Trigonometric Expansion based Random Fourier Feature Algorithm for Nonlinear Adaptive Filtering

Traditional nonlinear adaptive filters that rely solely on either functional link network (FLN) or kernel method encounter constraints in their capabilities of nonlinear modeling. This paper presents a novel nonlinear adaptive filter that inte-grates both FLN and kernel method to improve estimation performance in dealing with nonlinear filtering problems. By employing trigonometric expansion to create FLN model and random Fourier feature (RFF) technique to design kernel method, a novel trigonometric expansion based random Fourier feature (TERFF) algorithm relying on stochastic gradient descent optimization is derived. Simulation results validate the superiority of the proposed algorithm in terms of filtering accuracy.