PERFORMANCE STUDY OF RECEIVERS FOR DIGITAL COMMUNICATION OVER FAST FADING CHANNELS

Abstract

Receiver design for signals over channels with fast nonselective Rayleigh fading is considered here. Past research has focused mainly on the case of slow fading where the fading process can be considered to be constant over the duration of a symbol interval. Until now, not much work has dealt with receiver design and analysis of performance over fast fading channels. This thesis attempts to look at this issue and presents two receivers for the purpose of coherently detecting MPSK signals over the AWGN channel with fast nonselective Rayleigh fading.

The conventional method to combat the fading problem is to make use of the phase-locked loop to perform carrier recovery, which is necessary for coherent detection. However due to its limitations, channel estimation will be used and the generated estimates of the quadrature amplitudes of the channel fading gain will be employed in the likelihood ratio test. Either the Kalman filter or the adaptive filter can be employed to track the rapidly fluctuating fading process. Unlike the Kalman filter, the adaptive filter does not require any a priori knowledge of the channel parameters. Also, both the first-order and third-order Butterworth fading spectra are investigated.

In a fast fading channel, the transmitted pulse shape is distorted in that it is multiplied by a rapidly varying complex gain even within the symbol interval. Therefore, the two receivers proposed here, namely the sampled matched filter (SMF) receiver and the optimum sampled (OS) receiver, will have to use multiple samples of the received signal within each symbol interval. By taking multiple samples within each symbol interval, the fading gain can be approximated as being constant within each sampling interval. This forms the basis of the two new receivers.

From simulation results, it is found that by using pure MPSK signals, both the SMF receiver and the OS receiver can provide up to 3 dB gain over the conventional differential detector. However due to decision errors, these receivers eventually perform worse than the differential detector at very high SNR. In order to achieve better performance than MDPSK at high SNR, we must combine coherent detection with differential encoding and differential decoding of MPSK signals for both the two proposed receivers. By doing so, both the receivers outperform the conventional differential detector, especially at high SNR where the error rate floor is reduced significantly. This error rate floor can be reduced by an even larger factor if a larger number of samples per symbol interval is used. The performance results obtained using simulations also indicate that the OS receiver provides good performance gain in terms of error probability over the SMF receiver. This is because the SMF receiver only produces an estimate for the fading gain within each symbol interval. Whereas, the OS receiver utilises more than one channel estimate within each symbol interval. To be more exact, the OS receiver actually generates one channel estimate within each sampling interval, not each symbol interval. Therefore, it outperforms the SMF receiver as expected.

To summarise, the two receivers proposed in this thesis perform well under the fast nonselective Rayleigh fading channel. Since signal processing techniques are used, a complete digital implementation of these two receivers is possible. Moreover, the new finding in this thesis should be of great interest to many mobile communication industries and users.