Uncooperative Rss-Based Emitter Localization In Uncalibrated Mobile Networks

This paper explores the problem of localizing an emitter of radio frequency energy using a network of mobile, uncalibrated receiver nodes. We show that this received signal strength localization problem can be expressed in the form of a nonlinear mixed effects model, by extending the log-distance path loss model to include random biases. Doing so models drop in received signal strength due to distance (path loss) as well as the individual link biases from an uncalibrated network. We estimate the unknown bias and noise variance parameters via closed-form variance least squares expressions. These estimates are then applied as weights in the nonlinear least squares algorithm, alternating until convergence. Our simulations show a substantial performance improvement over standard nonlinear least squares, comparable to that of the maximum likelihood estimator for our model. Our algorithm is much simpler to implement vs. maximum likelihood, and also allows the assumption of Gaussian noise to be dropped.