Peak-to- Average Power Ratio Reduction, Impulse Noise Mitigation and Synchronisation Effects of OFDM Systems

Abstract

In recent years many wired and wireless communication systems have used orthogonal frequency division multiplexing (OFDM) technology. OFDM in combination with multipleinput multiple-output (MIMO) techniques is the leading candidate for implementing fourth generation communication systems. Three major drawbacks of OFDM are its high peak-to-average power ratio (PAPR), its sensitivity to synchronisation errors and the performance degradation in the presence of impulse noise. The original work presented in this thesis is a study of these three key aspects in OFDM and MIMO-OFDM systems, and an investigation of the performance and the capacity distribution of MIMO-OFDM systems.The PAPR problem of the time domain OFDM signals is studied. Clipping, a simple and effective method to reduce the high PAPR, is applied to the baseband OFDM signals. It is shown that, while clipping eliminates the high PAPR, it introduces some peak regrowth at subsequent filtering stages. To reduce this peak regrowth, clip and discrete Fourier transform based filtering technique is modified to include spectral masking. The performance of this technique is studied for a Hiperlan/ 2 system and it is shown that the peak regrowth due to filtering stages is negligible. The clipping effects are also analysed for an Alamouti OFDM system. In Alamouti OFDM systems, the effective clipping noise significantly degrades the system performance. This is due to the Alamouti combining ofthe signals at the receiver. Hence, it is shown that Alamouti OFDM systems are more sensitive to clipping noise than OFDM systems.The performance degradation in OFDM systems due to impulse noise is also investigated. The Noise Bucket concept is invoked to explain this performance degradation. A decision directed impulse noise mitigation technique is analysed. In this technique, initial tentative decisions are made about the transmitted data. A threshold detection algorithm is used to estimate the time domain impulses and these are then subtracted before the final demodulation. Simulation results indicate several orders of performance improvement.The performance degradation in OFDM systems due to phase noise and carrier frequency offset (CFO) is then studied. To analyse this degradation, a theoretical approach is devised. This approach can more accurately predict error rates than existing methods. In addition, the work is extended to study the CFO effects in Alamouti space-time systems, space-frequency OFDM systems and OFDM-based V-BLAST systems. It is shown that up to a normalised CFO of 0.05 Alamouti space-frequency OFDM systems are more sensitive to CFO errors than Alamouti space-time OFDM systems. In addition, both systems exhibit a higher sensitivity to CFO errors than OFDM systems. In OFDM-based V-BLAST, the effect of layer decoding errors due to CFO is studied and it is shown that there is a significant performance degradation. Next, the performance of OFDM-based V-BLAST systems is studied and a modified precoding scheme is proposed. This precoding scheme improves the performance of OFDM-based V-BLAST systems by exploiting the frequency diversity. The performance degradation of Alamouti OFDM systems over rapidly time varying channels is also investigated. Since OFDM symbols have long periods, the assumption that the channel must remain constant over the Alamouti codeword period is relaxed. The resulting performance degradation of two receivers is examined and compared with the performance of a conventional combiner. It is shown that, in most cases, a decision directed detector can improve the performance. However, this detector increases the receiver complexity.The final topic which is studied in this thesis is the capacity of MIMO-OFDM systems. In rich scattering and in semi-correlated channels, it is shown that the MIMO-OFDM capacity distribution can be accurately modelled as a Gaussian random variable. Using a recently reported approximation to the logarithmic function, the variance of approximate narrowband Mlf\IO capacity is also computed. The difference between this approximate capacity variance and the exact MIMO capacity variance is large, even larger than the difference between the approximate MIMO mean capacity and the exact MIMO mean capacity.This research provides a detailed understanding of the various factors that influence the performance degradation in OFDM and MIMO-OFDM systems. In addition, it proposes several mitigation techniques. These mitigation techniques and performance enhancements can be used to implement efficiently future high speed communication systems.