Cooperative adaptive estimation of distributed noncircular complex signals

The problem of distributed (cooperative) adaptive estimation of complex signals is addressed using augmented statistics and widely linear modelling, which enables optimal second order estimation of complex signals with both circular (rotation invariant) and noncircular (rotation dependent) distributions. The widely linear distributed augmented complex Kalman filter (D-ACKF) and recursive least squares (D-ACRLS) algorithms are introduced, and shown to allow for a unified treatment of the generality of complex valued signals. Further, the D-ACKF proposed here avoids the typical assumption that the observation noises at different nodes in the network are uncorrelated; thus providing enhanced performance in realworld scenarios.