On Steepest Descent Adaptation: A Novel Batch Implementation of Blind Equalization Algorithms
Huy-Dung Han, Zhi Jun Ding, Junqiang Hu, Dayou Qian · 2010
Blind equalization typically achieves parameter optimization through cost minimization using stochastic gradient descent in both batch and adaptive algorithms. In general, stochastic descent algorithms typically require large number of iterations or long data samples to converge. The batch approach is generally based on data reuse (recycling) and re-filtering to recompute the cost gradient after each iterative parameter update, thereby causing long processing delays. In this work, we present a novel steepest descent batch algorithm that does not require data recycling. We consider the popular Constant Modulus Algorithm and the Minimum Entropy Deconvolution for normalized cumulant maximization. Both algorithms utilize 4-th order cumulants. The proposed steepest descent batch implementation of both algorithms converge rapidly in a few iterations and deliver superior performance without the delay due to data recycling and refiltering.