A high level parallel-pipelined network architecture for adaptive signal processing

This paper investigates a recently developed modular recursive estimation (MRE) [1,2] algorithm using discrete event models (DEM). It also proposes a concurrent, pipelined adaptive signal processing architecture based on parallel networking of these MRE's. The main feature of the MRE is a discerning update strategy for parameter estimates, in contrast to the continual update strategy of conventional algorithms. Not only does this discerning update strategy result in a higher degree of modularity, but also does it facilitate more effective use of data information.