Abstract: As a young white person in the science and technology circle, I am often confused by data. Where does data come from? Where to? What can you do? How do you tell who’s useful and who’s not? In today’s article, xiaobian will take you to check the data acquisition technology of the water meter.
As a young white person in the technology circle, I am often confused by data. Where does data come from? Where to? What can you do? How do you tell who’s useful and who’s not? In fact, data is all around us. Everyone is a data producer. Data has been with us since the birth of human civilization. In today’s article, xiaobian will take you to check the data acquisition technology of the water meter.
Where does the data come from
Data is acquired through acquisition, but the acquisition process is not simple. For example, some data grains are coarse grain and some are fine grain, and the probability of obtaining fine grain is too low. In order to screen out valid data, the common practice is to bring back roughage before processing, which results in less than 10 percent of the grain that is extracted from the process.
In addition, not all data is readily available. In some cases, it is necessary to be self-sufficient in planting the land. It is up to the nature to decide which seedling to start cultivating their own data, and which link goes wrong to cause the poor harvest of data.
Even if we make our own food and clothing, there is also the problem of inconsistent seedling growth. The data is very subjective, and we often grow according to our own ideas. Everyone is different, so the taste is not consistent.
Where does the data go
The data hides the operation rules of the system or equipment, also contains sudden state change information, and even hidden subtle clues before the occurrence of anomalies or disasters. Therefore, it has a very wide range of USES, the traditional industry data implementation system can collect production equipment running status monitoring and operational work, all kinds of data in the information industry is the cornerstone of supporting the whole industry development, in the future, the mass is applied to the artificial intelligence technology in all walks of life, but also inseparable from the abundant data support.
Since we take food to compare data, so data for artificial intelligence is to eat into the stomach. The difficulty factor of data collection is not low. How can we get data more easily and solve the food and clothing problem of artificial intelligence more quickly? Smart humans are starting to think about how to get data for themselves quickly.
First of all, we can get a high customization unified harvesting systems, with a flat machine data of grain harvest efficiency is very high, but it has also led to the final data of grain, while united, no surprise, after all, harvester height is so high, this crop is harvested, some uneven data continued to get away with rough growth cannot harvest. As you can see from the curve below, a high degree of customization leads to a minimalist experience, but at the expense of flexibility.
How to choose between minimalism and flexibility, still have to land in the scene. According to the 2-8 principle, we harvest 80% of the data using high-end customized mode, and harvest the remaining 20% manually, so that we can efficiently obtain uniform food and also mix some small surprises, so as to achieve the best of both.
Let’s take the network ARTIFICIAL intelligence business as an example to see the specific operation:
More than 80% of the data used in network artificial intelligence services are equipment data of telecommunication networks, so this part can be directly connected to the network, and the system automatically completes docking negotiation, data collection and standardized processing. Users can directly obtain the desired cell performance indicators and enter the next business link.
80% of data collection scenarios are minimalist, ensuring the experience of most users. For the remaining 20% of scenarios, you can use flexible and universal collection capabilities. You can configure data source interconnection parameters to complete data collection.
There are also data that cannot be simply collected, such as passive devices and software and hardware systems that cannot generate state data, requiring a wave of advanced operations — probe autonomous acquisition. The probe technology itself is relatively mature and widely used in various industries, but its disadvantages are high deployment cost and difficult popularization. As can be seen from the name, soft probe technology is to collect network experience data through independent executable software or SDK that can be integrated under user authorization. Hard probe is basically a sensor, detection equipment, professional strong, high accuracy of data collection.
What can data do
There are so many things that data can do. It can be used to analyze user preferences and needs, obtain real and objective feedback on the use of telecom networks, quickly learn the shortcomings of products, update more business models, and help to achieve business improvement. Good data can help companies gain greater competitiveness. But behind these data are people. The more subjective people are, the clearer the demand will be, and the more helpful it will be to the improvement of product capability. It’s like the top end of the food spectrum, the taste is great, the quality is great, but the price is expensive.
As shown below:
It is difficult to obtain experience data, which is basically obtained through questionnaire survey, experiment, and end-to-end experience index collection, etc., which not only costs a lot but also has a limited amount of available data. It’s like if you’re an online seller and you want to get five-star reviews or real reviews of more than 10 words for your product, you need to give customers a good review rebate red envelope. In view of the fact that everyone is busy and there are still few customers who give word evaluations, interpolation is needed to estimate the overall feelings of all users based on a few real evaluations.
How do we get high quality data in the network AI business? First of all, we have developed a professional APP, for users, in addition to perceive their own network at any time the speed and time delay, also can help the user signal simulation for indoor coverage, realize the wi-fi network planning, really solve the problem of the user experience, in addition, has the sense of the game and use experience, better able to attract users, Collect network experience data. Second, we also provide can be based on limited data, by the ability of data interpolation algorithm, for lack of sampling points or less area for data, this method introduces the geographical statistical computing algorithms, under the certain condition of sampling, RMSE of interpolation error is less than 5, higher than that of the physical simulation results of the measuring device.
Protecting data security is a serious matter
As we mentioned at the beginning, data collection has many difficulties and requires a more secure environment for collection, storage and transmission. Only by ensuring the safety and reliability of data acquisition system can the rights and interests of data demanders and providers be effectively protected. Network artificial intelligence has also done a lot of targeted measures in data acquisition security. Such as data desensitization in the data collection link, minimum collection range, secret level control, etc., to avoid “stealing”; In the process of transmission also improve the transport brigade of ammunition reserves and concealed ability, for the robbers can not find their own data, or meet the hijacking son can calmly fight to defend data security; In the storage link access control and permission isolation technology, to avoid unauthorized access and data leakage problems.
The measures taken by network ARTIFICIAL intelligence in data acquisition security are shown in the figure below:
With the constant improvement of data security and privacy protection laws and regulations, deal with data security software and hardware measures are also gradually improve, each of the security measures are constantly improve project for a long time, there is no end in data security, only advancing with The Times progress, can continue to meet people for data security, thus more fully play to the data value.
Fifth, data collection technology continues to grow
Data collection is not accomplished overnight. With the iterative upgrading of products or services, the development of technologies in various industries, the improvement of user needs, and the improvement of policies and regulations, data collection technology needs to evolve and develop in terms of security and privacy protection, data availability and experience. Data acquisition is not only the basis of digital transformation, but also the spiritual food of artificial intelligence. It is more necessary to ensure the unimpeded flow of data channels.
As an annual event of Huawei ICT Infrastructure business for developers around the world, Huawei Developer Conference 2021 (Cloud) will be held in Shenzhen on April 24-26, 2021. With the theme of “Every Developer is Awesome”, this conference will bring together industry leaders, Huawei scientists, top technical experts, talented young people and many developers to discuss and share the in-depth innovation and application of the latest ICT technologies in the industry, such as cloud, computing and artificial intelligence. In the age of intelligence, every developer is creating an indomitable pentium. The world has you, great!
Click the link to learn more about the conference. Developer.huaweicloud.com/HDC.Cloud20…
Click to follow, the first time to learn about Huawei cloud fresh technology ~