This paper considers the downlink scheduling issue in the Third Generation Partnership Project. It applies a genetic algorithm to look for the optimal solution. A fitness function is formulated to include various metrics of quality of service while a solution space is defined. The genetic algorithm is then applied to save the overwhelming computational complexity in the exhaustive search. Through computer simulations, it is shown that this method will lead to superior performance in both fairness and throughput when compared with deterministic algorithms such as round robin and max carrier-to-interference (C/I) schedulers.
It is interesting to learn how the genetic algorithm is applied in this problem. This paper provides a necessary introduction to the system model. Readers who are not in the area of wireless communications may occasionally find it difficult to understand. Overall, however, the paper is quite accessible. The illustrated example is a great help in understanding the big picture of the issue at hand. Although the training process is not discussed at length, the performance is presented clearly and looks promising.
In addition to experts in this field, professionals with a general interest in network scheduling may be inspired by this application of a genetic algorithm.