A new transceiver for data transmission over multipath fading channels employing precoding and differential detection is investigated. It comprises a precoder, a differential detector and a linear equalizer compatible with differentially coherent detection. During a startup phase, the channel is estimated from a known training sequence and subsequently relayed to the precoder at the transmitter. This combination effectively functions as a decision feedback equalizer (DFE) for differentially coherent demodulation. Simulation results using the LMS algorithm are presented for both differential BPSK and differential QPSK over a two path fading channel. The proposed system is able to equalize the fading channel with a performance very close to that of the conventional DFE for coherent demodulation.

A nonlinear decision-based adaptive equalizer compatible with differentially coherent phase shift keying (PSK) is proposed for frequency-selective fading channels. This equalization scheme is appropriate whenever conventional equalizers are not capable of tracking phase variations in selective fading channels. The received signal is first converted to a baseband signal and then sent through a differential detector. A nonlinear processor before the equalizer generates the needed nonlinear terms that are weighted and summed in the equalizer. Nonlinear intersymbol interference at the output of the differential detector is dealt with by minimizing an error signal between the output of the equalizer and the detected data. The adaptation algorithm can be any algorithm currently used for conventional equalizers. Our simulation results confirm that for channels with spectral nulls, equalization is achieved successfully with the proposed scheme, whereas, linear equalizers, either with coherent or noncoherent detection, fail.

We employ a fractionally spaced decision feedback equalizer for indoor fading channels and give the relevant analytical and simulated performance results in terms of Doppler frequency. An upper limit is determined for the channel variations which can be tracked by the equalizer. Also, the effect of error propagation on the performance is evaluated. Finally, a nonlinear decision-based equalizer compatible with differentially coherent PSK is proposed. This equalization scheme is appropriate whenever conventional equalizers are not able to track channel phase variations. Our simulation results indicate that, for channels with spectral nulls, equalization will be achieved successfully with the proposed scheme, whereas linear equalizers, either with coherent or non-coherent detection, fail.

An efficient ranked-order filter is proposed for removing the additive noise from colour images. In the approach presented here, vector ranking reduces to ordering vectors according to their distance to a centroid criterion, such as the sample mean. This technique significantly reduces the computational burden. Using this vector ranking scheme, α-trimmed mean filtering is considered. It can suppress both impulsive and Gaussian noise better than the median filter. A significant reduction in the complexity compared to the regular vector α-trimmed mean filtering is shown. A number of simulations were performed in order to quantitatively evaluate its performance. These simulations involve the estimation of colour images in additive short tailed and long tailed noise