MS, University of Cincinnati, 2001, Engineering : Computer Science

Mobile computing has gained increasing attention recently as a variety of mobile terminals and convenient wireless connection becomes available. One of the major applications of mobile computing is information dissemination. The three fundamental methods of providing information on wireless communication channels are push, pull and hybrid push-pull. In a hybrid push-pull environment, the data of most interest is broadcast in the form of a broadcast disk, and the rest of the data is pulled from the server via explicit client requests. To achieve efficient data dissemination, the mobile server dynamically collects users' query access patterns and allocates data in broadcast disks. However, few studies have been conducted in the area of channel allocation and broadcast disk organization that considerate dynamic access pattern collection and utilization. In this study, we use a bit vector technique to represent mobile users' access patterns. We introduce a timestamp and a modification indicator to increase the accuracy of using bit vector technology. Algorithms for collecting access patterns are proposed and studied here. Algorithms for determining the optimal channel allocation and broadcast disk organization given a set of access patterns are proposed. The optimal allocation algorithm searches exhaustively for the optimal solution while the heuristic algorithm searches for a solution by finding the optimal number of broadcast channels first and then organizing the data pages on those channels. We also propose and investigate an algorithm for incremental broadcast disk reorganization. Performance studies are conducted using average access probe-wait time as a criterion. Performance studies indicate that the proposed algorithms always outperform the existing flat approach in channel allocation and broadcast disk organization. When the heuristic approach is compared with the optimal approach, experimental results demonstrate that no significant performance differences ar (open full item for complete abstract)

Committee: Dr. Karen C. Davis (Advisor) Subjects: Computer Science

MS, University of Cincinnati, 2008, Engineering : Computer Science

In this thesis work, we focus on the design of a new channel allocation mechanism for the Cognitive Radio based xG networks, based on graph theory. We propose division of the users as well as the channels on the basis of the quality of service constraints. We then analyze the performance of our mechanism incrementally in two steps: (a) Channel allocation in a static environment. (b) Channel allocation in a dynamic environment. In the first scenario, we assume a static channel configuration, wherein the primary users' behavior has been strictly defined according to a schedule. We analyze the performance of our scheme in such an environment taking into account the various performance measures like packet drop rate. We compare our scheme against the bare bones scheme where no priorities are taken into account while allocating channels. In the second part of our work, we simulate the performance of our algorithm in a dynamic environment, where the behavior of the primary users is arbitrary. In such a scenario, the algorithm must adapt to take into account the changing channel qualities which directly impact the users. We again compare our scheme based on graph theory, against the bare bones scheme where the priorities aren't taken into account as well as against the greedy scheme which simply serves the users based upon their priorities. We show that our approach delivers vast improvements in packet drop rates compared to the random channel allocation scheme.

Committee: Kenneth Berman (Committee Chair); Fred Annexstein (Committee Member); John Franco (Committee Member) Subjects: Computer Science