Application of adaptive equalisation to microwave digital radio
M.C.S. Young · OpenGrey (Institut de l'Information Scientifique et Technique) · 1989
Fading due to multipath propagation remains one of the principle sources of signal loss or outage in microwave digital radio communications, introducing amplitude and phase distortion to the received signal. This thesis examines a number of adaptive equalisation techniques with the aim of reducing degradation due to fading. The chapters may be categorised into an examination of system performance, followed by details of practical equaliser implementation. The requirement for bandwidth efficient modulation techniques has placed greater demands on the equalisation process, and more advanced structures are continually required. Two main types of adaptive equaliser are considered, linear transversal equalisers and nonlinear decision feedback equalisers. Linear equalisers are currently available in certain commercial digital radio systems, however results in chapter 3 indicate that during severe fading conditions, linear equalisation provides insufficient compensation, and system outage is likely. A number of suggestions are made for enhancing equaliser performance during deep fading. Decision feedback equalisers offer considerable performance gain over linear equalisers, and one configuration is shown to have the capability of maintaining performance through a phase transition. Transitions between different phase types remain one of the fundamental problems for reliable microwave communications. To ensure a stable timing phase, a novel receiver structure is proposed, whereby timing information is obtained from an estimate of the received channel impulse response. The proposed structure is shown to be capable of dealing with the most severe fading conditions described by the simplified three-path channel model. Equaliser implementation is the second main area of research. Recognising the relatively straightforward implementation of linear equaliser designs, results are presented for this structure. Similar techniques and analysis may be applied to other equaliser types. The least mean squares (LMS) algorithm is commonly used to adaptively update the equaliser tap weights, while the slightly less complex zero-forcing (ZF) algorithm has found favour for certain high speed designs. Chapter 5 provides a comparative study of various adaptive algorithms suitable for high speed implementation, including block least mean squares (BLMS) algorithms. The BLMS algorithm is shown to be a practical alternative when arithmetic resolution is limited by high speed digital operation. Finally, a novel 'off-line' adaptation technique is proposed where adaptation is performed by a digital signal processor. Results are presented from the implemented system.