Robust adaptive semiblind channel identifier

A novel adaptive channel identifier based on least-squares techniques is provided. An additional requirement that relatively short training sequences are used to initialise the proposed algorithm is further introduced, leading to the so called semiblind approach. Simulations show that when the subchannels share common root(s), a condition notorious for being the main difficulty in the applicability of second order methods, the proposed technique is more robust than the existing algorithms