Openflow channel deployment algorithm for software-defined AFDX

Avionics full duplex switched Ethernet (AFDX) is a time-critical network for aeronautical communications. Based on the static routing configuration of the virtual links (VL), real-time packets can be transmitted before their deadlines through AFDX. The next generation aeronautical communication environment comprises of multiple networks and service providers. Packets transmitted in AFDX may be changed frequently in runtime when sharing resources among different networks. In this situation, dynamic routing configuration is required to change packets' routes dynamically according to switch work loads and packet delays to meet all packets' QoS requirements. Software-defined networking (SDN), which delegates the task of routing from switches to the controller, is a promising approach to achieve dynamic routing configuration. OpenFlow is one of the main protocols for the controller to communicate with switches. Switches have to consult the controller frequently through OpenFlow channels for instructions on how to handle incoming packets. In this paper, we focus on optimizing the OpenFlow channel deployment in software-defined AFDX. Each switch should be deployed an OpenFlow channel which may be composed of a main connection and multiple auxiliary connections. The switch may service auxiliary connections with different priorities. Therefore, different time-critical packets can be sent by the switch to the controller on different priority connections to meet all packets' QoS requirements. This paper proposes an OpenFlow channel deployment algorithm for AFDX based on the principles of genetic simulated annealing (GSA). It tries to find the optimal auxiliary connections deployment that minimizes the number of connections and balance the work loads of each connection. The algorithm guarantees that all realtime packets meet their deadlines. Experimental results show the validity of the algorithm.