After briefly introducing the application and origin of fog computing, the following will introduce an important technology of the Internet of Things-edge computing. The following will explain the origin of edge computing, and introduce its relationship with the Internet of Things, and take unmanned driving as a use case to introduce the shortcomings of cloud computing and the application of edge computing.
First, define edge computing (wikepedia, 20 19):
Many difficult technical terms mentioned here, such as "distributed computing" and "nodes", are actually just descriptions: edge technology is a technology that transfers some large-scale applications (that is, distributed computing) to everyday devices (that is, edge nodes) for processing.
In the typical structure of cloud computing (as shown above), it can usually be divided into three layers: "cloud (cloud layer)-network (fog layer)-end (edge)". The Terminal layer covers all terminal applications and is usually a managed role. As soon as the result is calculated in cloud computing, it will be executed by the application program that sends instructions to the end layer through the network layer. After receiving the data, the application program will send it to the cloud layer for calculation.
Edge computing can be imagined as giving the "end" layer a certain degree of "autonomy". In the architecture of edge computing, the endpoint is given simple storage and computing capabilities (unlike fog computing, which emphasizes "simple" functions), so that it can occasionally get rid of the management of the cloud and respond according to environmental data.
This may seem like a small step to increase the simple computing and access capabilities of the terminal system, but in fact, this layout has great benefits, including:
? -Low latency: data is generated by near field, which can respond quickly.
? -Independence: The system can run without network connection.
? -Compliance: Protect personal data without transmitting user data.
? -Simplify data: the terminal processes some data first, then simplifies the data, and then transmits it to the cloud server.
? -Security: Reduce data transmission and reduce network security risks.
Unmanned driving is one of the classic use cases of edge computing, and it is also a good example to explain the shortcomings of cloud computing and why edge computing is needed.
The following figure is a commonly used cloud computing architecture, including 1) a smart car (client) and using driverless function, 2) the Internet is used to transmit data, and 3) a cloud service (cloud computing) server is used to provide driverless service.
Suppose the car is driving at a speed of 60ms- 1, and an obstacle is perceived at the starting position of 3m ahead. Because the car uses the cloud computing architecture, and the car itself has no analysis function, the car will send the sensed image to the cloud server for analysis (step 1).
Unfortunately, because the car is far away in the north area, it takes 0.05s for information to stop in the cloud service, but it takes 0.05s for instructions to be sent to the car for execution (step 2).
In the process of sending information to the recovery instruction (~0. 1s), the car will continue to drive at a constant speed (60ms- 1s) and will not receive the stop instruction until 6m later (= 60ms- 1x 0. 1s). And it will hit a passerby 3 meters ahead, causing a car accident.
When the car senses an obstacle 3 meters ahead at the starting position, it will immediately execute the stop command (step 1). Then send the images and decision-making contents to the cloud server for advanced analysis (step 2) to improve the unmanned performance. (Note: This seems to be similar to the method of fog computing, but in this process, the application did not do any data analysis, but only responded according to the content of the sensor. If it is fog computing, it will send the sensor information to the fog service for further analysis, and then notify the terminal equipment to respond. )
It can be seen that the cloud server is far away from the location where the data is generated, so it will cause a big delay. Unmanned driving, which requires real-time decision-making, has a great opportunity to use edge computing to make computing services close to the source of data and computing closer to actual actions.
With the development of science and technology and the rapid increase of data transmission speed, many daily objects, such as household appliances and vehicles, are embedded with sensors, and exchange information with the Internet through network connection, forming a huge object network (that is, the Internet of Things).
When an object runs, it will collect a lot of environmental data. Some people may ask, why not process all the data locally and then transfer the other data to the cloud service for storage. This may be one of the methods that can be realized, but if all the data is processed locally, the object itself will be equipped with many storage devices and processing servers. This will greatly increase the power consumption and the weight of the object, and increase the cost.
So the best way is to combine the advantages of cloud computing and edge computing to make the best configuration. In some events that determine the major safety of objects (such as braking in the unmanned example above), the decision-making power can be put on the edge, and other non-urgent things can be concentrated on the cloud server at low cost. Through the good division of labor between cloud and edge, the cost can be greatly reduced and the calculation efficiency can be improved.