Scheduling and simulation of large scale wireless personal area networks

As the earliest standard for Wireless Personal Area Networks (WPAN), Bluetooth has been widely used in cell phone, headset, car, GPS, etc. As a frequency hopping based system, however, constructing a large scale network using Bluetooth technology presents a real challenge. This dissertation explores this problem and presents several feasible solutions. Firstly, bridge devices, which connect multiple piconets into a connected scatternet by participating in a time division multiplex basis in adjacent piconets, need to be carefully coordinated to enable smooth operations of the scatternet; secondly, the lengthy device discovery and link setup phases make scatternets impossible to maintain, without disruptive interruptions to normal data communications. To address the bridge coordination problem efficiently and effectively, this dissertation proposes a novel distributed dichotomized bridge scheduling algorithm, coupled with an adaptive Rendezvous Window based polling scheme. A new method for device discovery is also introduced to address the scatternet formation and maintenance problems. The proposed algorithms have been tested on our own Bluetooth simulator (UCBT) which models the lower part of Bluetooth stack in detail and provides several example large scale scatternet configurations for executing our proposed scheduling algorithms. Extensive simulations have been conducted, and the performance results illustrate that large scale scatternets can operate efficiently. This dissertation also looks at applying scatternets to sensor networks by constructing a 480 nodes scatternet in our simulator. The simulation results illustrate that Bluetooth scatternet can be a good choice for low duty cycle sensor networks. The scheduling technique developed in Bluetooth scatternet can be applied to newly introduced IEEE 802.15.4 based Zigbee network as well. This is a new standard introduced to save consumed energy by defining a beacon controlled low duty cycle. Beacon collision problem presents a real challenge in any large sensor network setting. By applying scatternet technique, each adjacent cell may operate in a different channel to avoid timing critical beacon collision. Inter-cell communication can be achieved by having bridge type devices participating in multiple channels in a time division multiplex basis. Initial simulation results show our technique to be very promising.