The edge computing dilemma of operators in the 5G era

With the continuous development of 5G, 5G will be widely used in all walks of life in the future. The emergence of the new "5G + edge computing + AI" model has also prompted operators to help vertical industries achieve digital and intelligent transformation. However, this has also brought four new challenges to operators' bearer networks. In order to build a 5G MEC Ready bearer network, operators need to solve six key problems.

MEC is the key to digital transformation of industries in the 5G era

Application localization (no data transmission outside the campus), high-bandwidth content distribution, and low-latency computing localization all promote the migration of service content, applications, and computing to the edge, thereby promoting the development of multi-access edge computing (MEC) and the downward migration of 5G core networks.

Figure 1 Migrating services to the edge promotes the development of MEC and the downward migration of 5G core networks

The 5G core network adopts a flexible architecture that separates the user plane function (UPF) and the session management function (SMF). In this way, the UPF can be flexibly deployed on demand. One SMF can manage multiple UPFs at the same time while ensuring the high performance of the 5G core network. 5G brings many new advantages to MEC, including:

• The core network UPF is moved down to the enterprise campus to ensure that key service data does not leave the campus and provide a low-latency bearer solution. Operators can configure UPF for each enterprise independently to customize wireless services for enterprise users.
• Operators provide open and programmable 5G communication service functions such as 5G positioning and wireless communication, which can be used by enterprise users and integrated into enterprise service systems. Enterprises can customize their own 5G innovative applications.
• The sunken 5G MEC system is directly interconnected with the enterprise network, enabling real-time integration and simplification of applications distributed on the two network systems, and facilitating the development of customized innovative applications.

Four major challenges facing operators’ 5G MEC bearer networks

Traditional 4G bearer networks, because traffic is mainly north-south, many operators use L2+L3 methods, which are no longer suitable for 5G MEC traffic localization requirements. 5G MEC poses four new challenges to operators' bearer networks:

Figure 2 New challenges for operators’ 5G MEC bearer networks

• On-site MEC (deployed in enterprise campuses) is a new application scenario. Operators need to have low-latency connections between base stations and MECs within enterprise campuses. Important business data of enterprises cannot leave the campus, which poses new challenges to operators' access networks.
• The downward movement of UPF in 5G MEC leads to the downward movement of service ports in the 5G core network (such as N4, N6, N9, 5GC OAM and other interfaces), which makes the L3 VPN (wireless core network) originally on the backbone network in the 4G era move down to the UPF access point; at the same time, the large-scale distributed deployment of UPF increases the coverage of L3 VPN. The carrier bearer network needs to support the downward movement of L3 VPN and the extensive coverage of L3 VPN network to adapt to the new challenges of large-scale deployment of 5G MEC.
• The UPF of MEC needs to communicate with its control plane (SMF) and the management and control system of the 5G core network in the central cloud, and meet the high-performance communication requirements of the telecom cloud. MEC applications may be part of cloud computing in the data center (DC) and deployed at the edge. They need to interconnect and collaborate with this cloud computing application, which brings new challenges to the edge-cloud collaboration on the operator's bearer network.
• MEC supports integrated access on fixed and mobile networks and provides seamless fixed-mobile convergence (FMC) services. The bearer network needs to provide MEC with connections across mobile and fixed bearer networks to provide service interoperability between MEC and the central cloud and between MECs. This poses new challenges to network architecture, especially for operators that have both mobile bearer metropolitan area networks and fixed bearer metropolitan area networks.

Six key issues for operators’ 5G MEC network architecture model and network construction

Figure 3: Operator bearer network architecture model from the perspective of MEC

The bearer network architectures of different operators are diverse. The following sections introduce the bearer network architecture model from the perspective of MEC in Figure 1-3. The above 5G MEC network communication model requires operators to solve the following six key issues when building a MEC bearer network:

(1) Shortest MEC access network: Operators need to provide the shortest path for the N3 interface service flow from gNB to MEC UPF. In the on-site MEC mode, the N3 interface service flow needs to be forwarded to the MEC directly through the mobile bearer router in the campus. In addition to ensuring low latency and saving bandwidth on the operator's network, this also ensures that the enterprise's critical service data does not leave the campus, as shown in Figure 4. This requires the MEC access router to forward data packets through the shortest path. To this end, the MEC access router is required to provide the necessary routing functions (L3 to edge).

Figure 4 MEC requires a low-latency access network

(2) Low-latency slicing: To meet the requirements of MEC applications for low latency, high security, and high reliability, the operator's bearer network needs to provide low-latency slicing network services for enterprise users. The MEC slice network includes gNB, mobile bearer network (between gNB and MEC), and UPF. That is, all network elements that the enterprise service flow passes through to MEC. The fewer network elements the data packet passes through, the simpler the slice and the shorter the transmission delay.

(3) MEC multi-point communication: The business flows between MEC and the 5G core network (N4 and OAM interface), MEP management platform, and other MECs are all multi-point to multi-point communication modes and require L3 VPN support. The MEC bearer network needs to provide L3VPN functions on the entire network (including the access network); that is, connect the L3 VPN to the network edge. In addition, L3 VPN needs to span multiple network segments, such as the metropolitan area network and the backbone network. Compared with the 4G bearer network, the MEC bearer network is much more complex in terms of the number of network elements (a large number of UPFs are moved down) and network coverage (from access to backbone). Therefore, a flexible and powerful L3 VPN is needed to support multi-point communication, as shown in Figure 5.

Figure 5 Management and control service interface across multiple networks

(4) Communication functions integrated in the router of the MEC system: Small and micro MECs are common in 5G MEC. Due to cost and communication requirements, MEC usually uses a one-layer integrated network model (as shown in Figure 6), while data centers generally use a complex multi-layer network architecture. The router of MEC needs to provide all the necessary communication functions, such as intercommunication between devices in MEC, reliable connections at Layer 2 and Layer 3 between VMs, intercommunication and reliable communication with the external IP network (IP RAN), and edge cloud synergy. UPF as a network function virtualization (NFV) can run on multiple VMs to improve performance and reliability. MEC routers need to provide equal-cost multi-path routing (ECMP) for the current high-performance UPF to achieve 16-path load balancing.

Figure 6 MEC network model

(5) Edge-cloud synergy: MEC UPF is a data plane that is sent down to the 5G core network, and the application in MEC is a real-time processing unit that is sent down to the cloud service. Both require the operator's bearer network to provide reliable cloud-edge communication capabilities, and also need to support edge-cloud collaboration in terms of automated deployment and operation and maintenance. The cloud-edge collaboration of UPF can refer to the bearer solution of the telecom cloud.

(6) Secure intercommunication between the two networks: The operator's MEC network needs to be interoperable with the enterprise network so that the enterprise can integrate 5G communication capabilities and MEC applications into the enterprise's business system. Currently, the router in the MEC is generally used to interoperate with the enterprise network. Network security is a matter of great concern to both the enterprise network and the operator network, and a firewall-based network security solution is required.

Summarize

5G mobile communication systems have made many improvements in supporting vertical industries, such as low-latency wireless communications, flexible core network architecture, and super uplink, which are the main features that distinguish them from 4G. MEC is a new model for operators to help vertical industries become digital and intelligent. MEC is the beginning of the widespread distribution of intelligence on the network. In the future intelligent world where everything is connected, intelligence based on edge computing will be scattered all over the network.

The 4G bearer network is built based on the 2C (for ordinary mobile phone users) concept. The traffic model is a simple north-south, wireless core network centralized model, and does not consider the MEC network requirements for vertical industries. Therefore, the construction of the 5G MEC bearer network is not a simple bandwidth upgrade of the 4G network.