Anthony D. Fagan

The Gauss-Seidel fast affine projection algorithm

In this paper we propose a new stable fast affine projection algorithm based on Gauss-Seidel iterations (GSFAP). We investigate its implementation using the logarithmic number system (LNS) and compare it with two other fast affine projection (FAP) algorithms. Simplified and multi-input GSFAP versions are also proposed. We show that the algorithm is only marginally more complex than NLMS and simpler than other FAP algorithms. Its application for acoustic echo cancellation is also investigated.

A minimum mean-squared error interpretation of residual ISI channel shortening for discrete multitone transceivers

P.J.W. Melsa et al. (see IEEE Transactions on Communications, vol.44, p.1662-72, 1996) presented a channel shortening technique for discrete multitone transceivers that reduces intersymbol interference (ISI) by forcing the effective channels impulse response to lie within a window of v+1 consecutive samples. G. Arslan et al. (see IEEE International Conference on Acoustic Speech and Signal Processing (ICASSP), p.2965-8, 2000) claim that although this method is intuitive, no previous study has been made on its optimality. They comment on its optimality by simulation. We demonstrate that Melsas approach is in fact theoretically equivalent to a minimum mean-squared error (MMSE) solution to the channel-shortening problem. As a corollary to this we are afforded an insight into MMSE channel shortening as originally proposed by D.D. Falconer and F.R. Magee (see Bell Systems Technical Journal, p.1541-62, 1973). Previously, it has not been intuitive as to why the desired impulse response (DIR) should be made adaptive in this approach. Our result demonstrates that allowing DIR adaptation achieves a minimisation of the effective impulse response energy outside the desired window of v samples.

Minimum mean-squared error impulse response shortening for discrete multitone transceivers

We reformulate the minimum mean squared error (MMSE) solution for time domain equalization (TEQ) in multitone transceivers. We show that the maximum shortening signal-to-noise ratio (MSSNR) and minimum intersymbol interference (min-ISI) TEQs are particular cases of our new formulation. With good channel estimation, the MMSE-TEQ achieves near-optimal rate performance. Our algorithm is less sensitive to channel estimation errors than other methods.

Optimal Wavelet Packet Modulation under finite complexity constraint

Wavelet Packet Modulation (WPM) uses an arbitrary time-frequency plane tiling to create orthogonal subchannels of different bandwidths and symbol rates in a multichannel system. The Wavelet Packet Tree is implemented by iterating a perfect reconstruction two channel transmultiplexer. We derive operating conditions for the capacity-optimal tree for a given communication channel and power budget. We present a fast tree-selection algorithm which achieves this optimum for the case of a finite complexity transceiver. It is found that optimal-WPM outperforms conventional multichannel systems of equal complexity for ISI channels.

A Versatile Variable Rate LDPC Codec Architecture

This paper presents a system architecture for low-density parity-check (LDPC) codes that allows dynamic switching of LDPC codes within the encoder and decoder without hardware modification of these modules. Thus, rate compatibility is facilitated without the performance degradation inherent in a puncture-based system. This versatility also allows the LDPC system to be used in a variety of applications since the encoder and decoder can be used with codes that span a wide range of lengths and rates.

Iterative Channel Estimation, Equalization, and Decoding for Pilot-Symbol Assisted Modulation Over Frequency Selective Fast Fading Channels

A pilot-based channel estimation scheme is proposed for frequency selective Rayleigh fading channels which work in conjunction with the existing paradigm of turbo equalization. The iterative nature of the channel estimation technique provides substantial gain over noniterative methods and makes it a suitable choice for iterative equalization and decoding. The channel estimator has low complexity, is decoupled from the equalizer soft-input soft-output module, and is capable of tracking significant channel variations within a codeword. The scheme is compatible with quadratic-amplitude modulation and with both parallel concatenated convolutional and low-density parity checks coding. The proposed scheme provides an attractive low-complexity alternative to iterative receivers based on state-space models for channel parameter evolution. The use of pilot symbols is demonstrated to aid the equalizer both directly through channel estimation and indirectly through pilot message insertion.

The Gauss-Seidel pseudo affine projection algorithm and its application for echo cancellation

In this paper we propose a new approach for adaptive echo cancellation: the Gauss-Seidel pseudo affine projection algorithm (GSPAP). It is proved by simulations that it is stable, fast convergent and has good tracking abilities. The computational complexity of the GSPAP algorithm is evaluated. It is shown that its simplified and division-less version achieves much improved performances and is only marginally more complex than the NLMS algorithm.

Wideband speech coding using multiple codebooks and glottal pulses

We propose a coder that achieves near transparent wideband speech coding by parameterising the prediction residual through the use of multiple codebooks and synthetic glottal pulses coupled with adaptive bit allocation. The use of synthetic glottal pulses improves the performance of the coder compared to a previous coder using a single impulse without increasing the bit rate. This multiple codebook approach results in a coder operating at 16 kb/s and 24 kb/s that provides comparable speech quality to the CCITT G.722 coder operating at 64 kb/s.

Walsh Hadamard Transform Precoded MB-OFDM: An Improved High Data Rate Ultrawideband System

This paper examines the performance improvement to be achieved by preceding the OFDM symbol by the Walsh Hadamard transform (WHT) in a multiband orthogonal frequency division multiplexing (MB-OFDM) ultrawideband system. An upper bound on the probability of error for WHT-MB-OFDM in the Rayleigh fading channel with convolutional coding is derived, and compared to the upper bound for MB-OFDM. For high rate and low constraint length codes, WHT-MB-OFDM is shown to perform better than MB-OFDM. Next, the effect of narrowband interference is studied, and the WHT-MB-OFDM system is shown to perform significantly better than the MB-OFDM system. Finally, a general amplitude probability distribution (APD) function is derived for linearly precoded MB-OFDM systems and the WHT-MB-OFDM system is shown to cause less interference than the MB-OFDM to the co-existing narrowband system

An edge classification based approach to the post-processing of transform encoded images

Quantisation noise prevalent in transform encoded images becomes increasingly objectionable as the required bit rate is reduced. The perceptual effect of this coding noise is highly dependent on the focal behaviour of the signal upon which it is superimposed. In this paper, a computationally-efficient edge classifier, employing a histogram treatment of image sub-blocks, is proposed. The classifier forms the basis of an adaptive, non-linear postprocessing algorithm incorporating adaptive /spl alpha/-trimmed mean filtering (where the /spl alpha/-value and window size are determined by the output of the edge classifier) and a transform domain dithering technique. Subjective test results confirm the efficacy of the approach.<<ETX>>