Detection and enhancement of multiple sinusoids using an adaptive filter bank structure

In this paper we investigate the use of an adaptive filter bank structure for the detection and enhancement of multiple sinusoids embedded in noise. In the proposed adaptive filter bank structure, the input signal is transformed by means of a DFT filter bank, and the transformed signals are modified by sparse adaptive subfilters. The properties of the proposed structure, such as its convergence behavior and optimal solution, when applied to the detection of multiple sinusoids, are investigated. The proposed structure is compared to the conventional FIR adaptive line enhancer structure.