Adaptive Capacity Management in Bluetooth Networks

With the Internet and mobile wireless development, accelerated by high-speed and low cost VLSI device evolution, short range wireless communications have become more and more popular, especially Bluetooth. Bluetooth is a new short range radio technology that promises to be very convenient, low power, and low cost mobile ad hoc solution for the global interconnection of all mobile devices. To implement Bluetooth network as a true mobile ad hoc wireless network operating in short radio range, highly dynamic network environment, low power, and scarce resources, many new research challenges occur, such as limited wireless bandwidth operation, routing, scheduling, network control, etc. Currently Bluetooth specification particularly does not describe in details about how to implement Quality of Service and Resource Management in Bluetooth protocol stacks. These issues become significant, when the number of Bluetooth devices is increasing, a larger-scale ad hoc network, scatternet, is formed, as well as the booming of Internet has demanded for large bandwidth and low delay mobile access. This dissertation is to address the capacity management issues in Bluetooth networks. The main goals of the network capacity management are to increase the network efficiency and throughput, reduce queueing size or delays, increase resilience, stability and fairness among users. To achieve these objectives, several adaptive distributed approaches have been proposed for dynamic capacity management in Bluetooth networks, including capacity allocation, network traffic control, inter-piconet scheduling, and buffer management. First, after a short presentation about Bluetooth technology, and QoS issues, queueing models and a simulation-based buffer management have been constructed. Then by using analysis and simulation, it shows some issues of the current Bluetooth specification, which lead to the following research to improve Bluetooth performance: Inter-piconet predictive scheduling, adaptive distributed network traffic control and hybrid distributed capacity allocation. These approaches are proposed as heuristic solutions of the convex optimization problem of maximization of the total network flows, and minimization of total cost functions, thus increase the network utilization and efficiency, while still maintaining a certain satisfactory level of Quality of Services. Their operations are distributed, comply with scarce resource constraints in Bluetooth networks and adapt to mobility and frequent changes of the network topology, as well as to bursty traffic of Internet data applications, which are supposedly very common in Bluetooth. Some performance characteristics of these approaches are illustrated by analysis as well as simulations. The result from these research are not only applicable in Bluetooth networks, but it opens many future research on Bluetooth and next generation of mobile communications.