Using AI for optimal Time Sensitive Networking in Avionics

Background: The communication architecture within an aircraft needs to support high speed data requirements with reduced wiring weight. With Time Sensitive Networking, we can have multiple gateways and interfaces for each subsystem, which substantially reduces point to point communication. Methods: In this study, we have used a System Level Modelling Tool called VisualSim Architect for evaluating various use cases and validating the results. Results: By using Ethernet as the backbone helped in providing very high bandwidth for critical and non-critical applications. By applying Time Sensitive Networking protocols, a deterministic behavior for time critical applications were observed. Further, it was found that by adding AI algorithms to shape the traffic and time slots increased the reliability. Conclusion: From the results of various use cases, it was seen that even with Time Sensitive Networking, few of the critical applications misses their time slot and as a result a spike in latency plots were observed. In order to prevent slot misses, dynamic time slot configuration as well as credit configuration with the help of AI algorithms were considered and the results obtained helped validate this statement.