ENVELOPE Demonstrates AI-Driven Network Programmability for Reliable CCAM Services
Connected and Cooperative Automated Mobility (CCAM) applications rely on communication networks that can respond dynamically to changing conditions. In a live demonstration, the ENVELOPE Project showcases how AI-driven network programmability can help maintain the performance of a cooperative perception service even when network congestion starts to affect the connection.
From network prediction to application adaptation
The demonstration focuses on a cooperative perception scenario involving two vehicles. The front vehicle streams live video to the following vehicle over a best-effort 5G connection, enabling the rear vehicle to receive real-time information from the road ahead.
As additional background traffic is introduced into the network, available resources become increasingly congested. Instead of waiting for the service quality to deteriorate, ENVELOPE uses predictive network intelligence to anticipate the degradation.
Through the CAMARA Connectivity Insights API, the CCAM application receives predictions about whether its network performance requirements will continue to be met. When the system detects that degradation is expected, an alert is delivered to the application, triggering the next stage of the adaptation process.
Programmable connectivity in action
Once the network conditions begin to affect the cooperative perception stream, the front vehicle uses the CAMARA Quality on Demand API to request a guaranteed bitrate session. This provides the necessary uplink resources and restores the application’s performance, bringing latency back to approximately 200 milliseconds even while background traffic continues to increase.
The demonstration then addresses the connection of the rear vehicle. Using Access Traffic Steering, Switching and Splitting (ATSSS), 100% of the rear vehicle’s traffic is steered over a non-3GPP access, such as Wi-Fi, reducing the load on the congested 3GPP network.
By combining predictive network insights, Quality on Demand and multi-connectivity, both vehicles are able to achieve latency below 200 milliseconds and maintain a stable, high-quality cooperative perception experience.
From static connectivity to intelligent network adaptation
The demonstration illustrates a key principle behind ENVELOPE: CCAM services should not have to operate on top of a static network. Instead, applications and networks can interact through programmable interfaces, allowing connectivity and network resources to be dynamically adapted to application requirements.
The demo uses an OpenCAPIF-based API Gateway to interact with exposed network APIs, bringing together network intelligence and programmable connectivity mechanisms in a single workflow.
This approach demonstrates how AI-driven network programmability can support the stringent performance requirements of next-generation mobility services, enabling networks to anticipate problems and adapt before service quality is significantly affected.
The ENVELOPE demonstration provides a practical example of how predictive intelligence and programmable connectivity can work together to make future CCAM services more resilient, responsive and reliable.