmation, authorization and transaction\nsupervision of various assets. For tangible assets, such as certificates of deposit, accounts\nreceivable and digital intelligent assets, real-world asset transactions can be realized under\n\n\n## Railway freight supply chain\n\n 75\n\n\n-----\n\n## SRT 2,2\n\n 76\n\n\nvirtual environment, such as asset authorization and use control, product traceability and\nother applications. Blockchain realizes the automatic confirmation of movable property\nrights for all participants in the supply chain, forms an unalterable rights account book, and\nsolves the pain points in the existing rights registration and rights realization:\n\n� Proof of transaction authenticity based on deposit certificate.\n\nThe proof of transaction authenticity needs to be recorded in the creditor’s rights\ninformation in the virtual world, and the consistency between the virtual information and\nthe real information must be guaranteed, which is the basis of financial services and risk\ncontrol. Supply chain finance needs to ensure that participants, transaction results and\ndocuments are based on real asset transactions. The authenticity of transaction is verified\nby manual means, which has some disadvantages such as high cost and low efficiency. To\nsolve one of the core problems of supply chain finance, that is, the authenticity of\ntransactions, it is necessary to obtain all kinds of information dynamically and in real time\nfrom the trading network under the virtual environment, and to carry out “cross-validation”\nof information to test the authenticity of transactions, which has become one of the key\ntechnologies of supply chain finance at present. Blockchain can solve this problem to a\ncertain extent through digital signature, encryption and decryption technology and various\noffline operations:\n\n� Credit disassembly based on shared ledger.\n\nThe goal of supply chain finance is to fully cover the financing of small and medium-sized\nenterprises. However, the financing needs of a large number of second-level and thirdlevel suppliers/distributors are still difficult to meet. How to provide financial services for\na large number of non-first-level suppliers and distributors in core enterprises needs to be\nfocused on. Blockchain technology can disassemble the credit of core enterprises and\npass it on to suppliers and distributors in the whole chain through shared books. Core\nenterprises can register their creditor’s rights and debts with suppliers on the blockchain\nplatform, and transfer relevant accounting vouchers step by step. The original debtor of\nthe accounting voucher is the core enterprise, so in the financing scenario of the bank or\nfactoring company, the process of reviewing the trade background can be seen at a glance\non the platform. The problem of credit transmission can be solved with the help of\nblockchain technology:\n\n� Automatic execution based on intelligent contract.\n\nIntelligent contract provides an automatic operation tool for the implementation of supply\nchain financial business, and relying on efficient, accurate and automatic contract execution,\nit can alleviate the problem of difficult contract execution in reality. Taking real right\nfinancing as an example, after delivery, payment instructions can be sent to banks through\nsmart contracts, thereby automatically completing fund payment, liquidation and financial\nreconciliation, improving business operation efficiency and reducing potential risks and\nlosses caused by human operations to a certain extent. At present, the intelligent contract\ndevelopment platforms mainly include: Hyperledger (Linux), Corda intelligent contract\nplatform (R3 Alliance), Ethereum intelligent contract platform, etc. Hyperledger (Linux) is\nwidely used in commercial projects.\n2.3.3 Integration framework of blockchain and supply chain finance. The overall\nstructure based on the integration of blockchain and supply chain financial business is\nshown in Figure 2.\n\n\n-----\n\nAmong them, the supply chain financial platform is the platform to be built by the\nrailway freight logistics industry; Other partner systems are the systems of relevant\nexternal partners. The blockchain platform is built with Hyperledger-based Fabric\ntechnology. Each module inside the blockchain platform is described as follows.\nAPI SDK: External business systems access various resources in the blockchain\nþ\nplatform network through SDK packaged into API interface, including account books,\ntransactions, chain codes, events, rights management, etc., especially the installation,\ninstantiation, upgrade and call of chain codes.\nAccount book: it is the core structure, which is responsible for recording application\ninformation, and the application records data in the account book by initiating transactions;\nTransaction: the logic of execution is carried by chain code; Events: Events occurring in the\nwhole network operation can be accessed by applications to trigger external processes and\neven other systems; Chain code: relying on container, state machine and other technologies.\nUser chain code provides state processing logic based on blockchain distributed ledger,\nwhich is written in Go/Java language and runs in Docker container.\nBooks and transactions depend on core technologies such as blockchain structure,\ndatabase and consensus mechanism. Sorting service: All transactions in the sending\nnetwork need to be sorted globally by sorting service before being verified and accepted by\nCommitter, which provides atomic broadcast sorting function.\nAuthority management: responsible for access control in the whole process, solving the\nproblem of who allows an operation in a certain scenario and using existing public key\ninfrastructure system, digital certificate, encryption and decryption algorithm and other\nsecurity technologies.\nThe bottom layer is composed of multiple nodes in peer-to-peer network, which\ncommunicates among different components through google remote procedure call channel,\nand uses Gossip protocol to provide message transmission, block distribution and state\nsynchronization among nodes.\n\n3. Design of supply chain financial platform for railway freight industry based\non blockchain\n3.1 Supply chain financial platform architecture design\nAccording to the basic processing ideas for blockchain integration of supply chain finance\nbusiness and the characteristics of railway freight industry, the overall design of supply\nchain finance platform for railway freight industry based on blockchain is shown in\nFigure 3.\n\n\n## Railway freight supply chain\n\n 77\n\nFigure 2.\nIntegrated\narchitecture diagram\nof blockchain and\nsupply chain\nfinancial business\n\n\n-----\n\n## SRT 2,2\n\n 78\n\nFigure 3.\nOverall architecture\ndiagram of supply\nchain financial\nplatform\n\n\nThe overall architecture of the platform includes supply chain financial platform, internal\nrelated system integration, external access management, third-party integrated\nmanagement, uplink processing with blockchain platform and data synchronization:\n\n� Supply chain financial platform: To realize various business management functions\nof supply chain financial business, Including the application and initiation of\nvarious businesses, subsequent business acceptance and handling, management of\nvarious financing activities, risk control quota management and basic business\nsupport management functions.\n\n� Integration of internal related systems: Realize the integrated management of related\nsystems such as customers and merchants, freight transportation and logistics\nrelated to the railway, and realize the information sharing between internal data and\nsupply chain financial platform.\n\n� External access management: To realize data access and sharing transmission with\nexternal production enterprises, logistics enterprises and other enterprises.\n\n� Third-party integrated management: To realize the integrated management with\nbanking system, national tax system, Zhongdeng system, etc. necessary for the\ndevelopment of supply chain finance business.\n\n� Interaction with blockchain platform: Realize the blockchain deposit, certification\nand uplink function of business data and document data by all parties involved in\nthe business of supply chain financial platform through blockchain technology.\n\n� Other external systems: For all kinds of external cooperation entities, such as asset\nside, capital side and other parties, access the platform according to the access\nspecification of blockchain platform, so as to realize the synchronization of\ninformation chain and data.\n\n3.2 Design of risk management and control system\nIn the field of supply chain finance, risk management has always been an important topic,\nand a perfect risk management system is an important factor in establishing the business\nmodel of supply chain finance .In the supply chain financial risk management system, new\n\n\n-----\n\ntechnologies such as big data, Internet of Things, artificial intelligence and corporate\nportraits should be used to obtain more timely and accurate data of the controlled objects\nand comprehensively identify, analyze and finely control various risks related to supply\nchain finance. The management of risk control should run through all links before, during\nand after lending: in the pre-lending link, corporate credit rating can assist financial\ninstitutions to conduct corporate credit operations and quota management for certain\nfinancing products. In the process of signing and lending in the loan, the examination and\napproval will be carried out according to the credit and quota in the pre-loan process. For the\npost-loan management link, the risk control system can monitor all kinds of information of\nthe borrower in multiple dimensions and in real time and promptly and actively push the\nrelevant personnel about the risk events and early warning reminders of the enterprise after\ndiscovering the risk events (Bin et al., 2016; Jinzhao and Ju’e, 2015).\n\n3.2.1 Risk identification and analysis. Typical risks faced by railway freight transport\nindustry in developing supply chain finance business include the following:\n\n� Fraud risk. This risk is reflected in the authenticity of the basic transactions on\nwhich the supply chain financial business depends, including possible fraudulent\nbehaviors such as constructing false transactions between buyers and sellers, or\nenterprises that do not have financing conditions pretending to be enterprises of the\nsupply chain financial transaction parties to apply for financing, or using false\ninformation such as enterprises and transactions to apply for financing.\n\n� Credit risk. According to the historical transaction records and business habits of\nenterprises participating in supply chain finance business, check the repayment\nperformance of financing entities and repayment enterprises, and check the bad\nrecords of the entities through third parties (such as national tax, courts, etc.), so as\nto comprehensively judge the credit status of the participants in supply chain\nfinance business, thus providing good basic data support for risk management and\ncontrol.\n\n� Risks such as price fluctuation of pledge. For the financing of goods, we should pay\nattention to the risks of the pledge itself, especially the position risk of the price\nfluctuation of the pledge after lending. The market price of the pledge may be\naffected by many factors and fluctuate. Therefore, the risks brought by the price\nfluctuation of pledge will be the key risks of the supply chain finance business of\ngoods rights.\n\n3.2.2 Design of wind control model. Compared with traditional financial business, the risk\ncontrol management of supply chain financial business should pay attention to not only the\nrisks of financing subjects but also the final repayment sources, so as to control the risks of\nfinancing business in all directions. The wind control management model given in this\nstudy is shown in Figure 4:\n\n� Risk control management is divided into three stages, namely, paying attention to\nthe preliminary screening and review of basic transactions and assets in the prelending stage, paying attention to the rating credit and use of financing entities and\nrelated participants in the middle lending stage, paying attention to the repayment\nof various financing businesses after lending, monitoring related parties and\ncollecting due claims.\n\n� Risk control information focuses on three dimensions and six aspects of\ninformation, namely, enterprise basic information, enterprise shareholder\ninformation and enterprise operation information, and six aspects, namely,\n\n\n## Railway freight supply chain\n\n 79\n\n\n-----\n\n## SRT 2,2\n\n 80\n\nFigure 4.\nWind control\nmanagement model\ndiagram\n\n\nenterprise basic static information, enterprise financial status, legal person basic\ninformation, major shareholder information, enterprise exchange information and\ninternal and external dynamic information. By collecting information about\ndifferent dimensions and aspects of financing entities and final repayment entities,\nwe can determine the credit line basis of enterprises, judge the current state of\nenterprises and predict the future dynamic information of enterprises, so as to better\nmanage risk control.\n\n� Multi-dimensional and multi-factor weighted risk control model algorithm, which\nclassifies various types of risk control information for risk control management,\ngives different weights to different types of information in calculation, and can\ncustomize the risk factors and the scores of each factor under various types of\ninformation. The enterprise information that meets the requirements of risk factor\ndata can finally get the total score of the enterprise under this model through this\nmodel algorithm.\n\nThe overall model is described as follows:\n\n� Multi-dimensional model division and weight setting of each dimension. Support the\nestablishment of multiple wind control models. Different dimensions can be set for a\ncertain wind control model, and different dimensions are allowed to set their own\nrisk factors. The model adopts a 100-point system, assuming that the weight of a\ncertain dimension m in the model is n, the model has k dimensions, and the weight\nof each dimension is 100%. Therefore, the calculation algorithm of the first-level\ndimension framework of the model is described as follows:\n\nTotal score of model\nFor example, for the P model, three dimensions of enterprise basic information A,\nenterprise shareholder information B and enterprise management information C are set, and\nthe full score of each dimension is 100 points. The weights of A, B and C are set to 30%, 30%\nand 40% respectively. It is necessary to calculate the scores of A, B and C, respectively, and\nthe overall score is: A according to the first-level dimension framework.\n\n\n-----\n\n� Setting multiple risk factors in each dimension. Set the risk factors included in each\ndimension of a wind control model and their weights under this dimension. Take the\nprevious example: for example, dimension A includes two risk factors A and B (each\nwith a weight of 50%), Dimension B includes four risk factors C, D, E and F (each\nwith a weight of 25%), and Dimension C includes five risk factors G, H, I, J and K\n(each with a weight of 20%). So far, a basic risk control model system with three\ndimensions and eleven risk factors has been established.\n\n� Preparation of business data input. According to all kinds of information of y\nenterprise in three dimensions of a, b and c, the business data required by eleven\nrisk factors (i.e., a, b, c, d, e, f, g, h, I, j and k) required by the risk control model are\nextracted, so as to facilitate the calculation through the risk control model.\n\n� Operational processing of wind control model. According to the data of eleven risk\nfactors that Y enterprise meets the requirements, input the data into the wind\ncontrol model, and calculate the final score of each risk factor according to the rules\ncorresponding to different data scores specified in different risk factors in the model,\nand give the final scores of each dimension after the model operation. To take the\nabove example: the input data based on eleven risk factors (i.e., A, B, C, D, E, F, G, H,\nI, J, K) and the operation and judgment process of risk control model judgment rules\nare shown in the following table: (Table 1).\n\n� Wind control score output. Y enterprise score calculation is as follows:\n\nY total score of enterprise model =\n= A * 30% B * 30% C * 40%\nþ þ\n=70 * 30% 60 * 30% 80 * 40%\nþ þ\n= 71\nAccording to the final risk control score of enterprise Y, 71 points will be scored, and then\nfollow-up business management activities such as credit rating and quota management\nrelated to supply chain finance business will be carried out.\n\n4. Value analysis of the combination of supply chain finance business and\nblockchain\n4.1 Value brought to external partners\nSupply chain finance business integrates business flow, people flow, logistics flow,\ninformation flow, capital flow and other data into the chain by using blockchain technology,\nrealizing the integration of “five flows”, ensuring data security and realizing transparent\nvisualization of data business. Buyers and sellers can better grasp the tracking logistics\n\nA B C\nDimensions\nFactor\nTo deal with a b c d e f g h i j k\n\nInput 2 3 3 2 4 2 3 4 3 5 1\ndecision rule 30:0–2 30:0–2 10:0–2 10:0–2 10:0–2 10:0–2 10:0–2 10:0–2 10:0–2 10:0–2 10:0–2\n40:2–3 40:2–3 15:2–3 15:2–3 15:2–3 15:2–3 15:2–3 15:2–3 15:2–3 15:2–3 15:2–3\n50:3–5 50:3–5 25:3–5 25:3–5 25:3–5 25:3–5 20:3–5 20:3–5 20:3–5 20:3–5 20:3–5\n\nScore of each factor 30 40 15 10 25 10 15 20 15 20 10\nScore of each dimension 70 60 80\n\n\n## Railway freight supply chain\n\n 81\n\nTable 1.\nRisk control model\nprocessing table\n\n\n-----\n\n## SRT 2,2\n\n 82\n\n\ninformation and reduce the risks of both parties through intelligent contract technology.\nGive full play to the divisible, traceable and tamperable features of blockchain technology.\nCore enterprises issue digital payment commitments to tier-one suppliers in blockchain, and\ntier-one suppliers can split the above commitments and transfer some of them to tier-two\nand tier-n suppliers according to settlement needs. Supply chain finance business takes\nadvantage of blockchain consensus mechanism and cross-validation can ensure business\nauthenticity and prevent fraud risks. At the same time, the traceability of blockchain\ntechnology should be brought into full play, so as to achieve one-to-one correspondence\nbetween receivable financing and basic business.\nThe traditional business process of confirming rights is complicated and takes a long\ntime, which affects the borrower’s experience. Issue digital assets based on blockchain,\ndigitalize the rights and interests of accounts receivable, and ensure the true expression of\ndebt subject by encryption, which is convenient for the division, circulation and\nconfirmation of rights and interests of accounts receivable, improves the liquidity of\naccounts receivable and optimizes business processes and customer experience.\n\n4.2 Value for railway freight industry\nRailway transportation, as the artery of China’s national economy and a popular means of\ntransportation, has achieved remarkable results after years of informatization construction.\nIt has established an integrated information integration platform and a number of business\ninformation systems covering passenger and cargo services, operation and management,\ndispatching and command, safety monitoring and other fields. Based on massive data\nassets, it has carried out innovative applications of big data and artificial intelligence\ntechnology, and accumulated sufficient technical foundation for the application of\nblockchain technology. At the same time, China’s railways are also facing the challenges of\nreducing costs and increasing efficiency under the new situation. The railway system is\nlarge in scale, long in business chain and involved in many fields. The use of blockchain\ntechnology can reduce the cost of supervision and management, help to reduce the overall\ncost of passenger and freight transportation of railways and even the whole society, and\npromote the high-quality development of railways in the new era.\nAt present, the application of blockchain technology is expanding from the financial field\nto other fields, and many industries are in the exploratory stage I n essence, railway\ntransportation is a multi-flow transportation organization mode, such as people flow, goods\nflow, car flow, information flow and capital flow. It is urgent to apply blockchain technology\nto realize trust, traceability and contract intelligence in the whole process of railway\ntransportation.\n\n5. Conclusion\nThe development of supply chain finance is beneficial to the healthy development of the core\nparticipants of railway freight transport business and its upstream and downstream\necosystems. It links commerce, logistics, warehousing and financial services together, and\nbuilds an industry integrated ecological service platform through information technology\nplatform and supporting system, taking data as the basis and combining information\ntechnology such as blockchain as innovative means.\nIn an era when the whole society is vigorously advocating and gradually moving\ntowards digital economy, all walks of life are seeking comprehensive digital transformation,\ntransforming and reconstructing traditional businesses through technological innovation,\ninjecting new kinetic energy into railway business expansion and cross-border integration\nand even the value enhancement of the whole industry. At the same time, emerging\n\n\n-----\n\ntechnologies represented by blockchain provide the underlying support for the development\nand innovation of information technology platforms, creating favorable conditions for the\nconstruction of integrated platform operation with “scenario-business-product-data”. In the\naspect of extended value-added services for freight transportation, this paper innovatively\ndesigns the supply chain financial platform of freight transportation industry integrating\nblockchain technology and analyzes its business model, technical system, risk management\nand control system and value system in detail, which will provide technical support for the\ninnovative reform of freight information technology and realize the stable and high-speed\ndevelopment of freight logistics informationization.\n\nReferences\nBin, Y., Weiming, Z. and Haiying, Z. (2016), “Research on supplier-led supply chain finance model”,\nFinancial Research, No. 12, pp. 175-190.\nFengmao, S. (2020), “Application of blockchain technology in supply chain finance”, Computer\nProducts and Circulation, No. 8, p. 70.\nGeng, L. (2014), “Analysis of B2C supply chain financial model in the era of internet finance”, Times\nFinance, No. 2, pp. 67-69.\nHua, S. and Qiang, L. (2017), “Supply chain financial model innovation based on virtual\nindustrial cluster: a case study of Chuangjie company”, China Industrial Economy, No. 5,\npp. 172-192.\nJiehui, L., Yan, J. and Limin, C. (2020), “Zheng Xiaozhen. Research on innovation of supply chain\nfinancial model driven by blockchain technology”, Journal of Fujian Financial Management\nCadre Institute, No. 2, pp. 14-20.\nJing, G., Xiang, C. and Xiang, D. (2017), “Research on innovation of supply chain financial model of\nSMEs”, Soft Science, Vol. 31 No. 2, pp. 83-86 97.\nþ\nJinzhao, S. and Ju’e, G. (2015), “Research on the development of supply chain finance model and\ndomestic practice from the perspective of internet”, Journal of Xi ’An Jiaotong University (Social\nScience Edition), Vol. 35 No. 4, pp. 10-16.\nShiqing, X. and Bin, H. (2013), “Analysis of three typical models of international supply chain finance”,\nEconomic Theory and Economic Management, No. 4, pp. 80-86.\nSixin, X. and Jun, Y. (2017), Strategic Choice for Railway Enterprises to Develop Supply Chain Finance.\nrailway Purchase and Logistics, Vol. 12 No. 4, pp. 30-33.\nTian, L. (2020), “Based on ‘blockchainþsupply chain finance’, the financing problem of SMEs is\nexplored”, Peasant Staff, No. 11, pp. 175- 181.\nþ\nXingxiong, Z., Qingsu, H. and Shanqi, G. (2018), “Application of blockchain technology in supply chain\nfinance”, China Circulation Economy, Vol. 32 No. 3, pp. 111-119.\n\nFurther reading\nAn, L.Y., Jian, Z. and Rong, A. (2019), “Research on the optimization of supply chain financial system\nbased on blockchain technology”, Southwest Finance, No. 1, pp. 72-79.\nFangzhi, F., Guoqiang, S. and Xiaoyan, W. (2017), “Research on credit risk evaluation and risk\nmanagement of SMEs under supply chain finance model”, Journal of Central University of\nFinance and Economics, No. 12, pp. 34-43.\nHao, J. and Wei, G. (2019), “Research on the application of new supply chain financial model in\nfinancing of small and micro enterprises”, Southwest Finance, No. 4, pp. 46-52.\nHe, Y., Zhao, Y. and Tsui, K.L. (2019), “Exploring influencing factors on transit ridership from a local\nperspective”, Smart and Resilient Transport, Vol. 1 No. 1.\n\n\n## Railway freight supply chain\n\n 83\n\n\n-----\n\n## SRT 2,2\n\n 84\n\n\nMinfeng, L. (2020), “Research on the application of financial technology in supply chain financial risk\nmanagement”, Journal of Hubei University of Economics, Vol. 18 No. 1, pp. 67-73.\nRui, W. and Jintang, D. (2018), “Supply chain finance in internet plus: new ideas for SME financing”,\nEnterprise Economics, Vol. 37 No. 2, pp. 108-114.\nShi, L., Jia, Z., Sun, H., Tian, M. and Chen, L. (2020), “Analysis of the factors influencing on bird nesting\nand its impact on railway operation”, Smart and Resilient Transport.\nWang, Y., Wei, Y., Shi, H., Liu, X., Feng, L. and Shang, P. (2019), “The unit train make-up scheme for\nloaded direction in the heavy haul railway”, Smart and Resilient Transport, Vol. 1 No. 1.\nYing, G. and Zhilai, Z. (2020), “Innovation of financing mode and path for small and micro enterprises\nunder the background of blockchain finance [J/OL]”, contemporary economic management: 1-10\n\n[[2020-07-21], available at: http://kns.cnki.net/kcms/detail/13.1356.f.2020](http://kns.cnki.net/kcms/detail/13.1356.f.2020)\n\nCorresponding author\n[Ping Qu can be contacted at: zhipengli2012@foxmail.com](mailto:zhipengli2012@foxmail.com)\n\nFor instructions on how to order reprints of this article, please visit our website:\nwww.emeraldgrouppublishing.com/licensing/reprints.htm\nOr contact us for further details: permissions@emeraldinsight.com\n\n\n-----\n\n"