Hybrid stochastic gradient for convergence of adaptive filters

摘要

The present invention provides a technique for updating the coefficients a(n) of an adaptive filter impulse response A(n), whereA′(n) = (a₀(n) a₁(n) . . . a_m(n))is the adaptive filter impulse response at time n. The impulse response is utilized for tracking the output y(n) of a communications channel, where where H is an unknown impulse response of the channel,H′ = (h₀ h₁ . . . h_m),the ′ character denoting transposition. The channel is driven with a known input signal S(n), whereS′(n) = (s(n) s(n-1) . . . s(n-m)).;According to the technique, the output x(n) of an adaptive filter, wherex(n) = S′(n) A(n)is compared with the channel output y(n) according to the Least Mean Squares (LMS) algorithm to develop an LMS error prediction signal. At the same time, the adaptive filter output x(n) is compared with the channel output y(n) according to the Sign Algorithm (SA) to develop an SA error prediction signal. If the LMS error signal is greater than or equal to a preselected value, then it is used to update the coefficients a(n) of the adaptive filter impulse response. If the LMS error signal is less than the preselected value, then the SA error signal is used to update the coefficients a(n).