Zero phase smoothing of radio channel estimates

Modern radio systems require accurate radio channel estimates in the presence of fast fading. This paper proposes an iterative radio channel estimation system where the detected and decoded symbols from the previous iteration are used as `virtual' pilot symbols to improve estimation accuracy. In each iteration of the receiver algorithm, a low order Kalman filter provides a rough estimate of the channel gains. This rough estimate is refined to a more accurate radio channel estimate with a zero phase de-noising filter. This two stage channel estimation system is capable of providing near optimal channel estimation at a much lower computational cost than prior art. It is demonstrated that the proposed radio channel estimation is capable of supporting low error data reception at high fading rates. The algorithm can be implemented efficiently on modern multicore computing systems.