Efficient Matrix Computations in Wideband Communications

Modern telecommunications put increasing demands on the efficient use of bandwidth in a channel. This is especially true in the upcoming wideband channels, where a large number of users are spread across the same large frequency band. There are two main techniques to achieve the spreading, one works ’in the time domain’ and uses randomized codes to characterize users. It is called ’Code Division Multiple Access’, and in the case of a wide band channel WCDMA. The other works in the frequency domain. It is called OFDM and utilizes orthogonal sinusoidal waveforms as carriers. In this paper we concentrate on WCDMA and show how new methods from the field of time-varying system theory can be applied to achieve efficient algorithms for signal separation of individual users. In particular, the efficient computation of different kinds of pseudo-inverses of a matrix with a jagged block-structure appear to be the main algorithms needed and they can be achieved using a state-space representation of the original matrix and keeping the state-space representation throughout the computation.