Intelligent Routing Scheme for SDN Satellite Network Based on Neural Network

In view of the existing SDN-based satellite network system, the tri-state content-addressable memory (TCAM) space occupied by the flow table is increasing, and the search and matching of flow entries are complicated, resulting in reduced routing and forwarding efficiency, which cannot meet the needs of diverse application requirements. Therefore, an intelligent routing scheme of SDN satellite network based on neural network (IRSSN) is proposed. The controller acquires the transmission mode of the data stream by training the neural network (NN), and replaces the flow table with the trained NN. An intelligent routing mechanism based on the orthogonal polynomial (Chebyshev) neural network is proposed. The switch predicts its forwarding path according to the traffic type of the data stream to meet the QoS requirements of the satellite network application. The simulation results show that IRSSN significantly reduces the occupied TCAM space and improves routing efficiency.