Redesigning CAVIAR: a framework for next-Generation 6G/B6G simulations

Abstract

The evolution of mobile networks toward the sixth generation (6G) and the consolidation of the Open Radio Access Network (Open RAN) require flexible, modular simulation tools capable of integrating communications, mobility, and artificial intelligence in ultra-realistic scenarios. This work presents the latest version of the Communication Networks, Artificial Intelligence and Computer Vision with 3D Computer-generated Imagery (CAVIAR) framework, a co-simulation framework aligned with the 3rd Generation Partnership Project (3GPP) guidelines, capable of integrating independent modules for three-dimensional (3D) simulation, mobility, communications, and artificial intelligence, orchestrated asynchronously for real-time execution. CAVIAR incorporates channel modeling via ray tracing using Sionna-Ray Tracing (Sionna-RT) and Wireless InSite, fifth-generation (5G) network simulation with Network Simulator 3 (ns-3) and the 5G-Lena module, propagation calculation acceleration techniques, and interactive monitoring through InfluxDB and Grafana. To validate the framework, three representative use cases aligned with studies described in the 3GPP Release 19 specifications were implemented: rural animal monitoring with dynamic Physical Resource Block (PRB) allocation via Open RAN, urban search and rescue with You Only Look Once (YOLO)-based human detection over Non-Terrestrial Network (NTN) links, and (iii) indoor robot coordination with multimodal Integrated Sensing and Communication (ISAC) combining Light Detection and Ranging (LiDAR) and Red, Green, Blue (RGB) imaging. The results show that CAVIAR achieves Key Performance Indicators (KPIs) with high accuracy and low computational resource usage even in complex scenarios, enabling latency-sensitive and high-reliability network applications, thus reinforcing its applicability for advanced research in 5G-Advanced and 6G.