Recent developments in the core of digital signal processing: Adaptive filters

An adaptive filter is a self-designing system that relies on a recursive algorithm which allows the filter to perform satisfactorily in an environment where the relevant statistics are not available. Adaptive filters are classified into two main groups: linear and nonlinear. The least mean square algorithm used to design a linear adaptive filter has two major drawbacks: slow rate of convergence and sensitivity to the eigenvalue spread of the correlation matrix of the input signal vector. Whether there will be an ultimate time-frequency distribution that effectively describes all-time varying signals, or there will be a number of densities tailored for individual applications is a major issue in the field and unsettled.