Edge Computing Node

Definition and Technical Positioning

An edge computing node is a physical device deployed at the network edge in close proximity to data sources, equipped with data acquisition, real-time processing, local storage, and communication capabilities. Within the vehicle-road-cloud integrated architecture, it serves as the "local computing hub" for the roadside perception system—fusing, analysing, and making decisions locally on vast volumes of sensor data that would otherwise require cloud upload, before distributing structured results to passing vehicles via roadside units (RSUs) with millisecond-level latency.

Roadside sensing equipment (cameras, millimetre-wave radars, LiDAR, etc.) continuously gathers raw data. If all this data were backhauled to the cloud, it would trigger bandwidth congestion and transmission delays ranging from tens to hundreds of milliseconds. Edge computing nodes transform data flow through a "proximity computing" model: terminal data is first routed to the nearest edge node, which performs real-time analysis, filtering, and pre-processing, uploading only critical data to the cloud. This ensures that even if a vehicle's line of sight is obstructed by heavy commercial vehicles, it can still receive an early warning—such as "Pedestrian ahead, emergency braking advised"—once the edge node completes perception fusion, marking a crucial shift from "passively seeing" to "actively knowing".

Core Functions and Performance Indicators

Edge computing nodes undertake several critical roles in vehicle-road coordination: through multi-sensor data fusion, heterogeneous data from roadside cameras, millimetre-wave radars, and LiDAR are aligned and fused under a unified spatio-temporal baseline to generate structured trajectory data of road users; equipped with GPUs or dedicated AI acceleration chips, they deliver high-performance processing for image/video streams and AI model inference; and by transmitting fused perception data to nearby vehicles via RSUs over low-latency communications, they complete a closed-loop vehicle-road collaboration system.

In terms of performance benchmarks, roadside edge computing nodes must comply with stringent engineering standards. For instance, the Roadside Computing Unit (RSCU) jointly unveiled by Inspur Information and Baidu AI Cloud supports data processing for dual-carriageway 8-lane junctions, achieving an average end-to-end roadside object perception latency of ≤300 milliseconds, with positioning and speed accuracy exceeding industry averages. Meanwhile, the edge intelligent unit by Boyan Zhitong delivers up to 21 TOPS of AI computing power with power consumption below 30W, supporting up to 32 channels of 1080P HD video decoding, an operating temperature range of -40°C to 55°C, and an IP40 ingress protection rating. These devices are widely deployed across holographic junctions, digital twin platforms, traffic signal management systems, and vehicle-road coordination platforms.

Technological Evolution and Industry Application

The rollout of edge computing nodes in vehicle-road coordination is gathering momentum. Shiyan City plans to upgrade the infrastructure across 228 urban junctions and road sections to achieve broad coverage across its primary urban districts, deploying millimetre-wave radars, LiDAR, and edge computing units to deliver real-time traffic light data and blind-spot alerts for commercial vehicles. In Beijing Yizhuang, a 60-square-kilometre core zone has already seen the completion of 305 intelligent connected standard junctions, empowering roadside infrastructure to interact with passing traffic in real time.

On the regulatory and standardisation front, the standard "Internet of Things - Edge Computing - Part 4: Technical Requirements for Nodes" (GB/T 41780.4—2025) has been officially gazetted. It classifies edge computing nodes into four distinct categories: edge computing servers, edge gateways, edge controllers, and IoT terminals, setting clear benchmarks for functionality, performance, and security specifications. This enables edge computing hardware from diverse manufacturers to achieve seamless interoperability under a unified framework, speeding up the large-scale deployment of vehicle-road coordination solutions.

Technical Limitations

The operational efficacy of edge computing nodes relies heavily on the deployment density and data accuracy of front-end sensing hardware, communication network stability, and algorithm maturity. In scenarios involving sensor malfunctions or network dropouts, local decision-making capabilities may be compromised. Furthermore, while national standards now guide the setup of roadside edge computing nodes, cross-regional integration and interoperability remain ongoing works in progress. Additionally, when roadside data conflicts with a vehicle's onboard sensors, there remains a lack of clear legal and regulatory frameworks to delineate liability and responsibility.