5G is coming, we all have to stand on the sidelines

Unconsciously, 5G has come to us. There is no need to study those market research reports, just look at the 5G labels on the mobile phone advertisements around you, which is a clear proof.

However, the impact of 5G on our lives is far more than just "changing a new phone". As the network backbone of the future connected world, 5G's ambition is to cover the "full-scenario" wireless interconnection needs that our imagination can reach. Specifically, it can include three categories:

•eMBB : Enhanced Mobile Broadband, which can achieve high-speed data transmission at a rate of 1Gbps;

•uRLLC : Ultra-high reliability and low latency communication, providing a bit error rate as low as 10-5 and a latency of no more than 1 millisecond;

•mMTC : Massive Machine Type Communications, which can support up to 1 million connections per square kilometer, covering low-speed and low-power IoT applications

Figure 1: 5G application scenarios and required network carrying capacity (Source: IMT-20205G Promotion Group)

However, although such "full-scenario" coverage looks beautiful, there are still considerable challenges in its implementation. When the massive amounts of data generated at the network terminals are transmitted through the network to the cloud computer room for processing and then returned to the terminal to complete the business closed loop, it will inevitably cause network congestion, increased latency, and lower connection success rate. Such user experience is somewhat distant from the "original intention" of 5G.

From the perspective of telecom operators, such bandwidth consumption will cause huge transmission pressure on the backhaul network and business center. At the same time, the three major application scenarios require different network performance. It is not easy to adapt to such diverse business needs on a single network architecture. All of this means huge costs.

Therefore, in order to realize the promise of 5G, it is necessary to transform the existing wireless communication network architecture. The focus of this transformation is "edge computing".

In the traditional cellular mobile network structure from 2G to 4G, information processing is mainly completed by the data center of the core network, that is, all information needs to be transmitted from the edge of the network to the core network for processing and then returned to the edge of the network. This model is a challenge to the economy and flexibility of network deployment and operation. The concept of "edge computing", which has been quite popular in the field of network cloud computing, can solve this problem.

Specifically, in the 5G transmission network, gateways, servers and other equipment are deployed in the computer room close to the access side (edge ​​end), and the computing power originally concentrated in the central computer room is moved down to the edge of the network. This is actually to establish a small-scale data center on the edge side of the network to directly process information and data from edge devices without "handing over" every bit to the central computer room for processing. This model is called Mobile Edge Computing (MEC).

Figure 2: MEC architecture in 5G (Source: Wireless Deep Ocean)

The benefits of MEC are obvious:

•Since there is no need to transmit data back to the core network, network latency is greatly reduced.

• Local, low-value data can be processed at the edge computing center, and if necessary, the “refined” data can be transmitted to the core network, which greatly reduces the pressure on network bandwidth and central computer room resources.

•Edge data centers can flexibly define the functions of edge networks based on the application scenarios they cover, improve the efficiency of content and service distribution, and enhance user experience. New business models are likely to be nurtured in them.

To be more specific, the deployment of MEC will provide strong support for the three major application scenarios of 5G.

Let's talk about eMBB applications first, which mainly include high-definition video, VR/AR and other application scenarios that require extremely large network bandwidth. The data processing method of traditional networks has a long transmission path, occupies a lot of bandwidth, and has a large network delay. After deploying MEC, you can configure video cache on MEC to process VR/AR data, and the experience in network speed and delay will be better. Operators and users can also use MEC to create special services for personnel and data-intensive places such as enterprises, shopping malls, and schools, such as supporting VR shopping guides and indoor navigation services in large shopping malls, highlighting the value of 5G.

In terms of uRLLC high-reliability applications, since MEC is closer to the network edge nodes, time-sensitive data can be directly processed in MEC, and millisecond-level latency is undoubtedly more guaranteed. This is critical for the automotive, industrial and other fields.

The mMTC service needs to connect to a large number of IoT terminals and collect the data they generate. For the core network, these unprocessed raw data will be a heavy burden. However, if the data is transmitted to MEC, the key information will be extracted and reported to the core network after analysis and processing by MEC, and the data quality and quantity will be optimized.

Of course, there are still many factors that need to be carefully considered in the actual deployment of MEC. After all, even if MEC has many benefits, if it cannot bring real money to operators, it will also affect everyone's enthusiasm for moving forward and constantly trying. In reality, different choices will have different effects. In summary, operators have three choices for MEC deployment:

• Site room deployment: This method is closest to the end user and can cover a range of about 1 km (≤5 base stations). It is small in scale and high in cost, but it can still be considered in areas with high user density (such as commercial districts).

• Integrated access room deployment: covers a radius of 5-10 kilometers (about 20 base stations), the service scale is neither large nor small, and the return on investment is not significant, so operators generally do not consider this deployment method.

•Edge data center deployment: covering 50-100 kilometers (more than 100 base stations). Generally, operators believe that deploying MEC here is most conducive to balancing the relationship between costs and benefits, and it is also the choice of most MECs currently.

Figure 3: Different deployment methods of MEC (Source: Wireless Deep Ocean)

How to deploy it in the end still requires specific analysis of specific issues. Fortunately, MEC itself provides such flexibility in deployment, which is also conducive to operators to explore and experiment, and finally find the best solution based on the demand scenario.

Given that the temptation of 5G is so great, there are indeed many promoters in the industry for "5G edge computing". In addition to traditional telecom operators, some Internet companies have joined in and provided experimental solutions. Active industry organizations are also exploring whether it is possible to extract the common needs of MEC and build a universal network. This standardization attempt will also provide support for the future development of MEC.

Figure 4: Tencent Cloud's 5G edge computing center can be deployed with just a flat surface, electricity, and internet access (Source: Leifeng.com)

In short, all efforts are telling us that 5G is so unique that if we want to maximize the power of 5G, introducing edge computing and standing on the "edge" is an inevitable choice. This is a challenge and an opportunity.