Deployment Of Visual Sensor Networks For Redundant Data Acquisition

This paper deals with the problem of deploying a sensor network to monitor an area. More specifically, we consider sensor nodes that are equipped with smart cameras. We refer to this network as visual sensor network (VSN). The cameras used in this study are directional and must be fixed in study field to ensure target coverage in an indoor environment. The optimization problem has three objective functions: (i) maximize the target coverage present in the study area, (ii) maximize the redundant observation of those targets and (iii) maximize the average redundancy. To deal with this problem, we propose an approach based on a genetic algorithm and a simulation tool to measure the cameras’ fields of view (based on ray tracing). The results obtained show the effectiveness of our method through comparison with the state of the art, where the proposed coverage is practically 7% higher, and its adaptation through coverage reaching 100% for more positions of freedom for our cameras. Moreover, the algorithm succeeds in simultaneously optimizing all three objective functions.