Data gathering in sensor networks using the energy*delay metric

Abstract

In this paper we consider the problem of data collection from a sensor web consisting of N nodes, where nodes have packets of data in each round of communication that need to be gathered and fused with other nodes’ packets into one packet and transmitted to a distant base station. Nodes have power control in their wireless communications and can transmit directly to any node in the network or to the base station. With unit delay cost for each packet transmission, if all nodes transmit data directly to the base station, then both high energy and high delay per round will occur. In our prior work [6], we developed an algorithm to minimize the energy cost per round, where a linear chain of all the nodes are formed to gather data, and nodes took turns to transmit to the base station. If the goal is to minimize the delay cost, then a binary combining scheme can be used to accomplish this task in about log N units of delay with parallel communications and incurring a slight increase in energy cost. The goal is to find data gathering schemes that balance the energy and delay cost, as measured by energy*delay. We conducted extensive simulation experiments with a number of schemes for this problem with 100 nodes in playing fields of 50m x 50m and 100m x 100m and the base station located at least 100 meters and 200 meters, respectively, from any node. With CDMA capable sensor nodes, a chain-based binary scheme performs best in terms of energy*delay. If the sensor nodes are not CDMA capable, then parallel communications are possible only among spatially separated nodes, and a chain-based 3 level hierarchy scheme performs well. These schemes perform 60 to 100 times better than direct scheme and also outperform a cluster based scheme, called LEACH [3].