Edge Native: Architecture and Technology for 5G MEC and Future Network Evolution

With the rapid development of 5G industry applications in 2020, existing edge computing capabilities show shortcomings that cannot fully meet diverse industry scenarios. A key limitation is that MEC platforms cannot operate completely independent of carrier 5G networks, requiring tight coordination between network and compute resources. In this context, the whitepaper proposes the Edge Native concept to support future network evolution and industry digital transformation.

The document first examines the development status of edge computing, the market demand for 5G MEC, and the pain points encountered during 5G MEC deployment, highlighting the need for a new Edge Native paradigm.

It then describes the Edge Native industry concept and its core technical capabilities, including edge orchestration, edge intelligence, and edge security, positioning Edge Native as an efficient coordination of connectivity and compute for carrier‑centric, industry‑focused architectures.

Edge Computing 2.0 is introduced, covering three deployment forms—cloud‑edge, edge‑cloud, and edge‑gateway—along with the focus on "cloud‑edge collaboration" and "edge intelligence" as primary capability directions. Software platforms must adopt cloud principles and provide real‑time, collaborative, trustworthy, and dynamically reconfigurable services, while hardware must support heterogeneous compute (x86, ARM, GPU, NPU, etc.).

Statistical data shows that over 5,000 industry pilot projects were launched in 2020, with about 60% involving MEC deployment; Chinese carriers have conducted more than 100 5G MEC commercial pilots across various sectors such as smart factories, mining, and energy.

The analysis identifies remaining technical gaps in network deployment, device capabilities, and software development. MEC must support richer heterogeneous devices, stronger AI‑driven data analysis, high‑reliability network collaboration, and improved system isolation and data security.

For developers, a typical MEC application workflow includes selecting a service framework based on language skills, defining REST APIs, leveraging 5G network capabilities (low latency, high bandwidth) via SDKs or templates, adapting to both IaaS and PaaS layers (e.g., OpenStack, Kubernetes), and employing DevSecOps toolchains for end‑to‑end security and orchestration.

Edge Native aligns with Cloud Native principles, emphasizing long‑term cultural and technical evolution. It shares the goal of making applications adaptable to underlying infrastructure characteristics such as elasticity, distribution, rapid deployment, and zero‑downtime delivery.

The architecture highlights the shift of ICT focus toward the edge, addressing challenges like long RTT to cloud centers, high transmission and compute costs, and privacy/security concerns. Edge Native abstracts telecom connectivity parameters (e.g., 3GPP) for developer friendliness.

Key technical components include:

Edge Infrastructure (EdgeInfra): heterogeneous hardware (CPU, GPU, NPU) and Kubernetes‑based platforms with enhanced multi‑network isolation, multi‑tenant support, and telecom‑specific features (e.g., TSN).

Edge Network: open 5G network capabilities defined by ETSI MEC and 3GPP NEF, enabling end‑to‑end compute‑network integration.

Edge Orchestrator: strong orchestration for small‑capacity edge nodes, supporting mixed container/VM workloads, multi‑cloud plug‑in architecture, and various deployment scenarios (Kubernetes, OpenStack, K3s, MicroK8s, RunC).

Edge Collaboration: coordination across edge‑edge, edge‑cloud, and edge‑device domains to ensure low‑latency, high‑throughput services.

Edge AI: unified edge intelligence framework offering model and data partitioning, distributed learning, and heterogeneous hardware scheduling via a HAL layer.

Edge Security: end‑to‑end security mechanisms, blockchain‑based immutable data storage, and trusted execution environments.

Edge Data: distributed, heterogeneous, and lightweight storage solutions (e.g., TSDB for IoT).

Although 2020 MEC pilots have made progress, Edge Native remains at an early stage; its architecture and technologies require further research and industry collaboration. The involved alliances and open‑source communities will continue to refine the Edge Native framework and promote its practical adoption.