Tensor-Based Adaptive Filtering Algorithms

Tensor-based signal processing methods are usually employed when dealing with multidimensional data and/or systems a large parameter space. In this paper, we present family of tensor-based adaptive filtering algorithms, which suitable for high-dimension system identification problems. The basic idea is to exploit decomposition-based approach, such that the global impulse response can be estimated using combination shorter filters. algorithms mainly tailored multiple-input/single-output problems, where input and channels grouped in form rank-1 tensors. Nevertheless, approach could further extended single-input/single-output scenarios, responses (of more general forms) modeled as higher-rank As compared conventional filters, involve single (usually long) filter estimation response, achieve faster convergence rate tracking, while also providing better accuracy solution. Simulation results support theoretical findings indicate advantages over ones, terms main performance criteria.