A coherent initialization approach for DFE filters has been presented and shown to offer a significant improvement when used together with the least mean squares (LMS) algorithm. In summary, new initialization technique has been the following advantages: improves convergence at the start of the training process, enables the simple LMS to work in difficult channel conditions, allows training and tracking with fast changing channel profiles, enables the implementation of a receiver at very low signal to noise ratios and the linear operation of LMS provides better data recovery for error correcting codes. The I-LMS algorithm performs as well as the much more complex RLS algorithm.

Inspec keywords: target tracking; decision feedback equalisers; MIMO communication; adaptive equalisers; least mean squares methods; training; convergence of numerical methods; radio receivers; wireless channels

Other keywords: complex RLS algorithm; coherent initialization methods; training process; low signal to noise ratios; new initialization technique; I-LMS algorithm; error correcting codes; channel conditions; least mean squares algorithm; LMS algorithm; adaptive MIMO equalizers; DFE filters; data recovery; linear operation; fast changing channel profiles

Subjects: Radio links and equipment; Signal processing and detection; Communication channel equalisation and identification; Interpolation and function approximation (numerical analysis)