hash stringlengths 32 32 | doc_id stringlengths 5 12 | section stringlengths 5 1.47k | content stringlengths 0 6.67M |
|---|---|---|---|
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.3.3.2.4.1 General | The simulator has used a Reduced TTI, related Abis improvements (product implementation) and a shorter RRBP. The latency gains from Abis improvements are 20ms per round-trip, and the shorter RRBP improves every downlink re‑transmission by 20 ms. The rest of the improvements, which is thus the major part, come from the ... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.3.3.2.4.2 Single-user cases | Reduced TTI, shorter RRBP and related Abis improvement give an end-to-end latency gain (as measured with Ping) of around 40 % in the single-user cases. Approximately 40 % of the RTTI Ping samples meet the objective in (for GERAN Evolution) of a round-trip below 100 ms in non-ideal radio conditions with the given MCSs.
... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.3.3.2.4.3 Multiple-user cases | Reduced TTI, shorter RRBP and related Abis improvement give an end-to-end latency gain (as measured with Ping) of around 35 % in the multiple-user cases. Note that the same Abis transmission time is used for all multiplexed mobiles, for a fair comparison.
For VoIP (as can be seen from the result tables and yellow marks... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.3.3.2.4.4 Capacity gain | Even though capacity gains have not been explicitly evaluated, it is clear that a shorter session time gives a capacity gain, since any pooled resources may be re-used earlier, for other sessions, from the pool.
In the simplest model, where up-to one user is located in each cell, the gain per application can be directl... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4 Variable-sized Radio Blocks | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.1 Introduction | One of the goals of GERAN Evolution is to reduce the latency between the mobile station and the Gi interface. Although a significant proportion of this latency is incurred in devices and interfaces which are outside the scope of standardization, we should not ignore the delay incurred over the radio interface.
This pap... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.2 Motivation | For the transfer of large amounts of data, the latency is primarily determined by the bandwidth available, and, for a given resource allocation and modulation/coding scheme, this is fixed. In fact, the delay (as noticed by the user) is only affected only by the delay of the last radio block in the LLC frame (since the... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.3 Concept Description | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.3.1 Overview | Under the proposed scheme, new burst mapping rules would be defined (using existing burst structures) which would allow a receiver to decode both the data and header parts of a frame using fewer than four bursts; the exact number would depend on the amount of data to be sent.
Convolutional encoding and puncturing would... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.3.2 Example: TCP ACK (52 octets) | Using the proposed scheme, an uncompressed TCP Acknowledgement (52 octets) could be mapped onto only 2 bursts, using either MCS-8 or MCS-9, allowing the receiver to decode the data and header 10 ms earlier than it would if using the existing burst mapping.
The new burst mapping for 'MCS-9_2' is shown below for the firs... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.3.3 Signalling/Detection | The signalling of the use of the new mapping schemes remains an open issue and FFS. Possible approaches may include:
- Using the stealing bits (subject to the restriction, obviously, that the receiver must be able to determine that this block should be decoded after n bursts, having received 2n stealing bits).
- Blind... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.3.4 Radio Block Capacity | The amount of (uncoded) data that can be transmitted in each radio block is shown in the table below for MCS-7 to MCS-9.
EXAMPLE: If 18 octets (or fewer) are to be sent using MCS-7 coding and puncturing, this can be sent in one burst, allowing a reduction of 15ms in the one-way delay.
Table 183
Bursts used
Reduction in... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.3.5 Retransmissions | Existing modulation and coding schemes belong to one of three families for the purpose of retransmissions. However, for the MCS7/8/9 1- and 2-burst options, the overall code rate (using all four bursts) would be lower than other code rates, and may not benefit from retransmission using a lower coding scheme.
For the 3-... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.3.6 Benefits | As stated in the introduction, the majority of the delay from MS to SGSN is incurred away from the air interface. However, the benefits described here may be combined with improvements to other interfaces/nodes.
As already stated, the latency for a TCP ACK can be reduced by 10ms; this benefit will apply (cumulatively) ... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.4 Performance Characterization | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.4.1 Bandwidth Efficiency | Obviously this proposal increases the overhead as a proportion of the amount of data sent; however, since this scheme would only be used when there was no other data to send, the reduced efficiency is not an issue. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.4.2 Latency | The delay incurred by a radio block currently is fixed at 20 ms. By using VSRB, the latency would be lower in those cases where fewer than four bursts are required to transmit the final radio block.
The figure below compares the total air interface delay for varying amounts of data to be sent. (The graph continues in ... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.4.3 Block Error Probability | Because the code rate for the reduced-sized blocks will be the same as in the current coding schemes, the block error probability will be very similar, although because of reduced diversity, the probability that a block can be decoded with just the minimum number of bursts may be slightly lower than at present.
However... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.4.4 Simulation results | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.4.4.1 Simulation Parameters | Simulations are performed for normal BTS and the physical channel used is the TU3 idFH channel. Interleaving is done by random permutation of the bits and new puncturing patterns are used to achieve the required number of output bits after coding for VSRB blocks. It should be noted that the performance of the VSRB migh... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.4.4.2 Header Error Rate | The proposed change in the way the header is mapped onto bursts is shown in figure 337. This allows the receiver to decode both the header and data having received only 2 bursts.
Figure 337: Header sent in first two bursts
Figure 338: VSRB Header Probability
In figure 338 we plot the probability of (MCS-7/8/9) header e... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.4.4.3 Equal Code Rate Comparison | As described in GP-052598, the underlying code rate for the data sent in the first two blocks would be the same as in the existing coding schemes. The proposed arrangement of data using VSRB is shown in figure 339.
Although the code rate will remain the same, the performance may suffer due to reduced frequency diversit... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.4.4.4 Equal Data Load Comparison | In practice, when transmitting small amounts of data, lower coding schemes may be used in preference to using a higher rate coding scheme with extra padding.
This has the benefit that small amounts of data are transmitted more robustly, reducing the probability that a retransmission will be required.
In order to give c... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.5 Impacts on Network Entities and Standards | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.5.1 Impacts to the Mobile Station | The following modifications are required to the mobile station:
- (Transmit) Addition of new interleaving and burst mapping schemes.
- (Receive) Addition of new de-interleaving schemes.
- (Receive) Addition of soft-combining using redundant data in 'end' bursts. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.5.2 Impacts to the BSS | The impacts on the BSS are the same as for the mobile station. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.5.3 Impacts to the Core Network | There is no impact on the core network |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.5.4 Impacts to the specifications | The impacted 3GPP specifications are listed in table 185.
Table 185
Specification
Description
3GPP TS 43.064
GPRS Stage 2
3GPP TS 44.060
Radio Link Control / Medium Access Control (RLC/MAC) protocol
3GPP TS 45.003
Channel coding |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.4.6 Comparison of VSRB and RTTI | In subclause 10.3 the use of a reduced TTI (RTTI) is described. In this subclause, we compare RTTI and VSRB.
Figure 343 shows the delay incurred over the radio interface for sending data where the allocation is 2 TS per TDMA frame; for the RTTI scheme this allows a TTI of 10 ms.
Figure 343: Delay over the radio interfa... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.5 Combining Methods | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.5.1 Preface | Latency can be improved during short transmissions by allowing earlier decoding of the final RLC block. As noted in [3] from Siemens, early decoding the final RLC block of a short transmission is a valuable contribution to reducing system latency since this decoding operation limits the earliest response time.
A number... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.5.2 Early Decode with Multi-Frequency | The use of multiple receiver frequencies inside the same TDMA period allows the early decode of a RLC frame as shown below, while maintaining a level of frequency diversity gain equivalent to legacy mobiles (diversity over 4 frequencies, denoted F1 to F4). In this case the mobile station must contain two receive paths ... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.5.3 Early Decode with VSRB | Latency can be improved when the final RLC radio block of a transmission is only partially full. The basic idea, called VSRB, is to ensure all the information needed to decode the RLC block is contained in earliest bursts, provided it fits in these bursts.
VSRB is fully described in [14] but for convenience is illustra... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.5.4 Early Decode with combined VSRB and Multi-Frequency | VSRB can be combined with multiple frequency/ multislot operation. In this case N RLC frames are sent in a single RLC period, assuming N slots are used. With legacy mobiles, each burst of the RLC frames must be transmitted over a different one of the 4 TDMA frames otherwise frequency diversity gain is lost.
During the ... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.6 Performance characterization of combined proposals | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.6.1 RTTI and Fast Ack/Nack Reporting | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.6.1.1 Simulation results for VoIP | In the following some results are shown regarding a VoIP service realized over a GERAN network that implements the following "latency reduction" features:
1. Reduced TTI radio blocks. Both "10 ms TTI" and "5 ms TTI" are considered. The "2-burst RTTI radio blocks" option is not taken into account. For the "5 ms TTI" cas... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.7 High Speed Hybrid ARQ | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.7.1 Introduction | A high speed hybrid ARQ (HS-HARQ) mechanism for the reduction of EGPRS latency is introduced. Several factors affect EGPRS latency in an ARQ mechanism including BLER, TTI, and rate of acknowledgement. The HS-HARQ mechanism increases the rate of acknowledgement by moving the ARQ mechanism from the PCU to the BTS and ope... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.7.2 Comparison of EGPRS ARQ, Fast ARQ, Reduced TTI, and HS-HARQ | Table 189 shows the round trip latency for a downlink block transmission, an acknowledgement, and a retransmission under the HS-HARQ scheme. A node delay of 5 msec is assumed at the BTS. The latency is calculated from when the BSC/PCU has started sending the block to the mobile and ends when the mobile receives the ret... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.7.3 HS-HARQ Proposal | Figure 354 shows the network architecture of the current EGPRS MAC and RLC protocols. Figure 355 shows the network architecture of the HS-HARQ mechanism. In the existing scheme the MAC and RLC protocols are both terminated at the PCU. Typically the PCU is located remote from the BTS at the BSC location or elsewhere.
Fi... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.7.4 Channel Structures | HS-HARQ RLC data blocks have identical coding to EGPRS RLC data blocks (MSC1-MCS9) with modified header contents. HS-HARQ uses a synchronous ARQ mechanism for both downlink and uplink TBFs.
For a downlink TBF the timing between a transmission and its acknowledgement over the air interface is fixed. The downlink TBF tim... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.7.5 Stop and Wait ARQ | In order to limit latency for conversational services, the HS-HARQ mechanism could allow the number of retransmissions for each block to be limited. This limit could be negotiated or implied as part of the QoS negotiation and communicated from BSS to mobile during TBF establishment.
The HS-HARQ mechanism could utilize ... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 10.8 References | [1] 3GPP TS 44.060: "Mobile Station (MS) - Base Station System (BSS) interface; Radio Link Control/Medium Access Control (RLC/MAC) protocol," Rel-6, 2005-09.
[2] 3GPP TR 45.912: "Feasibility Study on Future GERAN Evolution," V0.2.0, output from 3GPP TSG GERAN#25, Schaumburg (Il.), U.S.A., August 29 - September 2, 2005.... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11 New burst structures and new slot formats | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.1 Introduction | This subclause describes a candidate enhancement for the uplink based on the definition of a new set of transmission bursts deriving from the aggregation of timeslots at Layer 1. The aggregation relies on the removal of guard times and training sequences from a subset of bursts within a multislot allocation.
The new fo... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.2 Concept description | The idea articulates in two fundamental components: removal of training sequences and removal of guard times. The combination of these two aspects generates the new burst format.
Within a multislot allocation of n slots, the first component consists in the removal of training sequences from all slots except one. For ex... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.3 RLC/MAC Aspects | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.3.1 Introduction | The following subclauses will consider the 2-slot aggregation as an example for the discussion of the RLC/MAC Aspects. However, the principles discussed here can be easily extended to an aggregation of multiple slots. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.3.2 RLC/MAC and the New Burst Structures | When the physical layer uses the new burst structures, each TDMA frame offers to the MAC and RLC layers a larger number of bits.
Further, it is assumed that, in correspondence of a 2-slot aggregation, two RLC blocks will still be carried, i.e. one RLC block in correspondence of every timeslot, with the exception of MCS... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.3.2.1 Option 1: Increased RLC block size for the 2-slot aggregation | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.3.2.1.1 Principles | The working principles and inter-relations of RLC, MAC, and physical layer for this option are summarized as follows for the case of the 2-slot aggregation.
- Data comes uncoded from the upper layers (IP/SNDCP/LLC).
- BCS (12 bits) is added.
- Data is coded with the same convolutional mother code as EGPRS:
- Constraint... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.3.2.1.2 Pictorial representations | Figure 361 is a pictorial representation of the coding/puncturing/multiplexing process for MCS1-4 for larger block sizes.
Figure 361: Pictorial representation of MCS1-4 for 2-slot aggregated bursts
with increased RLC block size
Figure 362 is a pictorial representation of the coding/puncturing/multiplexing process for M... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.3.2.1.3 A numerology | The following table summarizes the numerology associated with the schemes described in subclause 11.3.2.1.2. This table also captures the fact that MCSs are still grouped in families (A, B, C).
All figures in bits.
The nomenclature is as follows.
Uncoded Header: Header before coding. It changes for the GMSK blocks, the... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.3.2.2 Option 2: Reduced code rate for the 2-slot aggregation | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.3.2.2.1 Principles | The working principles and inter-relations of RLC, MAC, and physical layer for this option are summarized as follows for the case of the 2-slot aggregation.
The same working principles illustrated in subclause 11.3.2.1.1 are maintained. Additionally:
- The header is assumed to be larger to indicate the multiple blocks.... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.3.2.2.2 Pictorial representations | Figure 364 is a pictorial representation of the coding/puncturing/multiplexing process for MCS1-4 for reduced code rate.
Figure 364: Pictorial representation of MCS1-4 for 2-slot aggregated bursts with reduced code rate
Figure 365 is a pictorial representation of the coding/puncturing/multiplexing process for MCS5-6 fo... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.3.2.2.3 A numerology | The following table summarizes the numerology associated with the schemes described in subclause 11.3.2.2.2. This table also captures the fact that MCSs are still grouped in families (A, B, C).
All figures in bits.
The nomenclature is as follows.
Uncoded Header: Header before coding. It changes for the GMSK blocks, the... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.3.2.3 Discussion | The previous options are summarized by the following scheme.
Table 193: Summary of RLC/MAC options for Aggregated Bursts
Option 1
Option 2
Option 3
Option 3
Basic Principle
Increase the RLC block size
Keep the code rate unchanged
Keep the RLC block size unchanged
Reduce the code rate
Increase the RLC block size
Reduce ... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.4 Performance Characterization | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.4.1 Performance calculations | For a two timeslot allocation the gain of the new slot format (consisting of removal of guard times and TSC) measured at L1 would be 18.53 %. Within a n-timeslot allocation the gain would therefore be proportional to n, as illustrated by table 194.
Table 194: Bandwidth gain of the new slot format (see note)
Allocated T... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.4.2 Link Level Simulations | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.4.2.1 GMSK Modulated Channels with legacy equalizers | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.4.2.1.1 First simulation run | Simulations have been run for the following settings:
- RawBER simulation results.
- Results based on SNR-limited scenarios (i.e. no variation of interference within the aggregated timeslot).
- Results expressed in terms of performance degradation (or lack of performance degradation) due to the timeslot aggregation, i.... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.4.2.1.2 Second simulation run | Further simulations were run independently of the former.
Those confirmed that the BER performance of aggregated new burst became worse for certain scenarios. The performance degradation of 3 burst aggregation in co-channel interference condition is listed in table 196 for GMSK channels.
Table 196: BER degradation of 3... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.4.2.2 8PSK Modulated Channels with legacy equalizers | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.4.2.2.1 First simulation run | Simulations have been run with the same settings as subclause 11.4.2.1.1. Additionally
- "EQ-A" is a reference EDGE equalizer, essentially replicating the same performance of an EDGE mobile receiver available today.
- Note that the complexity of running EQ-A over a 2-slot aggregation is lower than running EQ-A twice ov... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.4.2.2.2 Second simulation run | Further simulations were run independently of the former.
Those confirmed that the BER performance of aggregated new burst became worse for certain scenarios. The performance degradation of 3 burst aggregation in co-channel interference condition is listed in table 198 for 8PSK channels.
Table 198: BER degradation of 3... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.4.2.3 8PSK Modulated Channels with advanced simulation settings | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.4.2.3.1 First simulation run | Simulations for more complex cases have been run with the same settings as subclause 11.4.2.1.1. Additionally, Random SNR variations on a per-timeslot basis were simulated.
Effectively, this has been done by applying a random variation ( 1 dB) around the "nominal SNR" on a per-timeslot basis within a 2-slot and a 3-sl... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.4.2.4 Interference limited scenarios | Interference Limited scenarios are relevant to an evaluation of the new formats from two points of view.
First, the SINR could change within one aggregated slot in those scenarios where the interference conditions change on a per-timeslot basis.
Secondly, some interference cancellation implementation at the receiver (i... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.4.2.5 RLC Simulations | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.4.2.5.1 RLC Simulations with New Coding Schemes | As the aggregated RLC block can provide the longer data block length before channel coding, so using turbo codes as channel coding become realistic. In this subclause, turbo codes is introduced in the aggregated RLC data block and LA and IR simulation results are provided based on new defined coding schemes.
This subcl... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.5 Additional technical aspects | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.5.1 Influence of TSC Position | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.5.1.1 New slot formats in simulation | In modern mobile communication, coherent time of the wireless channel is defined as [1].
is the wave-length of carrier, is the velocity of mobile terminal and is the maximum Doppler frequency. At 250 km/h, is about 2.03 ms for 900 MHz GSM system, a little shorter than the time duration of 4 legacy slots. So the maxi... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.5.1.2 Simulation Results | In the following, the simulation results are presented to show the influence of the TSC position in the new slot on the link performance.
In the simulation, both GMSK and 8PSK modulation are applied. The wireless channel model used in the simulation is TU50, HT100 and RA250, which defined in [2].
Nomenclature is as fol... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.5.2 BER Distribution Aspects | With the aggregated bursts, the bit error rate would vary significantly over the whole burst. This is illustrated in figure 372.
NOTE: Higher BER is experienced by every third bit in the 8 PSK symbol hence there are two distinct traces of BER profiles in the plot for each physical channel mode.
Figure 372: Bit error pr... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.5.3 Relationship between performance penalty and aggregation size | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.5.3.1 Simulation Setting | The purpose of this subclause is to investigate the degradation structure of the new bursts.
The simulation is designed to have a 26-symbol training sequence in the middle of the aggregated burst, and 4 segments of symbols are placed on each side of the training sequence and 3 tail symbols on each end as shown in figur... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.5.3.2 Simulation Results and Analysis | The raw BER of the Segment-I, Segment-II, Segment-III, and Segment-IV are shown in figure 374.
The Segment-I performance can be considered as the performance of the standard burst. The equalizer is a well-known adaptive equalizer, which is per-survivor based adaptive RSSE. It is shown that the penalty accelerates as ea... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.6 New Burst Structures with Turbo Codes | According to the simulation results, even though the TSC is placed in the middle of the new burst, the BER performance of new burst structure will degrade under many scenarios, especially under faster fading channel. There are many ways to reduce the degradation, such as increasing the trellis size of the equalizer, us... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.6.1 BER Degradation of New Burst | The BER performance of aggregated new burst became worse. The performance degradation of 3 burst aggregation in co-channel interference condition is listed in tables 206 and 207.
Table 206: BER degradation of 3 burst aggregation compared to legacy burst, GMSK
Scenario
8 % BER
1 % BER
TU3iFH 900 MHz
0 dB
0 dB
TU50 900 M... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.6.2 New Coding schemes and Simulation | The target of GERAN evolution is to improve the data throughput. In [2], RLC aggregation is proposed to get more gains of data throughput, but the gains listed in the paper is the theoretical results. No proposals to get the gains are provided.
As the aggregated RLC block can provide the longer data block length before... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.6.2.1 Coding schemes | Refer to the current EGPRS uplink MCS coding schemes, new uplink coding schemes is defined in the simulation, listed in table 208. It is defined as the same data rate (per legacy burst) with legacy MCS. MB3MCS10 and MB3MCS11 are added to the coding schemes to reach peak data rate. In MB3MCS1 to MB3MCS9, the redundancy ... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.6.2.2 Simulation setting | New burst of 3 aggregated legacy burst are considered in simulation and ideal LA (link adapt ion) and IR (incremental redundancy) assumed.
Scenarios:
- Co-channel interference.
Equalizer type:
- GMSK: MLSE + PSP (LMS).
- 8PSK: RSSE + PSP (LMS).
- The equalizer is not optimized and based on normal implementation complex... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.6.2.3 Simulation results | The curves in figures 376 to 379 show data throughput vs. C/I of new defined MCS and standard MCS. Different gains of new defined MCS against standard MCS with different C/I can be seen in the figure. The CDF of C/I presented by TeliaSonera in [4] indicates very few occurrences of C/I above 22 dB. According to the C/I ... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.6.3 Compatibility | Compatibility problems are reconsidered here. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.6.3.1 Impact to the current frequency planning | Coexist with current frequency planning. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.6.3.2 Multiplexing loss with legacy EGPRS | No loss in multiplexing with legacy EGPRS when used in uplink. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.6.3.3 Impact to BTS | No hardware change needed. Turbo decoding can be implemented by software upgrade. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.6.3.4 Applicable of DTM | Yes. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.6.3.5 Applicable for the A/Gb mode interface | Yes. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.6.3.6 Impact to the mobile station | For MS that already supports both GERAN and RAN, the existing Turbo coding capability can be re-used from RAN capability. Otherwise, the capability would need to be included. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.7 Timeslot Aggregation for RTTI TBF and Link Performance | |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.7.1 Introduction | Reduced TTI (RTTI) and Fast Ack/Nack Reporting (FANR) are accepted as effective enhancements to reduce latency [14], [15] and [16]. To get room for the FANR short bitmap, the coding schemes (convolution, puncturing and interleaving) of RLC data shall be modified due to the reduction of RLC data payload, or the RLC data... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.7.2 Timeslot aggregation for RTTI | At least two timeslots will be allocated to a RTTI TBF. If the two timeslots are consecutive, they can be aggregated to form a fat timeslot, thus can bear a fat burst. [18] That is, the radio blocks would still span two TDMA frames after aggregation, but would now consist of two "aggregated" bursts, instead of four ord... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.7.3 Definition of the new coding schemes | NBRTTI (New Burst RTTI) coding schemes are defined as shown in table 211.
Table 211: Definition of new coding schemes
MCS-2
w/o bitmap
RTTI
2 w/ bitmap
NBRTTI
2 w bitmap
MCS-6
w/o bitmap
RTTI
6 w/ bitmap
NBRTTI
6 w/ bitmap
Raw
Header
31
31
31
37
37
37
Bitmap
0
20
32[1]
0
20
32[1]
Data
226
194
226
594
530
594
Coded
(+CR... |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.7.4 Header and Data coding | The coding of RLC header and data is kept unchanged. Only the burst mapping is changed in order to allow Reduced TTI operation and aggregated bursts. The header and data bits are mapped on two aggregated fat bursts spanning two consecutive TDMA frames. The process is illustrated in the figure 380. |
5692480e365e21a881d790cc4eaea394 | 45.912 | 11.7.5 Bitmap coding | The bitmap payload has a length of 32 bits, thus is more robust or able to take more Ack/Nack information compared with 20 bits usually used for FANR. The coding of the bitmap is independent of RLC header and data.
To get a similar code rate as RLC header, 8 CRC bits are added to the bitmap payload. The method for gene... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.