Abstract:
Wireless sensor networks have exhibited their powers in a wide range of applications, including military surveillance, environmental monitoring, precise agriculture, preventive monitoring, smart buildings, assisted living, and digital entertainment. However, each individual sensor device is tiny and limited, in terms of computation, memory, communication bandwidth, and power. Hence, it is extremely challenging to network all these resource limited sensor devices into powerful systems.

In this talk, I will first review four major challenges for sensor networking: radio irregularity, radio interference, the limited bandwidth, and lack of Quality of Service. Then I will briefly go through my contributions in taming these four aspects. (1) For radio irregularity, a new energy model is proposed to regenerate the irregular radio patterns, and a set of solutions are developed to conquer radio irregularity in running systems. (2) For radio interference, a runtime scheme is developed to detect the interference relations among neighboring nodes, and the results are used for better TDMA designs. (3) To improve bandwidth, a multi-frequency MAC design is proposed that considers two sensor network limits: single half-duplex radio transceiver, and small application packet sizes. (4) A QoS framework is also proposed for body sensor networks. This BodyQoS is designed asymmetric and radio-agnostic to meet the sensor networking challenges. Plus, effective bandwidth is provided for high priority streams. During this talk, I will focus on (3) and (4) in more detail.

Biography:
Gang Zhou is a Ph.D. candidate at the University of Virginia, advised by Professor John A. Stankovic. He holds an M.CS. in Computer Science from UVA (2004), and a B.S. in Computer Science from Nanjing University, China (1999). He has published 19 distinct technical papers on wireless sensor networks, wireless ad hoc networks, and content distribution networks. Several of his papers have been used as graduate course materials in top research universities, like MIT, Berkeley and UIUC, and also 4 Intel patents are pending for his work.