CORDIC scenario for Kalman-based channel estimation

In this paper we discuss the use of special-purpose hardware to support radio channel estimation. This hardware will parallelize some of the operations in the matrix inversion, tap predictor, and channel compensation, which in turn allows higher order dynamic models to be used in the Kalman filtering operation without increasing computation time. In particular, we analyze CORDIC computing units, and propose a computing scenario that will allow Kalman filtering to be used effectively in the process of channel estimation. The main gain or our approach is that a large amount of computation can be expressed in terms of CORDIC primitives, and, therefore, can be deeply pipelined. This result is very promising when large amount of data are to be processed, especially in the context of joint channel estimation for multiple channel reception.