Paul D. Baxter

Paraunitary oversampled filter bank design for channel coding

Oversampled filter banks (OSFBs) have been considered for channel coding, since their redundancy can be utilised to permit the detection and correction of channel errors. In this paper, we propose an OSFB-based channel coder for a correlated additive Gaussian noise channel, of which the noise covariance matrix is assumed to be known. Based on a suitable factorisation of this matrix, we develop a design for the decoders synthesis filter bank in order to minimise the noise power in the decoded signal, subject to admitting perfect reconstruction through paraunitarity of the filter bank. We demonstrate that this approach can lead to a significant reduction of the noise interference by exploiting both the correlation of the channel and the redundancy of the filter banks. Simulation results providing some insight into these mechanisms are provided.

Robust Broadband Adaptive Beamforming via Polynomial Eigenvalues

A novel technique for robust broadband adaptive beamforming (ABF) is proposed. The technique, referred to as domain-weighted polynomial matrix eigenvalue decomposition (DW-PEVD), is founded on a basic paradigm shift from one of broadband noise cancellation to one of signal separation. It uses the second-order sequential best rotation (SBR2) algorithm to perform second order convolutive blind signal separation after applying a simple transformation to the data. The transformation is designed to exploit prior knowledge in the form of an estimated steering vector. The method is quite distinct from existing algorithms for robust broadband ABF and can offer improved performance in many cases. The results of some computer simulations that demonstrate this point are presented

Blind signal separation of convolutive mixtures

A novel algorithm is described for the blind separation of signals which have been mixed in a convolutive manner. It involves an initial process of strong decorrelation and spectral equalisation, based entirely on second order statistics. This is followed by the identification of a hidden paraunitary matrix which necessitates the use of higher (fourth) order statistics. The hidden paraunitary matrix is built up as a carefully chosen sequence of elementary paraunitary matrices.

Blind signal separation on real data: tracking and implementation

There are numerous algorithms available for blind signal separation (BSS) of multiple signals, but most of these are optimised for short blocks of data, stationary signals and time invariant mixing matrices. As such, they are unsuitable for real-world applications, which often require tracking BSS carried out in real time with as small a lag as possible. This paper looks at the problems encountered in applying BSS to real data sets and addresses the issue of computationally efficient tracking BSS based on well-understood two-stage block-based approaches. An example is included where the technique is applied to a five-minute section of twin foetal electrocardiogram (ECG) data.