A Collaborative Active Routing Information Exchange Mechanism Based Reinforcement Learning

The packet-switched networks are an important part of today's global communication infrastructure. IP network with other network technology will become the new generation network. Existing routing protocol implementations are monolithic, bundling together a database with an optimal path calculation algorithm and a network state distribution mechanism. Active routing is a new routing strategy target on the above difficulty. This paper presents a collaborative active routing information exchange mechanism based reinforcement learning. The active routing agents get current state information of network attribute, and set up a routing table, realize active routing mechanism.