Energy-Efficient Design of Adhoc and Sensor Networks.

Sameh Gobriel (CS Grad/Pitt)

PhD Defense

Friday, February 8, 2008
11:30 a.m. - SENSQ 6106 - Eli Lilly Room

Abstract

Adhoc and sensor networks (ASNs) are emerging wireless networks that are expected to have significant impact on the efficiency of many military and civil applications. However, building ASN's efficiently poses a considerable technical challenge because of the many constraints imposed by the environment, or by the ASN nodes capabilities themselves. One of the main challenges is the finite supply energy. Since the network hosts are battery operated, they need to be energy conserving so that the nodes and hence the network itself does not expire. In my thesis different techniques for an energy-efficient design for ASN's are presented. My work spans two layers of the network protocol stack; these are the Medium Access Layer (MAC) and the Routing Layer.

In my talk I first identify and highlight the different sources of energy inefficiency in ASN's, and then I describe how each of these inefficiencies is handled. I first focus on the Medium Access (MAC) Layer and present my work that handles the wasted energy in transmission and describe how the transmission distance is optimized to extend the network lifetime. I then describe BLAM, an energy-efficient extension for the IEEE 802.11, that handles the wasted energy in collisions. Next, TDMA-ASAP, a new MAC protocol for sensor networks, is introduced. TDMA-ASAP targets the wasted energy in idle listening.

I also investigate energy-efficiency at the routing layer level. First, the ``Flooding-Waves'' problem is identified. This is a problem in any cost-based energy-efficient routing protocol for adhoc networks, different ways of solving this problem are presented. For sensor networks routing trees are usually used, I introduce a new routing scheme called RideSharing which is energy-efficient and fault-tolerant. RideSharing will deliver a better aggregate result to the end user while masking network link failures. Next, I present how to extend the RideSharing scheme to handle different link quality models. Finally, I introduce GroupBeat, a new health detection system for sensor networks, which when combined with RideSharing can deliver the information to the end user even in case of node failures.

Dissertation CO-AdviserS

Dr. Rami Melhem, Department of Computer Science
Dr. Daniel Mosse, Department of Computer Science

Committee Members

Dr. Ahmed Amer
Dr. Tarek Abdelzaher (University of Illinois at Urbana Champaign)