Multiple objective fitness functions for cognitive radio adaptation.

This thesis explores genetic algorithm and rule-based optimization techniques used by cognitive radios to make operating parameter decisions. Cognitive radios take advantage of intelligent control methods by using sensed information to determine the optimal set of transmission parameters for a given situation. We have chosen to explore and compare two control methods. A biologically-inspired genetic algorithm (GA) and a rule-based expert system are proposed, analyzed and tested using simulations. We define a common set of eight transmission parameters and six environment parameters used by cognitive radios, and develop a set of preliminary fitness functions that encompass the relationships between a small set of these input and output parameters. Five primary communication objectives are also defined and used in conjunction with the fitness functions to direct the cognitive radio to a solution. These fitness functions are used to implement the two cognitive control methods selected. The hardware resources needed to practically implement each technique are studied. It is observed, through simulations, that several trade offs exist between both the accuracy and speed of the final decision and the size of the parameter sets used to determine the decision. Sensitivity analysis is done on each parameter in order to determine the impact on the decision making process each parameter has on the cognitive engine. This analysis quantifies the usefulness of each parameter.