Higher order statistical approach for channel estimation using matrix pencils

We study the blind identification of single input single output linear finite impulse response (FIR) channels via output higher-order cumulant information. Our aim is to find simple and effective approaches for channel estimation. The proposed algorithms identify channel impulse response by solving a cumulant matrix pencil. Compared with several existing linear approaches, the new methods are computationally simpler and less sensitive to channel order over-estimation. We also present simulation results that demonstrate the robustness of the new methods to various channel conditions in practical applications.