Distributed estimation and tracking for radio environment mapping

We study distributed estimation and tracking for radio environment mapping (REM). Comparing to existing REM using centralized methods, we provide a distributed solution eliminating the central station for map construction. Based on the random field model of the REM with shadow fading effects, we adopt consensus-based Kalman filter to estimate and track the temporal dynamic REM variation. The unknown parameters of REM temporal dynamics are estimated by a distributed Expectation Maximization algorithm that is incorporated with Kalman filtering. Our approach features distributed Kalman filtering with unknown system dynamics, and achieves dynamic REM recovery without localizing the transmitter. Simulation results show satisfactory performances of the proposed method where spatial correlated shadowing effects are successfully recovered.