5G MEC Networking Platform

Introduction

In order to upgrade their infrastructure to meet the latency and Quality of Service (QoS) requirements of 5G while reducing costs, commercial or private network operators can implement a MEC (Multi-access Edge Computing / Mobile Edge Computing) architecture, and utilize Network Functions Virtualization (NFV) together with general purpose servers to implement Virtual Evolved Packet Core (vEPC) technology, replacing expensive proprietary hardware and software. GIGABYTE's server products provide an effective hardware base to implement a MEC solution.

Network & Cloud Virtualization at the Edge

A MEC server platform combines NFV and Cloud Virtualization technology to minimize mobile backhaul bandwidth requirements and provide an ultra-low latency edge cloud platform. Adopting a MEC network architecture based on GIGABYTE servers to replace propriety networking hardware and software is an ideal way to reduce infrastructure costs while enabling network operators to successfully transform their 4G network to support 5G technology.

Autonomous Driving

Data from the vehicle's sensors and external traffic conditions from the cloud needs to be combined and processed in a split second and transmitted back to the onboard vehicle computer to make a decision to turn, brake etc. There is not enough time for this data to travel back to a central cloud to be processed. Mobile Edge Computing can process and transmit back the data in a timely manner.

Mobile AR / VR

AR & VR require powerful graphics processing capabilities which have previously relied on local compute / GPU host servers physically located next to the user. 5G and Mobile Edge Computing will allow the user to experience high definition and low latency AR / VR using only a lightweight mobile headset or device.

Massive IoT Network

The high bandwidth and low latency of 5G networks will allow for a mMTC (Massive Machine-Type Communications) IoT network. However, this huge amount of data will place a heavy strain on network backhaul capacity if it needs to be transmitted back to the cloud to be processed. MEC solves the backhaul issue by processing and re-transmitting the data at the network edge.

The Challenge of Implementing 5G

5G - the next generation of mobile telecommunications technology, is on the horizon and promises to deliver a myriad of new services and capabilities such as Enhanced Mobile Broadband (eMBB), Ultra Reliable Low-Latency Communications (URLLC) and Massive Machine-Type Communications (mMTC). A new band of the radio frequency spectrum between 30 – 300GHz has been opened for use, and a new telecommunications standard for 5G has been defined encompassing network speed, latency, the number of devices that can be connected, QoS and other conditions.

However, enabling 5G technologies requires not just a simple upgrade of User Equipment (UE) with a 5G modem and front end Radio Access Network (RAN) equipment. To meet the specifications of 5G such as speed, end-to-end latency and QoS, network operators will need to upgrade their entire front to back-end network topology. Upgrading this network infrastructure to enable technically complex 5G service could be extremely costly. In addition to the spectrum purchase cost, infrastructure investment costs will be huge: the number of base stations required for deployment will be four times that of the past, and construction costs will be 10 to 20 times higher than that of the 4G period.

The Solution: A GIGABYTE MEC Server Platform

Reduction of Infrastructure CAPEX and OPEX

Better Service Quality

High Reliability

Network Maintainability

High Scalability

Reduction of Infrastructure CAPEX and OPEX

The market size of general-purpose server equipment is much larger than that of dedicated telecommunications equipment. Therefore, the initial purchase cost will be lower, and performance / price ratio will be higher. A MEC platform can help reduce Capital Expenditure (CAPEX) for building a 5G network to be more in line with market expectations.

Using a MEC network topology, the data uplink / downlink between the base station and the core network will no longer be needed to support such a heavy load of local IP traffic. Therefore, private lines or dark fiber leases can be changed to general internet access links, allowing more competitive operating expenses (OPEX) .

Better Service Quality

A MEC server features Virtualized Evolved Packet Core (vEPC) functionality, allowing the UE to change signals between base stations without having to go back to the core network to complete the inter-base station handover. Only after the successful handover, will the necessary information be returned to the core network EPC. This can improve the 5G network service quality, especially when the user is moving at high speed.

High Reliability

A MEC server provides redundancy at the infrastructure level with NFVI, without relying on physical modules. The abstract nature of cloud virtualization resources ensures the virtual network functions (VNF) running on it will not be affected by service interruptions due to physical server or network connection errors. For example, EPC key network entities such as Mobility Management Engine (MME) VMs can run multiple copies, with hot standby technology providing N+1 backup for each other. If the main VM fails, the standby VM can take over within a fraction of a second, without the user feeling the EPC service has been temporarily interrupted.

Network Maintainability

5G Standards include specifications for a wireless IoT (mMTC: Massive Machine-Type Communications) - which places a huge burden on a network's Evolved Packet Core (EPC) in terms of system scale and scalability. The MEC platform can form a QoS policy through the Policy and Charging Rules Function (PCRF) network entity. The Software Defined Networking (SDN) controller will convert this policy to a ruleset, that will be published to the topology of the switch that groups traffic, and will use a Virtual Local Area Network (VLAN) method to manage operations.

High Scalability

Virtualization technology allows a MEC server to flexibly increase or decrease resource usage according to demand, so as to effectively utilize the overall resources. And in addition to the above-mentioned aspects of providing redundancy protection in NFVI, virtual resources and VNF can be dynamically managed, including the addition and removal of VMs, ensuring that every vEPC running can maintain proper configuration and operation.

The Core Value of MEC

A MEC platform can support a heterogeneous configuration of an ultra-dense 5G RAN with various vertical services, serving as an ingress point from the RAN to the core network as well as performing Self Organization Network (SON) functionality for the RAN, and local traffic breakout to offload traffic toward the core network. It is designed to minimize end-to-end delay, perform QoS negotiation/management with the RAN and the Software Defined Networking (SDN) / NFV based core network, and also manage some virtualized RAN functions.

Due to the rapid deployment capabilities of virtualized / cloud computing, the network entity setup / configuration period for the core network can be greatly shortened using the MEC platform. Virtualized devices can also be easily scaled out according to additional demand and remotely & dynamically updated, ensuring a greatly reduced system maintenance burden. Furthermore, the load balancing capabilities of cloud computing further improves QoS. In addition to the migration and re-generation ability of Virtual Machines (VMs), a hot standby feature (in addition to the main VM, a redundant spare will also be available to take over) ensures a high level of network reliability.

MEC Platform Features

1. Hypervisor-Based and Container-Based Hybrid NFV FunctionsA MEC platform can support both OpenStack NFV (OPNFV) and Kubernetes NFV Infrastructure (NFVI) solutions, providing a unified management interface for virtual machine and container application services, and providing application services for easy and quick deployment.

2. Service / Traffic Offload
Acting as a System Architecture Evolution (SAE) gateway, the MEC provides dynamic path judgment and can support the following usage scenarios:
・Traffic offload function for the specified Access Point Name (APN)
・If the user uses the MEC service, the packet is directed to the NFV platform of MEC.
・If the user is not using the MEC service, the packet is directed to the Packet Data Network Gateway (P-GW) and connected to the Internet through the carrier's core network.

3. Seamless Service Redirection

Seamless Service Redirection can work with various cloud platforms such as AWS, Google Cloud etc. to meet the low latency requirements of 5G application services. The application service can seamlessly run on the MEC platform, and the user application can connect to the application service of the MEC platform without any modification needed.

4. Just-In-Time Service Initiation

The MEC is a distributed cloud platform with limited system resources, as opposed to a more traditional data center cloud. Therefore, in order to make more effective use of these limited resources, the MEC features Just-In-Time Service Initiation technology, so that application services can start dynamically when needed and shut down when not in use, reducing resource idleness. A containerized application service can be dynamically launched within 1 second.

GIGABYTE Servers as the Hardware Base of a MEC Platform

GIGABYTE offers a range of server solutions ideal to be used as a MEC platform. These range from compact edge server platforms such as the H242 Series which can be deployed at base station towers, to larger and more powerful platforms such as the G291-Series (G291-280 / G291-Z20) or the G591-HS0 which feature GPU acceleration to be used for AR / VR or mobile gaming applications that can be deployed at aggregated edge data centers.