Q: How does MEC discover services and configurations?

The service registry provides visibility into the services available on the MEC server. It uses the concept of loosely coupled services to provide flexibility in application deployment.

The traditional DNS-based service discovery mechanism is mainly oriented to the scenario where the service is static or the service address changes slowly. When the service changes, the DNS server usually takes a certain amount of time to complete the synchronization of the domain name service, which is not suitable for the large-scale and dynamic MEC scenario.

E69F-TI2-S In the scenario of vehicle networking, users and computing devices are dynamically registered and revoked, and services usually need to be migrated, which will lead to a large number of burst traffic, and the bandwidth of the WAN may not be dynamically adapted. How to support the rapid configuration of services from the device layer?

The edge device generates a lot of data, and the edge server provides a lot of services. How to dynamically schedule the data to the appropriate computing service provider according to the edge server and the network condition?

Q: Where does MEC’s business plan come from?

MEC In order to respond quickly, not only the location should be close to the data source, but also should be simple, for which MEC only has small data processing capabilities. Mathematical models and Al algorithms require global, long-period, big data training, which is not MEC’s strength, so MEC needs to obtain models and rules from outside. At the same time, MEC requires model slimming and limits model capabilities to specific subdomains.

E69F-TI2-S Currently, some companies open source development frameworks that connect cloud and iot devices, allowing applications with machine learning capabilities to be developed directly on the device or in MEC. For example, Microsoft has open-source Azure loT Edge Runtime, and DJI uses this framework to achieve local image recognition functions of drones. Qualcomm also uses the platform to quickly build image processing capabilities on mobile devices.

Q: How is computing power configured in the cloud?

Migrating terminal computing tasks to the cloud. With 5G transmission, terminals do not need to download and install core software, reducing hardware requirements and costs, plug and play, and smooth experience; The data perceived by the Internet of Things is directly uploaded to the cloud through 5G low latency, IOT and Al are seamlessly integrated, AI+IOT=AlOT (Smart Internet).

E69F-TI2-S Cloud computing power is diverted to edge computing and terminals. The edge computing of the industrial Internet is set up in the workshop or branch factory; The iot module is embedded in the Al chip and the operating system, which is equivalent to the edge computing power sinking to the iot terminal.

Q: How does the cloud edge computing capability work together

Mecs need to be connected and coordinated. Multiple MECs managed by the same central cloud may have uneven workloads. Can the capabilities of the idle MEC be invoked to support the busy MEC? It is not easy to cooperate storage computing resources between MECs, and the central cloud and MEC not only need to collect data and transfer models, but also need the central cloud’s resource management capabilities for large-scale MECs.

MEC needs regular backups to a secure remote location or central cloud to avoid data loss in the event of a local disaster. In addition, MEC’s billing and security management may also need to be part of the central cloud management.

Q: Does the MEC have a formula?

The location of the MEC is flexible, and the functions of MEC can even include the functions of CDN and micro DC, which can be small enough to fall on the terminal. The key is to enhance the MEC function at the same time need to pay attention to weight loss, otherwise the cost is uncontrollable.

In many scenarios, MEC may not have a separate physical existence and is integrated into related devices in the form of application layer software. If MEC stands for computing power, computing will be everywhere in the future, which means that MEC will not be visible in the future.

Communication network function concentration and function distribution iteration, it is best to concentrate the distribution of the two appropriate, rare is the optimal distribution of functions, but this is difficult to do.

At present, it can be said that there is no formula for edge computing, and it can be said that because of this, there is a lot of room for innovation.