Edge computing is the most important application mode of 5G, and it is the key to 5G enabling all walks of life at the network edge. The issuance of 5G licenses in the middle of this year, and the three major operators officially launched 5G commercial use on November 1, which makes 5G edge computing move from trial and pilot to trial commercial use. However, for operators, how to establish a commercial edge computing network, how to meet the needs of edge computing at the same time, but also have good benefits, these are difficult.

UR9KH Start with the demand and the blade goes inwards and outwards

Lei Bo, co-chairman of the ECNI Working group of the Edge Computing Industry Alliance and director of the IP and Future Network Research Center of the China Telecom Strategy and Innovation Research Institute, said in an interview with the China Electronics News reporter: “What is the edge computing network, the computing node is placed in the computer room on the edge of the network, near the base station, is the edge computing network?” In fact, this is the early understanding of operators to develop edge computing, in the face of commercial use, this understanding has proved to be wrong.

UR9KH When operators put servers on the edge of the network, they gradually found that this alone could not realize the imagined edge computing, because the existing network carried the edge computing traffic, the traffic flow in the network, that is, the design of the network route, did not meet the expected goal of edge computing, and could not meet the promised 5G low delay and high bandwidth value to customers. Also can not realize the data is not out of the local requirements.

In 4G networks, the network layers from the bottom up are respectively the access network composed of base stations, the bearer network above the access network, and the core network above the bearer network. The industry often refers to the traffic in the same level of the network as east-west traffic, the traffic of different levels of the network as north-south traffic, the traffic itself carries data information, and the network routing according to the information to determine which direction to send the traffic.

UR9KH “The original traditional wireless carrier network is north-south traffic, the user’s data service will be sent to the core network, and the core network is the provincial network.” Dr. Song Jun, co-chairman of the ECNI working group of the Edge Computing Industry Alliance, told the reporter of China Electronics News that “MEC is now a clear localized traffic model, a lot of traffic is local and as close as possible to users, and some data traffic does not leave the enterprise campus.”

“It made us rethink what we should do from an edge computing point of view and see what kind of needs it has for the surrounding network.” Raber said.

Start from the benefit of general technology individual use

UR9KH At the same time, the prospect of 5G edge computing has made more and more industries interested in edge computing. “It used to be one or two customers talking to us operators about how to meet the requirements through edge computing, and as the number of customers increased, we found that we couldn’t do it.” “Now we need a platform to tease out the different customer needs and figure out how the network can adapt to the key needs,” Mr. Lebo said.

This idea has reached a consensus among operators and equipment manufacturers. In September of this year, the Edge Computing Industry Alliance formed the ECNI(Edge Computing Network Infrastructure Working Group). The working group has two main goals. One is to understand the network challenges of deploying MECs at scale. The second is to gradually establish a complete and clear network hierarchy and technical system like cloud computing while accumulating MEC technology.

“We want to use ENCI as a platform to put all the edge computing requirements in a pool to discuss, to reach the different needs of customers, preferably a network can solve, which is one of the goals.” Raber said.

UR9KH To achieve this goal, the challenges are considerable. Because MEC applications have strong individual needs, how can the universal network provided by operators meet various individual needs?

For example, in the fixed-mobile converged service scenario, the mobile network and fixed network must simultaneously access edge computing services, and the connection requirements of low latency and high reliability must be met to achieve seamless service experience. For example, in the integration scenario of carrier network and campus network, the demand mainly focuses on the access of new mobile network technologies such as 5G and network interconnection, interoperability and interoperability. In the field edge computing network, the requirements are mainly the integration of OT network and IT network, and the determination, real-time, reliability, and security requirements of on-site services.

“What we want to do is to abstract out the common needs of edge computing, and if the future edge computing network can match these key common needs, it means that it can meet 80 to 90 percent of the needs of edge computing, so that you can find a balance between personalization and universality.” Song Jun said.

“In the long run, if we can design the architecture of this network well, it can be fed back to various industries, such as some needs in the field of industrial Internet, which can be appropriately modified to achieve the best match between the two sides.” Lei Bo said, “Some industrial Internet fields set the indicators too high, in fact, he does not have such a high demand, just think that operators should provide such capabilities, and these high indicators often mean high costs, in this case we can do coordination, we put some indicators up, he put the requirements a little lower, we in the cost and efficiency to achieve optimization.”

The biggest advantage of general-purpose technology is the scale of application, so that the marginal cost continues to fall. “Edge computing in the future can’t be too expensive, or users won’t be able to accept it. We need to drive down the cost of construction, so that the business price of edge computing networks can also fall. Industrial customers can really use it.” Raber said.