FFT/IFFT Based Blind SIMO Channel Identiflcation
Song Wang, Jonathan H. Manton, Non-members · 2008
This paper presents an FFT/IFFT based blind identiflcation method for estimating the flnite im- pulse response of single-input multiple-output chan- nels driven by an unknown deterministic signal. The proposed algorithm successfully handles a very small size of received data, for which the existing blind channel estimation methods, including the subspace, cross-relation and shifted correlation algorithms, are known to be inefiective. Moreover, with no assump- tion of the precise knowledge of channel order, the proposed algorithm is capable of estimating channel parameters as well as detecting channel order. Sim- ulations show that the proposed algorithm outper- forms the existing methods in small sample size situ- ations. knowledge of the channel order and achieves channel parameter identiflcation and order (over)estimation. Despite difierent individual strengths, these methods share a common deflciency - they become inefiective when the observation data size is very small. This issue needs addressing because in practical commu- nication applications, there exist situations where a long data sequence is unavailable. In this paper, we present a novel blind SIMO chan- nel identiflcation method targeted at a small size of observation data. Motivated by the CR between each channel output pair, which serves as the basis of the CR approach (4), we extend the CR property to the frequency domain via the discrete Fourier trans- form (DFT) and take advantage of the computational e-ciency of the fast Fourier transform (FFT). The new method makes two contributions to blind chan- nel identiflcation. First, with a null guard interval introduced in the input, the proposed algorithm re- quires less observation data than the existing meth- ods, and a single short-duration output block su-ces for channel identiflcation. Second, assuming only the knowledge of the upper bound of channel order, the proposed algorithm is capable of estimating channel parameters as well as detecting channel order. In the case when channel order is unknown, compared to the SC approach (5) (6), our method performs better in small sample size scenarios. With FSC, the pro- posed algorithm provides a substantial performance improvement over the SC approach in the high SNR region.