Source localization in wireless sensor network via second-order cone programming

Abstract

In this paper, we propose novel algorithms for source localization problem in wireless sensor networks. Under the log-normal and Gaussian models for received-signal-strength (RSS) and time-of-arrival (TOA) measurements, the maximum likelihood (ML) estimation problems have been widely studied. However, these traditional ML estimation problems are highly nonconvex. In this paper, by applying convex relaxation technique, the ML estimation problem for the RSS case is transformed into a second-order cone programming (SOCP) problem. In the case of TOA, the ML estimation problem is more complex. An iterative procedure is thereby introduced. At each iteration, the similar convex relaxation technique is also employed to formulate the ML estimation problem in an SOCP form. The proposed algorithms can be readily extended to localization problems in the minimax sense for both RSS and TOA cases.