RLS adaptive filter

Recursive least-squares with forgetting factor lam and initial inverse-correlation P0 = delta * I. Faster convergence than LMS at the cost of O(order^2) per sample.

Usage

morie_dsp_rls(x, d, order = 16L, lam = 0.99, delta = 100)

Arguments

x

Input (reference) vector.

d

Desired vector, same length as x.

order

Filter order (taps). Default 16.

lam

Forgetting factor in (0, 1]. Default 0.99.

delta

Initial P diagonal. Default 100.

Value

List with y, e.

References

Rangayyan & Krishnan (2015), Ch. 3, sec. 3.6; Haykin (2002).

Examples

if (requireNamespace("ranger", quietly = TRUE)) {
  set.seed(1L)
  x <- rnorm(1000)
  d <- x + 0.01 * rnorm(1000)
  out <- morie_dsp_rls(x, d, order = 8L, lam = 0.99)
  mean(out$e[200:400]^2)
}
#> [1] 1.070665