Scaling Real-Time Traffic Analytics on Edge-Cloud Fabrics for City-Scale Camera Networks

2026-03-05

AI Summary

This work addresses the challenges of real-time, city-scale traffic analysis under constraints of latency, bandwidth, and computational resources when processing high-concurrency video streams. It proposes an edge-cloud协同 architecture that integrates a DNN inference pipeline, spatiotemporal graph neural networks (ST-GNNs), and Jetson Orin edge accelerators to dynamically construct traffic graphs from multi-stream video inputs and enable short-term traffic forecasting. The system introduces a capacity-aware scheduling mechanism, SAM3-assisted annotation, and a continual federated learning framework to support online model evolution and elastic scalability. Evaluation on a real-world testbed demonstrates stable throughput of 2000 FPS at the edge, low-latency aggregation, and accurate, scalable traffic prediction across up to 1000 concurrent video streams.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Computer Vision: Video Understanding & Activity Analysis

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Cloud, edge and content delivery systems for the Web
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines

Abstract

Real-time city-scale traffic analytics requires processing 100s-1000s of CCTV streams under strict latency, bandwidth, and compute limits. We present a scalable AI-driven Intelligent Transportation System (AIITS) designed to address multi-dimensional scaling on an edge-cloud fabric. Our platform transforms live multi-camera video feeds into a dynamic traffic graph through a DNN inferencing pipeline, complemented by real-time nowcasting and short-horizon forecasting using Spatio-Temporal GNNs. Using a testbed to validate in a Bengaluru neighborhood, we ingest 100+ RTSP feeds from Raspberry Pis, while Jetson Orin edge accelerators perform high-throughput detection and tracking, producing lightweight flow summaries for cloud-based GNN inference. A capacity-aware scheduler orchestrates load-balancing across heterogeneous devices to sustain real-time performance as stream counts increase. To ensure continuous adaptation, we integrate SAM3 foundation-model assisted labeling and Continuous Federated Learning to update DNN detectors on the edge. Experiments show stable ingestion up to 2000 FPS on Jetson Orins, low-latency aggregation, and accurate and scalable ST-GNN forecasts for up to 1000 streams. A planned live demonstration will scale the full pipeline to 1000 streams, showcasing practical, cross-fabric scalability.

Problem

• real-time traffic analytics
• edge-cloud fabric
• city-scale camera networks
• latency constraints
• scalability

Innovation

• Edge-Cloud Fabric
• Spatio-Temporal GNN
• Real-Time Traffic Analytics
• Continuous Federated Learning
• Capacity-Aware Scheduling