Datasets:
zuh
/
ITDataset

Dataset card Data Studio Files
xet
Community
1
text
stringlengths
0
134
data to and receive data from different and changing locations. This results in the
need for various kinds of proxy solutions, where users are handed off from one proxy
to another as they move. Protocols must be constructed in such a way as to be able to
tolerate such handoffs without breaking. Mobility also raises the need for intelligent
data staging and pre-staging, so that data can be placed close to where the users will
be when they need it (particularly in slow or unreliable communications situations).
Secondly, mobility adds location as a new dimension to applications that does not
typically play a role in stationary scenarios. For example, some of the most useful
applications for mobile devices are location-centric. Consider a system that can
answer questions such as find the drugstores within 2 miles of my current location .
Such a system must track the location of the current user and be able to access
information based on relative locations and distances. On a broader scale, servers
may have to track large numbers of moving objects (people, cars, devices, etc.) and be
able to predict their future locations. For example, an automated traffic control
system would have to track numerous cars, including their current positions,
directions, and velocities. Location-centric computing requires special data structures
in which location information can be encoded and efficiently stored, as well as ones in
which the dynamic positions of objects can be maintained.
Requirements Due to Context-Awareness
Context-awareness imposes significant demands on the knowledge maintained by the
system and the inferencing algorithms that use that knowledge. In order to be context
aware, a system must maintain an internal representation of users needs, roles, and
preferences, etc. One example of such a system is a smart calendar that routes
information to a user based on knowledge of the user s near-term schedule (as can be
determined from the user s PIM calendar). If, for example, a user has a meeting with
a particular client scheduled for the afternoon, such a system could send the user
information that would be highly relevant to that meeting, such as data about the
client s account, results of previous meetings with that client, news articles relevant to
the topic of the meeting, etc.
More sophisticated systems might use various types of sensors to monitor the
environment and track users actions so as to be able to assist in the tasks the user is
performing. Such sensor-based systems require the ability to process data streams in
real-time and to analyze and interpret such streams. Thus, data-flow processing will
play a key role in ubiquitous computing.
Whether the system obtains its context information from sensors, user input, PIM
(personal information management) applications, or some combination of these, it
must perform a good deal of processing over the data in order to be able to accurately
assess the state of the environment and the intensions of the user. Thus, context-
aware applications impose demanding requirements for inferencing and machine
learning techniques. These processes will have to cope with incomplete and
conflicting data, and will have to do so extremely efficiently in order to be able to
interact with the user in a useful and unobtrusive manner.
Requirements Due to Collaboration
The final set of requirements we discuss here are those that result from the need to
support collaborative work by dynamic and sometimes ad hoc groups of people. As
stated above, a prime requirement that stems from such applications is adaptivity. In
addition, however, there are several other types of support that are required beyond
what has already been discussed. First, there is a need for synchronization and
consistency support. At the center of any collaborative application is a set of shared
data items that ean be created, accessed, modified, and deleted by participants of the
collaboration. This coordination function must be lightweight and flexible so that
many different types of interactions can be supported, ranging from unmoderated chat
facilities, to full ACID (Atomic, Consistent, Isolated, and Durable) transactions, as
provided by traditional database systems.
A second requirement stemming from collaborative applications is the need for
reliable and available storage of history. In particular, if the collaboration is to be
performed in an asynchronous manner, users must be able to access a record of what
has happened earlier in the collaboration. Likewise, if the participants in the
collaboration can change over time (e.g., due to mobility, failures, or simply due to
the nature of the collaboration), then a durable record of participants and their actions
is essential to allow new members to join and come up to speed. Such durable storage
is also useful for keeping a log of activity, that can be used later to trace through the
causes of various outcomes of the collaboration, or as input into learning and data
mining algorithms whieh may help optimize future collaborations.
Example Data Management Technologies n On-Going Projects
The preceding discussion addressed some of the data management challenges that
must be addressed to support ubiquitous computing scenarios and outlined the
application properties from which they arise. In this section, we briefly describe two
on going projects that are addressing some of these aspects. The first project, called
Data Recharging, exploits user interest and preference information to deliver data
updates and other relevant items to users on their portable devices. The second
project, called Telegraph, is building an adaptive data-flow processing architecture to
process long-running queries over variable streams of data, as would arise in sensor-
based and other highly dynamic data environments.
Challenges in Ubiquitous Data Management
Data Recharging: Profile-Based Data Dissemination and Synchronization
Mobile devices require two key resources to function: power and data. The mobile
nature of such devices combined with limitations of size and cost makes it impractical
to keep them continually connected to the fixed power and data (i.e., the Internet)
grids. Mobile devices cope with disconnection from these grids by "caching".
Devices use rechargeable batteries for caching power, while local storage is used for
caching data. Periodically, the device-local local resources must be "recharged" by
connecting with the fixed power and data grids. With existing technology, however,
the process of recharging the device resident data is much more cumbersome and
error-prone than recharging the power. Power recharging can be done virtually
anywhere in the world, requires little user involvement, and works progressively -
the longer the device is recharged, the better the device-stored power becomes. In