Performance modelling for fading channel communication scheme

Markov models are well suited to the modelling of correlated communication channels. However, most conventional techniques include iterative methods of optimising their parameters, or require protracted calibration with field data or simulators for some given specific system. Both are laborious and prone to criticism of invalidity, and the mapping of model to the physical processes is often indirect or obscure. An advantage inherent in the Markov structure of emulating camel memory is therefore lost and provides motivation to a novel methodology of constructing Marico's proposed in this thesis. The resulting model is also novel. By applying Gaussians as components of the Markov model state descriptions, a direct mapping to a principal physical factor, calibration of the model as a once-only operation, and representation from memory less to correlated channels by PDF transformation, all become possible using published and well-proven mathematical techniques. By proposing a topology of a finite state Markov chain, the variance per Gaussian per state represents noise power, or equivalently SNR, and for a single Gaussian PDF per state, supports the concept of alignment of the ideal to the physical as received SNR over a memory less channel for some given modulation scheme via published BER curves. To model time-variant channels, each state PDF is made joint with a conformant random variable generated to suit channel type: tri-variate for Rayleigh; quad- for Ricean. Again, this is achieved using well-proven mathematical methods, and the validity of modelling both memory less and fading channels in this way is shown by close alignment of model output with published reference curves. A filer representation of the fading dynamic is made with the state transition probabilities whose values are calculated based on partitioning of received SNR to state. The methodology is therefore a well defined process with a single fixed calibration without iteration and is shown to be efficient against a popular conventional model perceived as also highly efficient. Performance and further validation of the methodology and resulting model are also shown by comparison with models of similar topology for specified realistic fading scenarios. Output of the proposed model is shown of superior performance by overcoming criticism of the limitations of conventional uni- vaiiate finite state Markov chains.