Robust Distributed Positioning Algorithms for Cooperative Networks

Paper in proceeding, 2011

The problem of positioning targets based on distance estimates is studied for cooperative wireless sensor networks when there is limited a priori information about measurements noise. To solve this problem, two different methods of positioning are considered: statistical and geometrical. Based on a geometric interpretation, we show that the positioning problem can be rendered as finding the intersection of a number of convex sets. To find this intersection, we propose two different methods based on projection onto convex sets and outer-approximation. In the statistical approach, a partly novel two-step linear estimator is proposed which can be expressed in a closed-form solution. We also propose a new constrained non-linear least squares algorithm based on constraints derived in the outer-approximation approach. Simulation results show that the geometrical methods are more robust against non-line-of-sight measurements than the statistical approaches while in dense networks with line-of-sight measurements statistical approaches outperform geometrical methods.