Optimal Robust Adaptive LMS Algorithm without Adaptation Step-Size

The main problem in the LMS algorithm is to find the most suitable step-size. Here, we propose an algorithm to solve this. While the conventional LMS uses a pre-determined constant step-size, this algorithm uses stimulating input data and error sample to determine the optimum step-size, a method that makes the algorithm more adapted with time variant channel, improves the convergence characteristics, and utilizes the training period without any increase in the complexity cost.