Two methods of adaptive controlled channel resource allocation using reinforcement learning and supervised learning techniques

摘要

Two methods of dynamic channel allocation for a cellular telephone network using neural networks are investigated. Both methods continuously optimize the mobile network based on changes in calling traffic. The first method uses backpropagation (supervised learning) model predictions to aid the channel allocator. The second method uses the same backpropagation models along with actor-critic (reinforcement learning) models to perform the channel allocation. A comparison shows that both methods significantly outperform fixed channel allocation, even when the call traffic activity deviates from the previously learned models of the call traffic activity.