A Mixed-Critical Consistent Update Algorithm in Software Defined Time-Triggered Ethernet Using Time Window

Software Defined Networking (SDN) presents a tremendous opportunity for developing abstract management of network updates. However, network updates introduce challenges in terms of consistency. Many consistent update algorithms are proposed for Ethernet, but there is seldom update algorithm for distributed time-triggered system to satisfy requirements of strict real-time and mixed-critical. The contribution of this paper is to present a distributed update mechanism and a mixed-critical update algorithm for the time-triggered (TT) and rate-constrained (RC) traffics to reduce the memory space complexity, computational time complexity, update time, and keep per-packet consistent strictly. More specifically, consistent update is defined and some basic update algorithms are described. And then a packet-based network model is built to describe network resources. Subsequently, the update mechanism is changed for the TT and RC traffics. Meanwhile, a mixed-critical update algorithm, which contains a time window-based update and a rate-constrained update for TT and RC traffics respectively, is proposed. Whether the mixed-critical update algorithm is consistent updates or not is proved logically. Finally, experiments analyzed the performance of the mixed-critical update algorithm, compared with four conventional updates. The results show that there is no black hole, loop and inconsistent path for the mixed-critical update algorithm. The most important thing is that the memory space complexity is reduced by 19.93% at least, compared with two-phase update. Meanwhile, the computational time complexity and update time are reduced by 17.14% and 7.96% respectively, compared with time-based update. The mixed-critical update algorithm provides a method to optimize policy update tasks for distributed time-triggered system, so as to improve the ability of system reconfiguration.