Multi-access edge computing (MEC) is an edge cloud platform that provides a new network architecture based on the operator network. It uses the wireless access network to provide IT services and cloud computing functions necessary for telecom users nearby, thus creating a carrier-class service environment with high performance, low latency and high bandwidth and allowing consumers to enjoy high-quality service experience.
MEC can be described simply by the following equation:
MEC = (Connection + Computing) × Capabilities
—Connection: It solves the problem of multiple access. Users can access the same service deployed at the edge and have the same user experience no matter through 4G, 5G or even fixed network access.
—Computing: It provides computing resources needed for edge deployment, including CPU general computing capabilities and heterogeneous accelerated computing capabilities provided by GPU and FPGA. More importantly, computing resources can be flexibly scheduled according to service needs.
—Capabilities: It provides cloud computing resource capability, wireless network capability, and service-enabling capability, facilitating local computing and processing of edge applications.
"Connection + Computing" is the foundation and a support platform for MEC, while network capabilities can tap optimal business value and ensure differentiated and high value-added services. Therefore, MEC can be used not only as edge computing infrastructure, but also as a sharp tool for operators to increase the value of their core network assets.
MEC Capability Platform Framework
MEC deeply integrates traditional mobile cellular networks with internet services to reduce the end-to-end latency of service interaction and improve user experience by exploring inherent capabilities of wireless networks. MEC needs not only the support of communication network capability (CT capability) but also the support of cloud computing capability (IT capability), so it is the best choice for ICT convergence.
The basic MEC architecture can be divided into four layers (Fig. 1). The first layer is the infrastructure layer, including various types of servers such as computing servers (computing capability), storage servers (storage capability), and hardware acceleration cards, which meet the needs of AI reasoning, graph and image rendering, and high-speed network forwarding.
The second layer is the virtualization layer, which provides virtualization platform resources and management for upper-layer capability services and APPs, including VMs and containers, so that different applications can share the same infrastructure.
The third layer is the network and service capability layer. The network capabilities include local traffic offload and basic service capabilities such as network address translation (NAT), virtual firewall (VFW), domain name service (DNS) and load balancing (LB). Radio network information service (RNIS), bandwidth management service, service routing rule, and wireless indoor positioning service are also provided in a network capability exposure framework through the API interface. The edge service capability layer uses a microservice framework and introduces new capabilities such as AI and big data to enrich and improve the MEC capability layer.
The fourth layer is the applications deployed at the edge, such as video surveillance, AR/VR video and campus applications.
Wireless Network Capabilities
Operators need to consider how MEC can leverage the advantages of mobile networks and take CT capabilities as the focus to provide a unified MEC platform with ICT convergence. MEC can provide CT-specific wireless network capabilities such as local traffic offload, NAT/VFW/DNS/LB, and wireless indoor positioning.
Local Traffic Offload
Local traffic offload is a core capability of MEC. In application scenarios such as local computing or industrial parks, the first problem to be solved is how to flexibly and efficiently offload data streams locally and enable nearby access. The optional solutions include TOF+ and CUPS for 4G networks. The 5G-oriented UPF solution can use local area data network (LADN), uplink classifier (UL CL) or IPv6-based multi-homing for traffic offload.
ZTE provides a 4G/5G converged MEC local traffic offload solution. On the same MEC platform, the above traffic offload functions are used as service plug-ins. The flexible plug-in mode supports traffic offload in 4G, 5G NSA, and 5G SA.
After the local offload, data traffic is tunneled to the MEP platform that provides NAT, VFW, DNS and LB. In this way, data traffic is offloaded from the operator network to each application (APP).
Wireless Indoor Positioning
By integrating multiple positioning technologies such as indoor base stations and Bluetooth, MEC provides indoor positioning capability within a range of 3–5 meters. It can also implement coordination management through the MEC-based IoT management platform and wireless sensors such as geomagnetic fields, fire sprinklers, and fire alarms. The indoor positioning capability can be exposed to third party applications and the big data platform in shopping malls through the API mode, to provide users with service applications such as indoor navigation and intelligent parking, so as to provide location service capability based on the existing communication network.
With its QCell indoor equipment and MEC location service, ZTE has cooperated with application partners including Innsmap and INNS Big Data in smart shopping malls, smart buildings, and smart parks.
The MEC platform needs to provide service rule management and configuration, so edge applications can dynamically change local traffic offload policy through the configuration interface of relevant service rules and flexibly control local services by domain name, IP 5-tuple, user, and base station location.
DPI/TCP optimization aims to optimize network performance and improve user QoE. The MEC-based DPI function implements deep packet inspection on the MEC platform and notifies the inspection result to the base station through the accompanying packet. The base station guarantees differentiated scheduling algorithms for specific service types according to the set policy, so that users can get a better service experience.
The MEC-based TCP optimization solution combines the unique TCP air interface optimization of the wireless network and the TCP congestion optimization at the wired side to improve TCP service performance through HTTP fragment proxy, TCP transparent proxy, TCP congestion control, and wireless resource scheduling optimization.
The results of ZTE's test on China Unicom's existing network in Shenzhen show that after the MEC-based TCP optimization function is deployed, typical HTTP and video services can obtain about 15% performance improvement in the uplink and 30% in the downlink.
RNIS provides radio network information services for MEC applications and the MEC platform, and the information can be used to optimize existing services. RNIS can provide such information as cell ID, radio channel quality, cell load and throughput. With the introduction of AI and other artificial intelligence analysis and reasoning capabilities, QoS can be guaranteed at a finer granularity from the user level to the stream level and then to the packet level, and new network capabilities such as location awareness and link quality prediction can also be provided.
In the 4G era, the smart terminal technology that enables terminals to provide computing capability has completely promoted the development of PC internet towards mobile internet. In the 5G era, the MEC technology that allows network edge to provide computing capability will promote the convergence of cloud computing platform and mobile network, bringing a new business ecosystem and new business models.
MEC is not only a resource computing platform but also a wireless network capability platform. Through the deep convergence of mobile access network and internet services, MEC can improve user experience and save bandwidth resources. By deploying computing capabilities to network edges, MEC provides converged third-party applications, offering more space for service innovation at the mobile edge entrance.
Network connection of the MEC platform is the key, its computing capability is the effective guarantee, and its network capability and openness are the driving force. By deploying the MEC platform, leveraging 5G network advantages, and fully exploring radio network capabilities, operators can enable digital transformation of the industry, and provide infinite possibilities for creating more network value in the future.