NSF Workshop on Context-Aware Mobile Database Management (CAMM)

 

 

 

Introduction

This is the first of the three planned NSF workshops. Its objectives are to (a) stimulate and focus interest in the emerging area of data management for context-aware mobile and pervasive computing, (b) establish consensus on the key research problems, (c) expose natural collaboration among the participants, (d) inform the larger research community of the interest and importance of context-aware data management, (e) provide NSF with input for future funding initiatives, (f) encourage the use of standards and shared infrastructure, and (f) create a permanent forum for evaluating mobile research activities.

The workshop aims to achieve its objectives by concentrating on the following topics.

·        Application level issues (data types, data structure, data distribution, data classification, etc.) of context-aware databases,

·        Develop and promote common platforms that can be used for cross-fertilization,

·        Develop efficient algorithms,

·        Develop prototypes, necessary testbeds, and benchmark suites or case studies for this discipline, which will evolve over time,

·        Inherent relationship between data mobility and data processing mobility (moving data, data caching, data push, data pull, data broadcast, etc.).

·        Web and data warehouse interface to mobility,

·        Mobile and intermittent connectivity (location management, handoff, reachability, optimization, etc.),

·        System security and recoverability, and

·        Topics related to terrorism (bioterrorism).

 The other major activity of the workshop will be to formally define those terms, which are frequently used in mobile computing literature. For example, terms such as “location dependent and location independent data”, “context aware data”, “mobile query”, “pervasive computing”, mobile security, and so on, are consistently used but a formal definition of the underlying concept has not appeared anywhere.  The following sections explain in detail the aspects of context-aware information processing in a mobile environment.

The Structure of the Information Space

One of the themes of the workshop is how to manage connectivity and data consistency, availability, sharability, etc., in an information space, which could be packed with thousands of mobile heterogeneous data processing nodes.  Each node could be a peer and capable of processing its needs independently or in a cooperative manner.  Client-server connectivity may also appear.  Such global continuous connectivity is necessary to achieve ubiquitous processing, which is also referred to as pervasive processing.  Thus, for example, a refrigerator can be a peer node, which can detect the predefined lower level of its contents and subsequently inform the right component (i.e., the owner of the house) about this situation.  The owner may be mobile or static and may be located anywhere in this information space.  This trigger from the refrigerator may prompt the owner to trigger next set of activities from his/her present location.  Similarly a mobile person can be fully connected and can perform necessary activities such as bill payment, sending mail, web browsing, sharing data with other node, etc., from anywhere at anytime.  It is important to note that in this information space both data and nodes are mobile and the application data requirements will depend on geographical location.  The structure of such information space can be illustrated with the following figure.

 

A fully connected information space

Conventional data processing occurs when there is only data mobility.  Conventional mobile data processing occurs when only some processing nodes are mobile.  Dynamic or ad-hoc data processing occurs when all nodes are mobile and every node is a peer.  The cooperation among these nodes to perform an activity usually requires a coordinator, which is also referred to as “Leader”.  We have, therefore, an information space where everything is dynamic, ad-hoc, and unpredictable and our job in such an environment is to maintain consistency, provide efficient data access, unlimited sharability, dependable security, system reliability, system recoverability, and adaptability.  As an example of the kind of system that we envision, consider a traffic information system that does not depend on a centralized server.  More conventional traffic information systems store all current traffic conditions on a central server, which introduces its own delays and becomes a single point of failure.  While the centralized model is easy to implement, drivers are typically only interested in information about their immediate vicinity.  Thus, in contrast, on-board computers could dynamically configure local, ad-hoc networks with other nearby cars.  A particular car would ask other cars in the area about the traffic conditions that they have encountered recently.  From the data that is returned, a model of the local traffic would be constructed.

Data mobility in distributed environments has been studied extensively and we have a number of efficient schemes for managing it.  The mobility of processing nodes is more complex, especially in the presence of continuous connectivity, intermittent connectivity, and disconnected processing.  We need a new discipline to define research directions requiring innovative solutions.  In this environment we believe that many of our conventional notions, for example, consistency, system recovery, integrity, etc., will need to be redefined.  The workshop will thoroughly discuss these issues and their efficient solutions.

Context-Aware Mobile Databases

We are surrounded by the information space that is packed with all possible types of data.  The contents of the information space are continuously changing and an application's data requirements may change with time and space.  As a result, some data may become irrelevant (e.g., previous location), some data could acquire new values (e.g., the best route to Boston), some may acquire new meaning (e.g., the rules governing sales tax), some data may become stale, some data may become hot and some may become cold (e.g., the traffic report at rush hour).  Thus, not only the component and the processing activities, the relationship and interaction among data and its environment are also highly dynamic and complex.  This phenomenon will increase as more and more devices (e.g., my cell phone) become computationally enabled and intelligent.  The study of the information infrastructure that is required to support this environment is referred to as context-aware data management.  This forms one of the central themes of this workshop.

As wireless bandwidth becomes more widely available, the communication infrastructure will be heavily wireless oriented. It will provide adaptive connectivity among immobile systems and portable devices such as cell phones, laptops, and other such future devices. Under this platform the computing will begin to migrate away from the desktop toward these devices consequently it will become necessary to managed data more carefully.  Industry and academia recognized wireless devices as an important area for development, however, the problems of data management have not been addressed in a coordinated manner. This workshop aims to fill this gap by providing a forum for leading researchers from both industry and academia in the areas of information management, wireless networking, communications and signal processing systems to set the research agenda for the future.  This workshop will lead to a more cohesive research community and will result in a quicker development and adoption of technologies for this new area. It will facilitate the development of new standards and will provide the synergy necessary for the US technical community to become a dominant player in this field.

Workshop Web

The results of the workshop will be made available online through workshop web site which cab be accessed through the following two URLs.  It will also be reported in periodicals (e.g., ACM and IEEE) and relevant conferences, workshops or meetings.

·        http://www.sice.umkc.edu/nsfmobile/wshop.html/

·        http://www.cs.brown.edu/nsfmobile/wshop.html/

Workshop model

The workshop will be conducted by invitation only.  There will be (a) invitees from academia, (b) invitees from Industries, and (c) invitees from NSF.  Total number of participants will be around 30 out of which about 10 may be from industries, about 5 from NSF and the rest from industries.  The participants in this workshop will include well-established senior researchers and new emerging researchers and advanced developers from both academia and industry involved in all aspects of mobile systems.  Each participant will present his/her original innovative work, which will be followed by extensive discussion.  We ask that the presented work should be of highest quality, original, and should be of practical value. 

There will be two invited talks.  One invited speaker will be from academia and the other will be from industries. 

Conclusions

It is well established that mobility is one of the essential trends in today's information processing fabric.  An information consumer will demand mobile access to their data from any platform available to them.  Each of the platforms with their widely varying capabilities will place unique demands on the data management infrastructure.  This set of workshops aims to provide solutions to complex problems of mobility and will set new research directions for the community. Industrial support is crucial to the objectives of this workshop. Through this workshop we would be able to reach relevant industrial organizations and together diffuse the results of our efforts into tomorrow's information access capabilities.  It should be emphasized that this series of workshops will be unique. While there are many meetings on the topic of mobility, we intend to design this series to address context-aware data management.  To our knowledge, no such series exists or is planned.  We believe that this workshop series can, thus, serve as catalyst to the formation of a more focused research program.  Depending on the results, government agencies and industrial labs may want to invest in this program in the future.