Coherent gain through a frequency domain adaptive LMS algorithm

Signal processing to achieve gain is an important area for detection, localization, and classification purposes. Classical techniques include DFT as one of the best means to achieve coherent gain. Various long term time integration methods usually follow to provide extra non-coherent gain to avoid over-resolving the signal. An alternative method investigated here is to use an adaptive algorithm to track the signal in successive DFT´s so that further coherent integration can be performed. The tracking part of this new algorithm is simular to Widrow´s LMS algorithm (B. Widrow et al., Proc. of IEEE, Vol. 63, p. 1692-1716, Dec. 1975.) Characteristics of the new algorithm will be compared with other conventional techniques based on simulation results.