Experimental Evaluation of Multiple Multipath Schedulers over Various Urban Mobile Environments
Papa Bara Diakhate, Tran‐Tuan Chu, Mohamed Aymen Labiod, Brice Augustin, Hai Anh Tran, Abdelhamid Mellouk · 2022
The use of multiple interfaces and multipath protocols for end-to-end devices, especially in a mobile environment, has recently gained much attention. At the heart of any multipath protocol lies a scheduler, whose ability to select the best path dictates the system’s overall performance. While a plethora of schedulers has been proposed so far, their evaluation was only conducted in simulated, static environments. A question remains on their behavior in real-world, dynamic, and mobile conditions. In this paper, we design a framework and conduct various experiments to assess the performance of four state-of-the-art schedulers for the cutting-edge MPQUIC multipath protocol. Our setup involves mobile situations within an urban area around Paris, France. We first design and implement an MPQUIC-based modular scheduler assessment framework to compare the schedulers. We then take advantage of this framework to measure their performance in real-world scenarios involving multiple transportation modes (train and car). The collected datasets allow us to perform an in-depth analysis to find insights as well as quantify the schedulers’ performances. Our experimental results confirm the dominance of the state-of-the-art learning-based multipath scheduler.