Multiobjective Path Optimization for Deadline-aware Multipath over SCION

The growing complexity of modern network applications demands flexible and timely data delivery. Time-sensitive applications, such as teleoperation, video conferencing, and online gaming, rely on meeting strict deadlines for data delivery. In such scenarios, data that fails to meet a deadline becomes outdated and useless. One of the potential solutions for this problem is the usage of multipath networking. Nevertheless, this presents a potential challenge: many existing multipath protocols prioritize reliability over latency requirements, making them less suitable for time-sensitive applications. In this paper, we present a path selection approach based on a genetic algorithm that, combined with the SCION Internet architecture, is designed to address the unique requirements of deadline-aware multipath communication. Our algorithm treats path selection as a multiobjective optimization problem and takes into account the need to meet deadlines, minimize cost, and reduce loss rates. We evaluate the algorithm’s performance by testing it in a SCION network and comparing it with a heuristics-based approach. Other evaluation metrics include optimization time, complexity, CPU usage, and memory usage. The results show that our algorithm presents a promising approach to deadline-aware multipath communication in SCION and has the potential to address the challenges that real-life time-sensitive applications pose.