Comparison of Optimization Goals for Resilient Routing

Abstract

With the introduction of 5G, flexible network management through Software-Defined Networking will become available in cellular networks. As this has not been designed for safety-critical application such as autonomous driving or aeronautical applications, special care needs to be taken to use it in such contexts. In this work, different objective functions for optimization and their respective impact on performance characteristics such as link utilization and latency are investigated and compared. This allows to better understand the impact of resilient flow placement under constrained capacity and latency on network performance. Results show that for critical systems, objective functions relating to link capacity result in poorer network performance as latency and resilience are more limiting factors. While optimizations for maximum resilience increase network usage, with tight latency constraints they are comparable in performance with link capacity optimization, while providing more resilience against failures.