TOWARD RELIABLE AND ROBUST WIRELESS PERSONAL AREA NETWORKS

Wireless personal area networks (WPANs) have been widely deployed for providing low power, low cost wireless connectivity that facilitates seamless operation among wireless devices centered around an individual person. However, pervasive deployment of WPANs has introduced several challenging problems on coexistence and resource provisioning.

First of all, most commercial-of-the-shelf (COTS) WPAN devices typically provide limited information of the current condition of wireless channels, which would lead to misinformed decisions and under-utilization of available resources. In this dissertation, we revisit the issue of link quality prediction in IEEE 802.15.4 low rate WPANs, and decipher RSSI, LQI readings available in Zigbee radios with TI CC2420 chipset. In order to predict the instantaneous link quality, we develop an inference model under different channel environments that uses instantaneous LQI readings as input.

Secondly, neighbor discovery and contention relationship inference are two corner stones of operating and managing ad-hoc WPANs. In this dissertation, we investigate the problem of joint neighbor discovery and contention relationship inference while most existing work focuses on one of the two problems. An active inference algorithm is proposed, called the ternary inference algorithm that utilizes decentralized randomized schedules to infer the neighboring and contention relationships through mixed signal at the receiver nodes. We analyze the sources of errors in the proposed scheme, and evaluate the impact of loose synchronization, network size, and other parameter settings.

Lastly, newly developed high rate WPAN applications pose more stringent requirements on quality-of-service (QoS), which makes it more vulnerable to network dynamics and uncertainty in wireless channel environment. We investigate the relay placement and route selection solution to combat the uncertain link failure in the IEEE 802.15.3c mmWave WPANs. Specifically, two robust problems are formulated, robust minimum relay placement (RMRP) and robust maximum utility relay placement (RMURP), with the objective to minimize the number of relays deployed and maximize the network utility, respectively. Efficient algorithms are developed to solve both problems and have been shown to incur less service disruption in the presence of moving subjects that may block the LOS paths in the environment.