Serverless computing in 5g-enhanced MEC

Multi-Access Edge Computing (MEC) has the potential to support low-latency, high bandwidth, and massive connectivity, which are critical for applications such as autonomous vehicles, AR/VR, and industrial IoT. Despite its benefits, MEC faces challenges in resource management and dynamic service provisioning due to distributed, heterogeneous nodes. Serverless computing (Function-as-a-Service) offers a promising solution by abstracting infrastructure, enabling event-driven execution, and providing automatic scaling and fine-grained billing. The integration of 5G networks further enhances this synergy. 5G’s ultra-reliable low-latency communication (URLLC), massive device connectivity, and high throughput complement MEC by enabling real-time, latency-sensitive applications. Combining serverless computing with 5G-enhanced MEC can deliver scalable, cost-efficient, and responsive edge services for next-generation applications.

To fully realize this potential, intelligent orchestration is essential. Leveraging 5G network data, such as traffic patterns and mobility information, MEC can proactively forecast demand and optimize serverless function placement across edge nodes while meeting stringent SLAs for latency, availability, and privacy.

Work on optimal service placement and resource allocation for serverless MEC exists, but a key gap is the limited integration of these solutions with 5G, and potentially 6G control plane [1,2]. Consequently, the network remains unaware of serverless deployment state (e.g., pre-warmed instances) during Packet Data Unit (PDU) session establishment, resulting in sub-optimal traffic steering and avoidable cold starts. Prior work, including [3], advances orchestration within MEC but do not enable end-to-end coordination between serverless placement decisions and mobile network management functions. Furthermore, current 5G enhancements primarily optimize PDU setup for single pre-warmed functions and do not extend to multi-function service function chains (SFCs) that characterize realistic, composite edge workloads.

This research will address these limitations by integrating serverless orchestration with the 5G/6G control plane to enable computing-aware traffic steering based on live deployment state, and by extending from single functions to full-service function chains. The resulting high-level framework will coordinate placement, scaling, and routing across MEC sites using network analytics to meet end-to-end SLAs while improving efficiency and robustness under realistic mobility and demand dynamics.

References

1. Russo, G.R., Spaziani, P. and Cardellini, V., "Towards QoS-Aware Serverless Function Offloading in the Edge-Cloud Continuum through Reinforcement Learning," 2025 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW), Milano, Italy, 2025, pp. 1073-1080, doi: 10.1109/IPDPSW66978.2025.00168.

2. Bensalem, M., Carpio, F. and Jukan, A., "Towards Optimal Serverless Function Scaling in Edge Computing Network," ICC 2023 - IEEE International Conference on Communications, Rome, Italy, 2023, pp. 828-833, doi: 10.1109/ICC45041.2023.10279357.

3. Tran, M.N. and Kim, Y., 2024. Design of 5G architecture enhancements for supporting serverless computing. IEEE Access.