News

Background

At MWC Las Vegas 2024, Tannera, Phluido, and Effnet introduced a commercial-ready Cloud-native Private 5G RAN solution, aligning with the AI-RAN alliance's mission to drive innovation in integrating AI and RAN technologies. The Private 5G RAN serves as a foundational building block, enabling the deployment of robust, scalable and flexible network infrastructure. This demonstration showcases how AI applications can be effectively integrated into RAN to enhance performance, addressing the objectives outlined by WG3 of the AI-RAN alliance.

Problem Statement

The rollout of Private 5G networks faces challenges in achieving cost efficiency, scalability, and the ability to support AI applications that demand low latency, high throughput, and consistent performance. Industries such as industrial automation and surveillance require solutions capable of meeting these stringent requirements while remaining adaptable to diverse deployment scenarios. Additionally, there is a need to benchmark and evaluate performance metrics such as latency, jitter, throughput, and encryption overhead to meet the evolving demands of AI-enabled use cases.

Key Features Demonstration

The demo highlights how the Private 5G RAN solution integrates AI and RAN seamlessly:

• Interaction between AI and RAN: The AI application leverages real-time data streamed over the RAN to perform object detection. The RAN’s low-latency and high-throughput characteristics enable rapid processing and feedback loops, critical for applications like industrial automation and surveillance.
• Before: Traditional setups required separate infrastructure for RAN and AI, leading to higher costs and latency issues.
• After: The Tannera FEC acceleration IP reduces CPU load, enabling simultaneous RAN and AI workloads on a single, low-cost, low-power server.
• Comparative Aspects: The solution demonstrates significant improvements in cost efficiency and performance by reducing Total Cost of Ownership (TCO).
• Benchmarking Key Metrics (WIP/TBC):
  • Latency (real-time object detection is achieved within stringent packet delay budgets)
  • Throughput (the system supports bi-directional data streams for AI and RAN workloads)
  • Jitter (consistent latency variations ensure reliable AI task execution).
• Improvements Highlighted: The integration reduces TCO while opening new opportunities for AI-driven gains. Real-time object detection using OpenCV is seamlessly integrated, with marked detections streamed via a Grafana dashboard alongside live KPI metrics for RAN performance.

Check out the demo

Check out the demo in the following video.