Project URL: www.ee.gatech.edu/research/labs/bwn/projects/nsf www.ee.cornell.edu/~haas/nsf.html In this proposal, we deal with two central issues in Multimedia Mobile Computing, namely Mobility Management and Connection Admission. More specifically, we ask the questions of how to efficiently locate users within massively accessed internet (we refer to a network of networks as an internet), based on the micro-networking communication environment and how to limit the multimedia connection admission in such an environment, as to guarantee quality of service to already accepted connections. Our approach is based on the idea of User Mobility Profile (UMP). As a user roams within an internet, the system continuously collects information on the users' mobility and connection patterns. This information is used to anticipate the future location of the active users, thus, reduces the search space and, consequently, the cost of locating the user. This is achieved through the dynamic calculation of location area. In our approach, the location area is defined per mobile and is calculated, using the UMP, ``on the fly,'' as the user roams between different networks. Calculation of the location area is done as to minimize the user's mobility management cost. Furthermore, the statistical knowledge of users' future positions embedded within the UMP allows prediction of resource utilization and, thus, reduces the probability of resource shortage. The result is that multimedia connections can be supported with a much stricter Quality-of-Service (QoS) guarantees. To achieve the goal of guaranteed QoS, we employ a new concept, called the Shadow Clique, which is the representation of the probability of user's future locations, given its current position in the network. The Shadow Clique is, in fact, a digest of the UMP and allows easy and fast determination of the future resource requirements in different network s. In our research we will address the following issues using both analytical and simulation tools: 1. The optimal strategy for UMP information reporting and collection 2. The use of the UMP for reducing the registration requirements and the user location costs 3. Methods for grouping the networks into paging groups, as to reduce the mobile paging costs, subject to delay constraints 4. The use of the UMP to create the Shadow Clique digest 5. The use of the Shadow Clique concept to predict the future network resource utilization 6. The use of the Shadow Clique concept in the Connection Admission algorithm It is obvious that as the six topics are highly related, a close collaborative interaction and continuous exchange of information and results is essential for the success of this project. To estimate the potential impact of our work, one must envisage the future use of the Internet services and the mobile applications. These are responsible for much of the recent research in the Mobile Multimedia field. However, before a wide-spread use of the mobile applications can be realized, the fundamental questions of how to manage the mobility of such large volume of users and how to manage the resources for multimedia communication need to be addressed. Our proposal is a significant step in this direction.
