|M.Sc Student||Ledvich Ilya|
|Subject||Threshold-Related Throughput - A New Criterion for|
Evaluation of Sensor Network Performance
|Department||Department of Electrical Engineering||Supervisor||Professor Emeritus Adrian Segall|
Up-to-date technology makes possible the production of low-cost micro-sensor devices, which can perform short-range wireless communications and relatively complicated calculations. An ad-hoc network consisting of a large number of such devices, which are randomly distributed in a specified area, can be used in a variety of commercial and military applications. Such micro-sensor devices with small production expenses are battery powered and hence have extremely short lifeti me. Therefore, the use of wireless sensor network is ineffective without proper attention to utilization of energy resources.
Energy efficient and power aware protocols are of utmost importance in Sensor Networks. The most popular criteria so far for evaluating performance of energy-aware protocols are lifetime and throughput. One of the main contributions of the present work is to show that those criteria are often insufficient indications of the algorithm performance. Here we propose a new criterion, named threshold-related throughput, which provides a much better measure of the algorithm performance. The other main contribution is an extensive investigation of a large variety of on-line routing protocols and routing cost metrics activated on a variety of Sensor Networks topologies and initial energy configurations. Performance of these protocols and configurations is studied and compared using the new criterion.
In several previous works, the usage of devices of two types with two different power capabilities was proposed. In the second part of this work we present results of an extensive investigation of performance of several configurations of networks with two types of devices, (homogeneous, heterogeneous and hierarchical). Generally speaking, our simulations show that if the total amount of energy in the network is given, it is best to allocate it equally to all devices. Networks with two types of devices perform worse than the ones where all devices are of equal power.