Adaptive Forgetting-factor Gauss-Newton inverse QR-RLS per-tone equalisation for discrete multitone systems

Adaptive Forgetting-factor Gauss-Newton inverse Square-Root Recursive Least Squares (AFGN-iQRRLS) algorithm is introduced for Per-tone Equalisers in Discrete Multitone Systems. We describe shortly about the inverse Square-Root RLS algorithm which preserves the Hermitian symmetry of the inverse autocorrelation matrix by virtue of the product of square-root matrix and its Hermitian transpose. Such symmetrical property lends the benefit to the parallel implementation. By using the nonlinear algorithm, Gauss-Newton method is presented for the forgetting-factor optimisation with inclusion of RLS-based equalisation in each tone separately. Simulation results reveal that Signal to Noise Ratio and Bit Rate performance of the proposed algorithm can be ameliorated as compared to the existing Adaptive Forgetting-factor RLS algorithm.