Optimal Sensor Density in a Distortion-Tolerant Linear Wireless Sensor Network

Abstract

The optimum sensor node density in a large linear wireless sensor network with spatial source correlation is studied. Unlike most previous works that rely on the design metric of network capacity with an error-free communication assumption, this paper performs analysis under a distortion-tolerant communication framework, where controlled distortion in the recovered information is allowed as long as the information can be recovered beyond a certain fidelity. The impacts of node density and spatial data correlation on the information distortion are investigated asymptotically by considering a large network with infinite area, infinite node numbers, but finite node density. Under fixed energy per unit area, it is discovered that: 1) for applications that only need to recover data at discrete locations, placing exact one sensor at the desired measurement locations will generate the optimum performance; 2) for applications that need to recover data at arbitrary locations in the measurement field, the optimum node density is a function of the spatial data correlation.