Diversity: On the Evolution to MIMO and Beyond

This paper considers the evolution to multiple input, multiple output (MIMO) communications, which has its origins in both spatial processing and temporal processing. In the context of this paper, the term ´spatial processing´ is taken to include classical/adaptive beamforming and sub-space analysis, whereas the term ´temporal processing´ is taken to include the subjects of error control coding, interleaving and equalisation. The objective of this paper is therefore to provide an outline of some of the principal developments within both spatial processing and temporal processing, in order to relate MIMO to the prior art. The emphasis on this paper is on the results which steered this notion of diversity to yield MIMO. The benefits of MIMO have been achieved by integrating the functionality contained within both spatial processing and temporal processing. As shown within this paper, historically, the original motivation within spatial processing was to enhance target location and accuracy in terms of resolution and bias. Temporal processing, in the case of communications engineering, however, was driven by the need to control the number of digital errors between the transmitter and receiver. Each of these two subjects was for a significant number of years treated independently, and excellent process had been made in both. However, a significant barrier within each of these two subjects was the limitation of system processing capability, particularly in terms of both speed and memory. Improvements in the categorisation of physical phenomena associated with a range of wireless channels, and the processing power of recent hardware have enabled the realisation of MIMO to be achieved. As hardware capabilities improve and greater effort is made at the integration of the functionality between the various layers within a communications system, system efficiency and performance can be maximised. The technologies which will be instrumental in achieving this goal are likely- to include techniques as diverse as artificial intelligence, higher order statistics and the synthesis of natural phenomena (e.g., the use of chaotic models)