Weighted Myriad Filters - Nonlinear Signal and Image Processing: Theory, Methods and Applications - page 187

(a)

(b)

(c)

(d)

(e)

FIGURE 5.13
Robust highpass weighted myriad filter: (a) two-tone signal corrupted by α -stable noise, α = 1 . 4,

γ = 1, (b) output of the 30-tap linear FIR filter, (c) myriadization, (d) optimal weighted median

filter, (e) optimal weighted myriad filter.

0 .

EXAMPLE: Robust Blind Equalization
The constant modulus algorithm (CMA) may be the most analyzed and de-
ployed blind equalization algorithm. Many researchers believe that the CMA
has become the workhorse for the blind channel equalization just as the least-
mean square (LMS) has for the supervised adaptive filtering. 31 Some newly
emerging communication technologies, such as digital cable TV, DSL, etc.,
are found in great favor of blind equalization implementation, mainly be-
cause their training is extremely costly if not impossible to carry out. In CMA
applications, the linear FIR filter structure is assumed by default. But for