Mobility estimation using an extended Kalman filter for unmanned ground vehicle networks

An ad hoc unmanned ground vehicle (UGV) network operates as an intermittently connected mobile delay tolerant network (DTN). In this paper, we develop a mobility estimation algorithm that can be coupled with a cooperative communication routing algorithm to provide a basis for real time path planning in UGV-DTNs. A Gauss-Markov state space model is used for the node dynamics. The nonlinear measurement signals are constant-power RSSI (Received Signal Strength Indicator) signals transmitted from fixed-position base stations. An extended Kalman filter (EKF) is derived for estimating the position, velocity and acceleration of aUGV node in a two-dimensional spatial grid environment. We use Matlab to simulate a single mobile node traveling along a trajectory that includes abrupt maneuvers. Estimation performance is measured using zero-mean whiteness tests on the innovations sequences, root mean square error (RSME) of the state estimates, weighted sum squared residuals (WSSRs), and the posterior Cramer-Rao lower bound (PCRLB). Under these performance indices, we demonstrate that the mobility estimation algorithm performs effectively.