Accuracy improvement of connectivity-based sensor network localization

Abstract

The early results from connectivity-based sensor network localization suffer from disappointing accuracy. The reason is partly due to the limited information of the problem, and also the deficiencies of the algorithms. This paper proposes a two-level range/indication of connectivity between each pair of nodes, which would indicate three levels of connectivity: strong, weak or nil. Theoretically, the two-level connectivity localization problem can be modeled as a non-convex optimization problem in mathematics, which contains the convex constraints and non-convex constraints. Besides using two-level range to enrich the given information, a two-objective evolutionary algorithm is also used for searching a solution. The simulation is carried out using five different topology networks all containing 100 nodes. Simulation results have shown that better solution can be obtained by using two-level range connectivity when compared with the usual one-level range connectivity-based localization.