Adaptive Long-Range Prediction of Mobile Fading

Abstract

Long-range prediction of fading in mobile systems is the key element for many fading-compensation techniques. A linear approach, which is usually used to model the time evolution of the fading process, does not perform well for long-range prediction. In this article, we propose an adaptive channel prediction algorithm by using a novel statespace model for the fading process. Our simulations show that this algorithm significantly outperforms the conventional linear method, for both stationary and non-stationary fading processes, especially for longrange predictions 1 .

Adaptive modelling and long-range prediction of mobile fading channels

“…For the Jakes fading, a is analytically available [1]. In practice, a is estimated using the fading samples using one of the well-known methods such as Levinson method, Burg method, or Prony method.…”

“…Furthermore, this is particularly helpful in the tracking mode as it is addressed in Section II-C. Equation 11can be extended to provide a D-step linear predictor [1] as followŝ…”

“…The inversion of equation (25) is a scaler inversion (not a matrix inversion), and most of the other equations only need vector calculations as m n is a constant vector.2) Model Acquisition: The current parameters for the fading model are estimated according to Section III-A. We apply the Fourier method to estimate ω(k), k = 1, • • • , N ray using an FFT algorithm[1].…”

“…n+D • • • A n+1 = diag e j{D ω n(1)+(D+•••+2+1) δω n (1)}T s , • • • , e j{D ω n (N ray )+(D+•••+2+1) δω n (N ray )}T s (80) = A D n diag e j D(D+1) Block diagram of a channel prediction scheme Block diagram of the linear prediction algorithm (LP) Block diagram of proposed prediction algorithm (KF) Estimation and tracking of the doppler frequencies Kalman tracking of the amplitudes Error history for the Kalman filter Comparison of MSE versus prediction depth for the (stationary) Jakes fading at V = 25 and V = 100 Comparison of MSE versus prediction depth for (non-stationary) RT fading at V = 25 and V = 100 LIST OF TABLES I Variables used in the Kalman filter . .…”

Adaptive multi-step channel prediction in spatial channel model using Kalman filter

“…With the reliable fading prediction technique, adaptive transmission scheme is more applicable in diverse wireless network. Linear prediction and non-linear prediction methods are compared in [4], simulation demonstrates that non-linear approach outperforms linear one in the light of MSE, but with the increase in complexity. Based on the onestep Kaiman predictor, N-frame-ahead predictor and a feedback mechanism are proposed in [5] to reduce the feedback rate.…”