This thesis details an investigation into the application of genetic algorithms to indoor optical wireless communication systems. The principle aims are to show how it is possible for a genetic algorithm to control the received power distribution within multiple dynamic environments, such that a single receiver design can be employed lowering system costs. This kind of approach is not typical within the research currently being undertaken, where normally, the emphasis on system performance has always been linked with improvements to the receiver design. Within this thesis, a custom built simulator has been developed with the ability to determine the channel characteristics at all locations with the system deployment environment, for multiple configurations including user movement and user alignment variability. Based on these results an investigation began into the structure of the genetic algorithm, testing 192 different ones in total. After evaluation of each one of the algorithms and their performance merits, 2 genetic algorithms remained and are proposed for use. These 2 algorithms were shown capable of reducing the receiver power deviation by up to 26%, and forming, whilst the user perturbs the channel, through movement and variable alignment, a consistent power distribution to within 12% of the optimised case. The final part of the work, extends the use of the genetic algorithm to not only try to optimise the received power deviation, but also the received signal to noise ratio deviation. It was shown that the genetic algorithm is capable of reducing the deviation by around 12% in an empty environment and maintain this optimised case to within 10% when the user perturbs the channel.
展开▼
