Nguyen Thanh Trung

With the development and commercialization of new mobile network generations such as 5G and beyond, future communications are shifting from the traditional single-path paradigm to multipath transport protocols such as MPTCP and MPQUIC. One of the most critical issues in dealing with the multipath transmission is appropriately scheduling the pathways in order to guarantee QoS. Despite the fact that tremendous effort has been put into developing multipath scheduling algorithms, existing approaches suffer from several limitations when dealing with the network’s dynamicity, including congestion and packet loss. In this paper, we propose a novel Reinforcement learning-based multipath transport protocol named SATO, which efficiently schedules multipath communication in heterogeneous wireless networks. By leveraging the self-learning ability of reinforcement learning, a node equipped with SATO can capture the environmental changes and select transmission paths based on an appropriate policy to optimize QoS. Our evaluation results show that SATO improves the QoS by 10%-15% in simulation and 12% in a real deployment compared to the state-of-the-art algorithm.

In the 5G and beyond era, multipath transport protocols, including MPQUIC, are necessary in various use cases. In MPQUIC, one of the most critical issues is efficiently scheduling the upcoming transmission packets on several paths considering path dynamicity. To this end, this paper introduces FQ-SAT - a novel Fuzzy Q-learning-based MPQUIC scheduler for data transmission optimization, including download time, in heterogeneous wireless networks. Different from previous works, FQ-SAT combines Q-learning and Fuzzy logic in an MPQUIC scheduler to determine optimal transmission on heterogeneous paths. FQ-SAT leverages the self-learning ability of reinforcement learning (i.e., in a Q-learning model) to deal with heterogeneity. Moreover, FQ-SAT facilitates Fuzzy logic to dynamically adjust the proposed Q-learning model’s hyper-parameters along with the networks’ rapid changes. We evaluate FQ-SAT extensively in various scenarios in both simulated and actual networks. The results show that FQ-SAT reduces the single-file download time by 3.2%–13.5% in simulation and by 4.1%–13.8% in actual network, reduces the download time of all resources up to 20.4% in web browsing evaluation, and reaches percentage of on-time segments up to 97.5% in video streaming, compared to state-of-the-art MPQUIC schedulers.