# Network-in-the-Loop at Scale: GPU-Batched 5G Simulation for Massively Parallel Robot Learning

arXiv: https://arxiv.org/abs/2610.02370 · HTML: https://mlsanity.com/paper/2610.02370/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 25 of its 55 references that are in the map.
- First posted on arXiv 1 Oct 2026.

## Builds on (most cited first, 25 of 25)

- [Isaac Gym: High Performance GPU-Based Physics Simulation For Robot Learning](https://mlsanity.com/paper/2108.10470/) (2021)
- [Learning to Communicate with Deep Multi-Agent Reinforcement Learning](https://mlsanity.com/paper/1605.06676/) (2016)
- [Learning Multiagent Communication with Backpropagation](https://mlsanity.com/paper/1605.07736/) (2016)
- [TarMAC: Targeted Multi-Agent Communication](https://mlsanity.com/paper/1810.11187/) (2018)
- [Learning when to Communicate at Scale in Multiagent Cooperative and Competitive Tasks](https://mlsanity.com/paper/1812.09755/) (2018)
- [Learning to Schedule Communication in Multi-agent Reinforcement Learning](https://mlsanity.com/paper/1902.01554/) (2019)
- [VMAS: A Vectorized Multi-Agent Simulator for Collective Robot Learning](https://mlsanity.com/paper/2207.03530/) (2022)
- [Sionna RT: Differentiable Ray Tracing for Radio Propagation Modeling](https://mlsanity.com/paper/2303.11103/) (2023)
- [ColO-RAN: Developing Machine Learning-based xApps for Open RAN Closed-loop Control on Programmable Experimental Platforms](https://mlsanity.com/paper/2112.09559/) (2021)
- [srsLTE: An Open-Source Platform for LTE Evolution and Experimentation](https://mlsanity.com/paper/1602.04629/) (2016)
- [New Radio Physical Layer Abstraction for System-Level Simulations of 5G Networks](https://mlsanity.com/paper/2001.10309/) (2020)
- [Calibration of the 5G-LENA System Level Simulator in 3GPP reference scenarios](https://mlsanity.com/paper/2205.03278/) (2022)
- [An E2E Simulator for 5G NR Networks](https://mlsanity.com/paper/1911.05534/) (2019)
- [ROS-NetSim: A Framework for the Integration of Robotic and Network Simulators](https://mlsanity.com/paper/2101.10113/) (2021)
- [SynchroSim: An Integrated Co-simulation Middleware for Heterogeneous Multi-robot System](https://mlsanity.com/paper/2208.06569/) (2022)
- [CAVIAR: Co-simulation of 6G Communications, 3D Scenarios and AI for Digital Twins](https://mlsanity.com/paper/2401.03310/) (2024)
- [Network Digital Twin for 5G-Enabled Mobile Robots](https://mlsanity.com/paper/2502.02253/) (2025)
- [Open Wireless Digital Twin: End-to-End 5G Mobility Emulation with OpenAirInterface and Ray Tracing](https://mlsanity.com/paper/2503.12177/) (2025)
- [CORNET 2.0: A Co-Simulation Middleware for Robot Networks](https://mlsanity.com/paper/2109.06979/) (2021)
- [Learning To Communicate Over An Unknown Shared Network](https://mlsanity.com/paper/2507.06499/) (2025)
- [CALF: Communication-Aware Learning Framework for Distributed Reinforcement Learning](https://mlsanity.com/paper/2603.12543/) (2026)
- [AgentComm-Bench: Stress-Testing Cooperative Embodied AI Under Latency, Packet Loss, and Bandwidth Collapse](https://mlsanity.com/paper/2603.20285/) (2026)
- [SimART: A Unified and Open Real-world Multimodal Simulation Platform for 6G Integrated Sensing and Communication](https://mlsanity.com/paper/2605.13309/) (2026)
- [LUCID: An Agentic AI Framework on Digital-Twin in the Loop for QoS-Guaranteeing Robotic Control](https://mlsanity.com/paper/2608.28437/) (2026)
- [Coupled Control and Wireless World Models for Resilient Remote Robotic Control](https://mlsanity.com/paper/2609.04851/) (2026)

## Abstract (arXiv)

Massively parallel GPU simulators train multi-robot policies in thousands of environments, and many fleets use private Fifth-Generation (5G) networks, where each robot's delay depends on its teammates' traffic. Network-in-the-loop training places a simulated 5G network inside this loop. However, GPU robot simulators reduce the network to an independent delay per message, while packet-level simulators run one scenario per CPU process and cannot keep pace with thousands of parallel environments. To bridge this gap, we present Isaac-Net, a GPU-batched 5G New Radio (NR) module that advances the uplink of thousands of environments in lockstep with Isaac Lab physics. Isaac-Net simulates every slot, the 0.5~ms interval in which the base station decides which robots transmit, for all environments at once. Extensive experiments confirm that its NR engine reproduces the median delay of ns-3 5G-LENA across loads, with a median delay 5--10% low on an unseen carrier and 9% high at 32 robots per environment in closed loop. The engine also reproduces the Age of Information (AoI), the age of each robot's newest delivered report, while an independent delay per message leaves the AoI tail about three times too light. In a configuration validated against 5G-LENA, Isaac-Net keeps the network in the loop for about one million robots on one GPU at 83% of the Isaac Lab rate without the network, measured under a random policy. Isaac-Net is open source at https://github.com/ZzZTripleZzZ/isaac-net