Dynamic Voting Schemes to Enhance Evolutionary Repair in Reconfigurable Logic Devices

The area of fault-handling in reconfigurable logic devices is one that continues to receive research attention in the field of engineering. Field Programmable Gate Arrays (FPGAs) are reconfigurable logic devices that have become an essential element in electronic hardware used for space applications, for instance deep space satellites. When electronic devices such as FPGAs are launched into space, they are relentlessly exposed to fault-inducing hazards such as high levels of radiation and extreme temperatures. The ability of the device to maintain and correct its functionality while experiencing these harsh conditions is vital to a successful mission by today\u27s technological standards. Many techniques have been proposed for the purpose of detecting and repairing hardware faults that occur in reconfigurable logic devices. The implementation of a Genetic Algorithm (GA) as the means of repairing a faulty component has become a popular method among such techniques. A great deal of success has been demonstrated by the use of GAs in fault-repair, but there is room for improvement in the completeness of a given repair. This thesis addresses this issue by exploring the possible outcomes of implementing a voting system to work in parallel with a particular GA. Throughout the first two chapters, a general overview ofFPGAs and faulthandling techniques is provided. The advantages and disadvantages of each technique are mentioned to help re-emphasize the main purpose for the research being conducted. Once a solid background has been established regarding the main ideas behind this work, the thesis presents an in-depth description of the problem and the experimental approach that is taken. The work involves experiments which are run using a simulated FPGA that is coded in C++. A genetic algorithm is included in the program in order to simulate the repair process. By varying the parameters of the GA, as well as experimenting with the addition of a voting scheme to enhance the performance, meaningful results are discovered and presented. Fault-handling techniques proposed in the future will have a better idea of whether or not it would be beneficial to include a voting scheme to improve success