Non iterative channel estimation and equalisation strategy for burst /spl pi//4 DQPSK transmission

An inverse linear transversal equaliser algorithm (ILTE) which uses a non-iterative channel estimation and equalisation strategy is presented. It is shown to be more efficient in terms of computational load requirements as compared to the least mean squares (LMS) or recursive least squares (RLS) approaches or the linear solution using direct matrix inversion or solutions employing matrix decomposition algorithms such as Cholesky decomposition. The computational load of the ILTE approach is shown to be less than 1% that required when the taps are derived using iterative techniques such as LMS or RLS and of less than 5% that required when using direct matrix inversion to solve the equalisation linear equation problem. The significance of the results is seen when the technique is applied in distributed ad-hoc type digital radio networks where a channel has to be estimated and equalised on a packet-by-packet basis without buffering the received packets. Examples of equalisation of typical minimum and non-minimum phase three-ray channels using the technique are presented and show a good error rate performance with the PI/4 DQPSK detector. The ILTE method, which is based on inverse filter modelling, is compared with linear solution based on Cholesky decomposition and shown to perform much better in both minimum and non-minimum phase channels.