The low-cost implement method of blind two-step equalization algorithm
Tao Wu, Songyin Dai, Xin Wei, Lei Yu · 2012
To reduce the computational cost of two-step equalization algorithm brought by extracting the orthogonal basis of equalizer coefficient vector space using Singularity Value Decomposition (SVD), a low-cost implement method of blind two-step equalization algorithm is proposed, which obtains the orthogonal basis of equalizer coefflcient vector space using Gram-Schmidt orthogonalization to the first P columns of the inverse of the measurement auto-correlation matrix. It reduces the computational complexity from O(K3) to KP2, where P ≪ K. An adaptive implementation of the low-cost method is presented to update the equalizer coefficient vector real time, which has the computational complexity of O(K2). Numerical simulations show that the low-cost method has an advantage of computational simplicity and shares the same performance with the origin one, and the adaptive implementation has higher convergence speed and less steady residual error than the existing adaptive algorithm at present.