In order to improve the performances of recently-presented improvednormalized subband adaptive filter (INSAF) and proportionate INSAF algorithmsfor highly noisy system, this paper proposes their set-membership versions byexploiting the theory of set-membership filtering. Apart from obtaining smallersteady-state error, the proposed algorithms significantly reduce the overallcomputational complexity. In addition, to further improve the steady-stateperformance for the algorithms, their smooth variants are developed by usingthe smoothed absolute subband output errors to update the step sizes.Simulation results in the context of acoustic echo cancellation havedemonstrated the superiority of the proposed algorithms.
展开▼