Multi-Access Edge Computing

Description

Multi-Access Edge Computing (MEC), formerly Mobile Edge Computing, is a system architecture defined by ETSI and integrated into the 3GPP ecosystem that brings application hosting and cloud computing resources from centralized data centers to the network edge. The 'edge' is physically and logically close to the end-user, typically at base station aggregation points, central offices, or even within the radio access network (RAN) itself. A MEC platform consists of a virtualization infrastructure (e.g., a small data center) that hosts MEC applications and provides core MEC services. These applications run on top of a MEC host, which includes the MEC platform and the virtualization infrastructure.

The architecture is anchored by the MEC system, which comprises the MEC host and the MEC management. The MEC host contains the MEC platform (offering service APIs) and the MEC applications. The MEC management includes the MEC orchestrator (for lifecycle management of applications) and the MEC platform manager. Crucially, MEC provides a set of standardized APIs, most notably the Radio Network Information Service (RNIS) API, which allows authorized applications to access real-time, contextual information about the radio network conditions (e.g., UE location, cell load, throughput). Another key API is the Location API. This exposure of network capabilities is a fundamental aspect of MEC.

How it works involves traffic steering and application hosting. User plane traffic can be routed (steered) to a local MEC application instead of being backhauled to a distant internet gateway. This is achieved through mechanisms like User Plane Function (UPF) selection and traffic offload in the 5G Core network. For example, a latency-sensitive augmented reality application can be hosted on a MEC server at the edge. When a UE requests this service, the network's Session Management Function (SMF) selects a UPF that is co-located with the MEC host. The UE's data traffic is then routed to this local UPF and onward to the MEC application, resulting in minimal latency. The application can also use the RNIS API to adapt its service based on the user's radio link quality or location.

Purpose & Motivation

MEC was created to address the limitations of centralized cloud architectures for latency-sensitive, bandwidth-intensive, and context-aware applications. It solves the problem of network congestion and high latency caused by backhauling all traffic to distant core data centers. The rise of applications like autonomous vehicles, industrial IoT, immersive VR/AR, and real-time video analytics demanded single-digit millisecond latencies and efficient local data processing, which traditional mobile networks could not provide.

The historical context involves the evolution towards 5G, where key usage scenarios like Enhanced Mobile Broadband (eMBB), Ultra-Reliable Low-Latency Communications (URLLC), and Massive Machine Type Communications (mMTC) require edge computing support. Initial concepts from ETSI ISG MEC were integrated into 3GPP specifications starting with Rel-15 to ensure seamless interoperability with 5G system architecture. MEC transforms the network from a pure connectivity pipe into a distributed computing platform, enabling new business models for operators and vertical industries by allowing third-party applications to leverage edge resources and network information.

Key Features

• Hosts applications in close proximity to end-users at the network edge
• Provides standardized APIs (e.g., RNIS, Location) for application access to real-time network context
• Enables local traffic breakout and user plane optimization via UPF selection
• Supports ultra-low latency and high bandwidth for demanding applications
• Facilitates application lifecycle management (onboarding, instantiation, termination)
• Enables network and service exposure to third-party application developers

Evolution Across Releases

Defining Specifications