Prediction of the number of coherent signals for mobile communication systems using autoregressive modeling

Abstract

AbstractOne effective technique for reducing multipath fading in mobile communications is introduction of directional diversity with the help of adaptive arrays. This technique requires direction of arrival (DOA) estimation of coherent signals and relative time delay estimation. Eigenvector analysis or maximum likelihood localization is an efficient technique to estimate DOA. However, for coherent signals, spatial smoothing preprocessing is required and information about the number of signals is needed.This paper proposes to use spatial autoregressive modeling and final prediction error (FPE) to estimate the number of coherent signals. Since FPE drastically decreases with p when the length of prediction filter p is less than the number of coherent signals k and FPE converges to a constant value determined by C/N when p is greater than k, the number of coherent signals can be estimated. Moreover, by appropriately defining the FPE threshold, the number of dominant coherent signals can be estimated under the C/N concept. Numerical experiment indicates that this technique can estimate the number of signals and DOA with sufficient accuracy. Further, by appropriately defining the FPE threshold, the number of dominant coherent signals can be estimated.