PhD Proposal

Software Update Management in Wireless Sensor Networks

Weijia Li

Friday October 16, 2009
3:00 pm - SENSQ 6329 Board Room

Abstract

Wireless sensor networks (WSNs), composed of a large number of low-cost, battery-powered sensors, and a relatively powerful sink node, have recently emerged as promising computing platforms for many nontraditional applications. Although the code running on the sensors is preloaded to them before deployment, it may still need updates for many reasons.

The code update patches are often transmitted via wireless channels, because the sensors are usually left unattended after deployment. As the code is transmitted via battery-powered wireless communication, the energy consumed in the software update can be significant, especially when it happens frequently.

The goal of this proposal is to design a software update management framework, which optimizes the overall energy consumption in WSN software update. The proposed framework includes an update-conscious compiler, a high compression ratio patch generator, a light weight code retriever, and an efficient code dissemination protocol. First an update-conscious compilation technique is used to generate the binary code. It makes newly generated binary code more similar with the base version, in order to reduce the patch size. A high compression ratio patch generator will furthermore generate the update patch to make its size even smaller. A code dissemination protocol is then used to transmit the patches over the network efficiently. After the sensor nodes receive the software update packages, they will run a light weight code retriever to regenerate the executable binary.

This research solves an important problem in WSN study. The proposed software update framework will benefit all the WSNs users by making the software update procedure faster and more energy efficient. Besides that, it is also the very first research of update-conscious compilation techniques. The motivation of similar code generation in compilation research is another great contribution of this proposal.

Dissertation Adviser

Dr. Youtao Zhang, Department of Computer Science

Committee Members

Dr. Daniel Mosse', Department of Computer Science
Dr. Bruce Childers, Department of Computer Science
Dr. Daqing He, School of Information Sciences