Composition of IOT platform
IOT platform is based on the Internet and constructed through existing communication technology, without relying on specific hardware modules. Users can finally access the Internet of things based on their own equipment technology architecture.
It mainly connects hardware, handles different communication protocols, provides security and authentication for devices and users, collects data and makes visual analysis, and integrates with other web services.
A complete IOT platform includes four parts: application enabling platform, device management platform, connection management platform and IOT terminal agent. Other functional modules are extended based on these four functional modules. The following is the specific content of the core function module:
Application enablement platform (AEP), whose core competencies are business orchestration, big data analysis, industry suite and API management, and the mainstream camps are thingworx, cumulity IOT, etc;
Device management platform (DMP), whose core capabilities are device authentication, data collection, Internet of things protocol adaptation and subscription notification; The non core capabilities are firmware upgrade and alarm management, and the mainstream camps are Bosch BSI, digi, Nokia impact, Sierra wireless, etc; There are two modes of operation, namely, the platform service is mainly provided by integration, and a complete end-to-end solution is provided. The former, the general equipment management platform DMP, is integrated into the complete set of end-to-end M2M equipment management solutions and is not provided separately. Only a small number of manufacturers that provide equipment management cloud services independently charge a certain operation and management fee for each equipment every month; In the latter, mainstream DMP platform operators provide Internet of things connection devices, sensing products and management cloud platform services at the same time, so that they can provide complete end-to-end solutions at the same time.
Connection management platform (CMP), whose core capability is to connect SIM card to operator network and realize the docking of operator billing system; The mainstream camps are Cisco Jasper, Ericsson DCP, Vodafone GDSP, etc;
The core capability of IOT agent is to enable the pre integrated equipment to quickly access the platform to realize near-field control and edge computing.
Closed loop architecture of IOT system
Throw a brick to attract jade: at 2 p.m. on a weekend in the future, the sun is shining brightly. It shines in from the window and the breeze blows my face. I lie down leisurely on the sofa in the living room and shout: “Xiaosha, Xiaosha, play me a pure hypnotic music and massage my back for 10 minutes”. The sofa did as I said. In addition to doing what I said, it also analyzed my health by contacting my skin, and then told the information to the home air conditioner, which automatically adjusted my most suitable humidity and temperature. When I took a nap and woke up at more than 3 o’clock, I wanted to take a shower, so I told the sofa to help me heat the water. Of course, the sofa analyzed the hair quality by touching my hair and recommended which shampoo I should use. After taking a shower, I wanted to go out again, so I asked the robot servant at home to call a car for me… The whole life is very convenient.
What I said above has not been popularized, but some enterprises are already doing it. This system of networking and intelligence of all things is called IOT (Internet of things), which can also be called the Internet of things (including Zhilian). The ofo small yellow car that can be seen everywhere now is also an IOT mode. So, how to realize the complete closed loop of an IOT system?
In fact, the implementation principle is not difficult. It is nothing more than three parts: Hardware Product + software control + cloud. The so-called closed loop means that they can form a loop to realize a complete production product chain. The relationship diagram of these three parts is as follows:
The above figure reveals two IOT modes, one is cloud hardware off mode, and the other is cloud software hardware mode. Now common bike sharing belongs to cloud software hardware mode, and the software part is app. So, what are the specific details of these three parts? As a back-end developer, I am naturally responsible for the cloud. The cloud is also a heart / brain level role in IOT. How should the cloud be designed to support IOT scheduling?
I can’t give standard answers to these two questions because I’m also groping. Here I give a cloud solution. As for the hardware or software control system, an interface is provided for the cloud. The design schematic diagram of cloud is as follows:
Based on the architecture above, the functions of each part are as follows:
1. Store the data information of the third party, such as the product of an intelligent robot. The data of the robot should be collected to deal with various behaviors and give instructions
2. Storage user behavior system. Note that in the figure, there is a black solid line pointing from hardware / software to file storage. This part is designed for user behavior analysis. Traditional user behavior analysis (such as Youmeng and Alibaba datav) adopts the mode of sending behavior messages (buried points) to the server in real time. But I’m poor and can’t afford an online server dedicated to processing user data, because the configuration of this server is very high and the cost is too high! Therefore, I will store the user’s data of one day / week / month locally, and then upload it directly to the file storage every other period of time. The file storage can be the storage of a third party. Then I use my own computer to download data from file storage to local after work or at night, and run big data system analysis. The specific details will be introduced in the big data system.
1. MySQL Cluster, which stores logical data and provides it to various systems.
2 the database design should be combined with the terminal, especially the hardware. For example, if a UAV product wants to express it, the database design must be close to the product itself, and the fields include: chip (memory, CPU, resistance, capacitance), shape, public model and other parameters and indicators. Therefore, when designing the database, we should learn from the hardware department.
Big data system
1. Statistics of terminal behavior. The service object of IOT is the terminal, so it is necessary to make statistics and analysis of terminal equipment, such as the number of users, the global distribution law of users, etc. taking smart home as an example, if you can count the user’s usage habits, which temperature of air conditioner is used by the most people, which position of sofa is used by the most people, the work and rest time of users Health status and so on, which are helpful to the product design, point out the direction for the operation and the design of the next version, and recommend other related products by the way.
2 explore potential functional requirements
3 feedback to enterprises: when an enterprise accesses an IOT system, it naturally wants to see the real effect
4. As for my design, the non real-time method is adopted, which is very convenient in management. I am very clear about what data I want, so I don’t have to analyze all of them. The composition of data shall be combined with the actual functions of the terminal. Each event of the terminal shall be defined and sent to the file storage system with time and GPS. To distinguish different terminals, you only need to indicate it on the file, which can reduce the number of transmitted words. In this way, the statistical data will always be one day / week / month less than now, but the cost is greatly reduced, because it does not need an online server, just a local computer. I can work with my own computer during the day and run data analysis at night, 0 cost!!!
This should be combined with specific services. If it is facing the hardware terminal, send instructions to the terminal and process the designation of the terminal. If it is controlled by software, provide an interface.
2 the business system is not only business, but also applications such as nginx, cache and data layer. It just says that it directly faces the terminal and processes business.
Background management system
1. It is mainly open to enterprise users to customize the image, manage the terminal, and view the user behavior laws of their own products, such as what functions each instruction of a hardware terminal corresponds to, recommend other products, advertise, etc., which are specified by the background management system
1 is divided into two parts. The conventional part is the traditional network security, such as obtaining user token, data filtering, encryption, etc. it is mainly used in the transmission of background management system and business interface.
2. However, for the hardware terminal directly, the command transmission security is special. This mainly draws on the ID ^ 2 security model of Ali IOT, just like the smart home. Although it seems that the whole home system is unified, in fact, their work is independent and completely controlled by the cloud, which connects them together, but the terminal itself is not united, Therefore, each terminal independently receives instructions + transmission encryption. The sofa sends the user’s body information to the cloud for processing, calculates the temperature and humidity that the air conditioner should adjust, and sends this command to the air conditioner before the air conditioner can make adjustment.
Terminal (hardware, software)
1 if the hardware is directly connected to the cloud, the hardware must provide networking modules
2 if it is hardware software cloud, the connection part between software (generally APP) and hardware takes WiFi or Bluetooth as the carrier, and the protocol should also be defined first
The structure listed above seems quite simple, but once it is done, we know a lot to consider, such as hardware sound control. To distinguish different people’s voices, we need to do acoustic filtering. I’m ashamed to say that the knowledge used in acoustic filtering is from analog electronic circuits. It happens to be the first subject in my university, The data extraction of big data system should go through layer by layer statistics, and set primary data, secondary data and tertiary data. What is the basis for extraction? How to extract?, There are also log management, alarm system, fault tolerance… It’s hard to say. In short, I’m also an ant working hard on this road. Let’s work together!