Sensor placement in distributed sensor networks using a coding theory framework

Abstract

An important issue in the design of distributed sensor networks is the optimal placement of sensors for target location. If the surveillance region, also referred to as the sensor field, is represented as a grid (two- or three-dimensional) of points (coordinates), target location refers to the problem of pin-pointing a target at a grid point at any point in time. For enhanced coverage, a large number of sensors are typically deployed in the sensor field, and if the coverage areas of multiple sensors overlap, they may all report a target in their respective zones. The precise location of the target must then be determined by examining the location of these sensors. Target location can be simplified considerably if the sensors are placed in such a way that every grid point in the sensor field is covered by a unique subset of sensors. The sensor placement problem for target location is closely related to the alarm placement problem, which refers to the problem of placing "alarms" on the nodes of a graph G such that a single fault in the system can be diagnosed. The alarms are therefore analogous to sensors in a sensor field. It was shown by Rao (1993) that the alarm placement problem is NP-complete for arbitrary graphs. However, for restricted topologies, e.g. a set of grid points in a sensor field, a coding theory framework can be used to efficiently determine sensor placement. The sensor locations correspond to codewords of an identifying code constructed over the grid points in the sensor field.