An EM Technique for Multiple Transmitter Localization

We propose an expectation-maximization (EM) technique for locating multiple transmitters based on power levels observed by a set of arbitrarily-placed receivers. Multiple transmitter localization is of interest for uncoordinated cognitive radio systems, which must identify and transmit over unused radio spectrum without cooperation from conventional transmitters. We employ the EM algorithm to reduce the dimensionality of the maximum-likelihood estimation problem. Because the EM algorithm finds only a locally optimal solution, we explore the use of clustering to generate "smart" initial estimates of the transmitter locations. Simulation results show that, as the number of sensors increases, the proposed EM technique achieves gains of up to an order of magnitude over constricted particle swarm optimization, a popular global optimization technique.