Density-Based Sensor Selection for RSS Target Localization

In received signal strength target localization problems, the selection of sensor positions involved in estimation is a crucial factor in determining the network’s overall quality of estimation. In this paper, we consider a region of interest with: 1) several points of interest (PoIs) each assigned a localization error requirement and 2) candidate sensor positions at which sensors can be deployed/activated. Then, under sensor budget constraints, how can we select a limited number of sensors so that the overall localization error at the PoIs is best met in a squared error sense? Instead of directly using errors, we use corresponding sensor density information which results in the simplification of the selection problem. Using density information, we propose two novel sequential selection algorithms. In the first algorithm, selection is formulated as a sequential convex quadratic optimization problem in which the cost is a function of the difference between required and achieved sensor densities at the PoIs. Selection in the second algorithm uses density weights assigned to each candidate position and is performed in a greedy fashion. Performance evaluation of the proposed methods with respect to other algorithms shows their effectiveness.