Mobility estimation for wireless networks based on an autoregressive model

We propose an integrated scheme for estimating the mobility state and model parameters of a user based on a first-order autoregressive model of mobility that accurately captures the characteristics of realistic user movements in wireless networks. Estimation of the mobility parameters is performed by applying the Yule-Walker equations to the training data. Estimation of the mobility state, which consists of the position, velocity, and acceleration of the mobile station is accomplished via an extended Kalman filter using measurements from the wireless network. The integration of mobility state and model parameter estimation results in an efficient and accurate real-time mobility tracking scheme that can be applied in a variety of wireless networking applications. The mobility estimation scheme can also be used to generate realistic mobility patterns to drive computer simulations of mobile networks. We validate the proposed mobility estimation scheme using mobile trajectories collected from drive-test data obtained from a live cellular network.