強化学習による複数のメトリックを考慮したQoSルーティングアルゴリズム

抄録

Recently, there are numerous services, such as audio and video streaming, which are real-time communication services which guarantee bandwidth and low latency, called “QoS (Quality of Service).” To ensure the bandwidth and shorten the communication delay, it is required to solve the problem of finding paths which satisfy the QoS constraints made of two or more metrices, such as link costs and communication delay. However, this problem is shown as an NP-complete problem. We introduce a routing algorithm with a reinforcement learning scheme to obtain the paths which satisfy two or more constraints.