NetworkGym: Reinforcement Learning Environments for Multi-Access Traffic Management in Network Simulation

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Haider, Momin, Yin, Ming, Zhang, Menglei, Gupta, Arpit, Zhu, Jing, Wang, Yu-Xiang
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866916470836953088
author Haider, Momin
Yin, Ming
Zhang, Menglei
Gupta, Arpit
Zhu, Jing
Wang, Yu-Xiang
author_facet Haider, Momin
Yin, Ming
Zhang, Menglei
Gupta, Arpit
Zhu, Jing
Wang, Yu-Xiang
contents Mobile devices such as smartphones, laptops, and tablets can often connect to multiple access networks (e.g., Wi-Fi, LTE, and 5G) simultaneously. Recent advancements facilitate seamless integration of these connections below the transport layer, enhancing the experience for apps that lack inherent multi-path support. This optimization hinges on dynamically determining the traffic distribution across networks for each device, a process referred to as \textit{multi-access traffic splitting}. This paper introduces \textit{NetworkGym}, a high-fidelity network environment simulator that facilitates generating multiple network traffic flows and multi-access traffic splitting. This simulator facilitates training and evaluating different RL-based solutions for the multi-access traffic splitting problem. Our initial explorations demonstrate that the majority of existing state-of-the-art offline RL algorithms (e.g. CQL) fail to outperform certain hand-crafted heuristic policies on average. This illustrates the urgent need to evaluate offline RL algorithms against a broader range of benchmarks, rather than relying solely on popular ones such as D4RL. We also propose an extension to the TD3+BC algorithm, named Pessimistic TD3 (PTD3), and demonstrate that it outperforms many state-of-the-art offline RL algorithms. PTD3's behavioral constraint mechanism, which relies on value-function pessimism, is theoretically motivated and relatively simple to implement.
format Preprint
id arxiv_https___arxiv_org_abs_2411_04138
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle NetworkGym: Reinforcement Learning Environments for Multi-Access Traffic Management in Network Simulation
Haider, Momin
Yin, Ming
Zhang, Menglei
Gupta, Arpit
Zhu, Jing
Wang, Yu-Xiang
Networking and Internet Architecture
Artificial Intelligence
Machine Learning
Mobile devices such as smartphones, laptops, and tablets can often connect to multiple access networks (e.g., Wi-Fi, LTE, and 5G) simultaneously. Recent advancements facilitate seamless integration of these connections below the transport layer, enhancing the experience for apps that lack inherent multi-path support. This optimization hinges on dynamically determining the traffic distribution across networks for each device, a process referred to as \textit{multi-access traffic splitting}. This paper introduces \textit{NetworkGym}, a high-fidelity network environment simulator that facilitates generating multiple network traffic flows and multi-access traffic splitting. This simulator facilitates training and evaluating different RL-based solutions for the multi-access traffic splitting problem. Our initial explorations demonstrate that the majority of existing state-of-the-art offline RL algorithms (e.g. CQL) fail to outperform certain hand-crafted heuristic policies on average. This illustrates the urgent need to evaluate offline RL algorithms against a broader range of benchmarks, rather than relying solely on popular ones such as D4RL. We also propose an extension to the TD3+BC algorithm, named Pessimistic TD3 (PTD3), and demonstrate that it outperforms many state-of-the-art offline RL algorithms. PTD3's behavioral constraint mechanism, which relies on value-function pessimism, is theoretically motivated and relatively simple to implement.
title NetworkGym: Reinforcement Learning Environments for Multi-Access Traffic Management in Network Simulation
topic Networking and Internet Architecture
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2411.04138