Skip to main content
. 2018 Nov 25;20(12):902. doi: 10.3390/e20120902
Algorithm 1: RMCKRSL.
Input:σ, λ, d(k), x(k)
1. Initializations: δ=0.0001,  p0=0, w(0)=0, R0=δI, R0−1=δ−1I
2. While {x(k)d(k)} available, do
3. e(k)=d(k)−wH(k)x(k)
4. κσc(e(k))=exp(−|e(k)|2/2σ2)
5. h[e(k)]=exp[λ(1−κσc(e(k)))]κσc(e(k))
6. b(k)=R^k−1−1x(k)(h−1(e(k))+xH(k)R^k−1−1x(k))−1
7. w(k)=w(k−1)+b(k)e*(k)
8. R^k−1=R^k−1−1−R^k−1−1x(k)(h−1(e(k))+xH(k)R^k−1−1x(k))−1×xH(k)R^k−1−1
9. End while
10. w^0=w(k)
Output: Estimated filter weight w^0