Sub-Channel and Power Allocation for Multiuser OFDM with Rate Constraints using Genetic Algorithm

We demonstrate that the resource allocation problem in OFDM (for which there are no complete analytical solutions or numerical solutions that are practical) can be solved in real time using the Genetic Algorithm (GA). The sub-channel assignment and power allocation that maximize the throughput of the system with constraints on total power usage and users' transmission rates are obtained using an intelligent search based on GA. Our version of the GA uses two chromosomes per individual - one for the channel assignment and another for the power allocation. Users' transmission rate constraints are met by awarding points to individuals who satisfy the constraints and incorporating the points into the fitness function. There is no analytical method that produces the global optimum solution to the problem on its complete form with the constraints mentioned above for us to compare our result with. However, by comparing our solutions to the existing global optimum solutions for the cases with less constraints, we show that our algorithms produce results that are within 5% of the global optimum.