Smart home will also use various other new technologies, including essential communication technologies, such as wireless technologies Zigbee, Z-Wave, RF, Bluetooth, WIFI, EnOcean, UWB, NB-IoT, Lora, etc., and wired technologies RS485, RS232, Modbus, KNX, etc. Zigbee is a short-range, low-power wireless communication technology featuring proximity, low complexity, self-organization, low power consumption, and low data rate.

Zigbee is a short-range, low-power wireless communication technology, characterized by proximity, low complexity, self-organization, low power consumption, and low data rate. Zigbee has a transmission distance of 50 meters to 300 meters, a rate of 250kbps, a power consumption of 5mA, and a maximum number of 65,000 network nodes.

Z-Wave is also a low-cost, low-power, highly reliable RF-based short-range wireless communication technology. In terms of technology, Z-Wave has increased from 9.6Kbit/s to 40Kbit/s. In terms of nodes, a Z-Wave network can support 232 nodes.

RF is a non-contact automatic identification technology. Compared with the traditional magnetic card and IC card technology, it has the characteristics of non-contact, fast reading speed, no wear and tear, no environmental impact, long life, easy to use, and has the function of anti-conflict, and can handle multiple cards at the same time. Wireless RF technology provides non-contact two-way data transmission between reader and RF card to achieve the purpose of target identification and data exchange.

Bluetooth is a wireless technology standard that enables short-range data exchange between fixed devices, mobile devices and building personal area networks. Bluetooth has a transmission distance of 2 meters to 30 meters, a rate of 1Mbps, and power consumption between Zigbee and WIFI, and is mainly used in small smart hardware products.

WIFI is one of the most widely used wireless communication technologies, with a transmission distance of 100m to 300m, a rate of 300Mbps, and a power consumption of 10-50mA.

EnOcean is the world”s only wireless international standard using energy harvesting technology, and the EnOcean energy harvesting module can harvest energy from the surrounding environment and use it for EnOcean”s ultra-low power wireless communication module. Compared to similar technologies, it has the lowest power consumption, the longest transmission distance, can be networked and supports relay functions. It requires no batteries, and the power required for wireless signals is 1/30 to 1/100 of ZigBee, and is mainly used in wireless passive smart home and smart building products.

UWB uses nanosecond to microsecond non-sine wave narrow pulse to transmit data. By transmitting very low power signals over a wide spectrum, UWB can achieve data transmission rates of several hundred Mbit/s to several Gbit/s within a range of about 10 meters.

NB-IoT (Narrow Band Internet of Things) is built on a cellular network, consuming only about 180 kHz of bandwidth, and can be deployed directly in GSM networks, UMTS networks, LTE networks or 5G networks to reduce deployment costs and achieve smooth upgrades. NB-IoT can support 100,000 connections in one sector, supporting low latency sensitivity, ultra-low equipment cost, low equipment power consumption and optimized network architecture; third, lower power consumption, the standby time of NB-IoT terminal module can be as long as 10 years; fourth, lower power consumption. The fourth is lower module cost.
The LoRa network architecture consists of four parts: terminal node, gateway, web server and application server, and application data can be transmitted in both directions. Because of LoRa”s low power consumption, long transmission distance, flexible networking and other characteristics and the Internet of Things fragmentation, low cost, large connection needs are very suitable.

RS485 is also known as TIA-485-A, ANSI/TIA/EIA-485 or TIA/EIA-485. RS485 has two-wire system and four-wire system two wiring methods, four-wire system can only achieve point-to-point communication, rarely used, now more than two-wire system wiring method. This wiring method for the bus topology, in the same bus can be hooked up to a maximum of 32 nodes. Many mainstream wired smart home manufacturers are using the RS485 protocol.

RS232 is one of the communication interfaces on personal computers, developed by the Electronic Industries Association asynchronous transmission standard interface. Usually the interface to 9 pins or 25 pins type appears. There are generally two RS-232 interfaces on a personal computer, called COM1 and COM2, and a few smart home products use this protocol.

Modbus is an industrial communication system consisting of a programmable controller with intelligent terminals and a computer connected via a common line or a local dedicated line, supporting 247 remote slave controllers.

The KNX bus is a system independent of the manufacturer and the application area. The bus devices can be sensors or actuators for the control of building management devices such as: lighting, blinds/shades, security systems, energy management, heating, ventilation, air conditioning systems, signaling and monitoring systems, service interfaces and building control systems, large appliances, etc. large home appliances, etc. All these functions can be controlled, monitored and signaled through a unified system, without the need for additional control centers.

Status quo of smart home edge computing

(1) Smart home gateway

At present, the smart home mainly connects and controls the smart devices at home through the cloud platform, and the interaction of many devices in the home area network is also realized through the cloud platform. However, over-reliance on the cloud platform will bring many problems, such as network failure at home, and it is difficult to control the devices. In addition, the cloud platform to control the home devices, sometimes slow response speed, will bring a strong sense of delay, and with the increase of smart home single product categories this bad experience will become more and more frequent.

In the smart home, the application of edge computing is becoming more and more widespread. Smart home gateway is the heart of home intelligence, through which the system information collection, information input, information output, centralized control, remote control, joint control and other functions are realized, and it is an important carrier of edge computing.
On the one hand, with the support of edge computing, the smart home gateway can control the smart home devices directly through edge computing. For the smart components in the same gateway, the gateway can process the information received by these components and make decisions based on user settings or habits, and control the execution of the components to perform corresponding actions.

For the smart home components that can realize edge computing, they can be unaffected when the user”s external network is disconnected, which avoids the problem of paralysis of the smart home system when the user is disconnected.

On the other hand, in the interactive scenario of different products of smart home, edge computing will also act as a network management or central control system to realize the interconnection and interoperability between devices and scene control through the synergy of cloud computing and edge computing.

The smart home gateway connects the rich home and sensors in the family internally and the cloud platform externally. The smart home gateway provides infrastructure resources such as computing, storage, network, virtualization, etc., as well as lifecycle APIs for device configuration, monitoring, maintenance, optimization, etc. Edge computing requires support for plug-and-play, automatic multi-AP configuration connection, southbound management of peripherals such as cameras, access control, temperature and humidity sensors in the home, and rich southbound interfaces such as network cables, power cables, coaxial cables, Zigbee It can also process a large amount of heterogeneous data, and then upload the processed data to the cloud platform in a unified manner.

Users can not only connect to the edge computing nodes through the network and control the home smart terminals, but also realize the whole life cycle management of the edge home gateway by accessing the cloud platform, including network big data analysis to provide relevant resource optimization decision recommendations and realize the manageable operation of the home network.

Users can choose intelligent active services for family life, or customize their own home business scheduling, which is unified by the intelligent gateway for management and control. Under the authorization of the user, the intelligent home system of Cloud Edge Synergy will actively learn the user”s living habits and continuously optimize the intelligent model to better provide intelligent services to the user: the homes in the living room, bedroom, bathroom and kitchen will automatically and intelligently help the user to complete a series of family affairs according to the user”s customized needs, combining different scenes such as waking up in the morning, leaving home, returning home in the afternoon, relaxing and going to sleep. In the morning, the bedroom can automatically wake up with music, open windows, prepare toiletries automatically in the bathroom, and start baking bread automatically in the kitchen; when leaving home, it will automatically turn off air conditioning and other electrical appliances, and choose whether to close windows according to the weather; when returning home in the afternoon, it will open the door safely and activate the home mode, automatically turn on the lights, recommend recipes, teach cooking, and automatically place orders for daily needs; after dinner, the robot automatically cleans the bathroom, automatically turns on the After dinner, the robot automatically cleans the bathroom, automatically turns on the TV and switches to the corresponding program, etc.; before going to bed, it automatically checks and closes the doors, windows and appliances [2].

(2) Artificial intelligence applications

The deep combination of edge computing and artificial intelligence can realize smart home edge intelligence, from comprehensive voice control, to spatial physical control and visual control, to the final realization of the system after self-learning senseless experience.

(A) Voice control technology makes life smarter, and voice makes interaction more convenient and fast. No matter how advanced the brand and technology, how friendly the human-computer interaction interface is, it is not as convenient and direct as voice interaction control. Apple smartphone as the representative of Siri took the lead in the concept of voice control, and then the rapid development of intelligent voice control technology, Amazon Echo smart speaker device became the first devices in addition to cell phones, so that the role of voice assistant to control smart home devices. The Internet giants represented by Google, Microsoft, etc. also followed up to explore the application of voice control technology in intelligent home devices. General smart home products (such as lights, TVs, music players and other devices) can basically achieve their functions through voice control.

However, voice control still faces many challenges, for example, a necessary condition for users to be able to control smart home by voice at will is that the device can accurately recognize commands no matter where you are in the living room, and voice recognition technology must break through the distance barrier. At present, indoor voice interaction is affected by multiple complex factors such as background noise, other people”s voice interference, echo, and reverberation, resulting in low or no recognition rate, and can only be used in a relatively quiet and proximate environment.

In addition, the Chinese accent and language are very profound, and the percentage of people in China who have mastered standard Mandarin is relatively low, so the accent phenomenon is complicated, and there are even different accents in the same city. Therefore, when these people with light or heavy accents use voice input, if the usual method of using standard Mandarin data for model training is used, serious adaptation problems will arise, which will affect the recognition effect of voice input.

Therefore, smart home language control must be more intelligent, only recognize some basic words and phrases is obviously far from enough, they must be more understanding of human. They must understand human accent, speed of speech, dialects, pet phrases and professional vocabulary that pops up from time to time. This is the smart home language control to achieve personalized recognition. That is, the voice recognition system has the ability to automatically learn and adapt to the user”s habits, the more you use, the more it understands you. Generally speaking, personalized identification includes two aspects of pronunciation and language. The personalization of pronunciation mainly refers to the system”s learning of user”s accent, speech speed and other pronunciation habits, while the personalization of language mainly refers to the system”s better recognition of user”s specific vocabulary (e.g., personal names, place names, pet phrases, professional words, etc.).

(B) Physical and visual control Physical control refers to a practice that allows people to directly interact with the surrounding devices or environment using body movements, without using any complex control devices, so that people can interact with the content in an immersive way. For example, if a person is standing in front of a TV set and there is a somatosensory device that can detect the movement of the person”s hand, then if the person swings his or her hand up, down, left and right, it can be used to control the TV set”s fast-forward, rewind, pause and stop functions.

According to the different methods and principles of body sensing, can be divided into three main categories: inertial sensing, optical sensing, and inertial and optical combined sensing. Inertial sensing is mainly inertial sensors, such as gravity sensors, gyroscopes and magnetic sensors to sense the physical parameters of the user”s body movements, respectively, acceleration, angular velocity and magnetic field, and then according to these physical parameters to obtain a variety of user actions in space.

Optical sensing uses lasers and cameras (RGB) to obtain image information of the human body and capture 3D whole-body images and movements of the human body. Common 2D vision has primary applications such as video calls, network cameras, and 2D recognition. 2D recognition only allows the terminal to “see” and is not sufficient for intelligent obstacle avoidance and recognition. For example, equipped with 2D vision sweeper robot is not smart enough to identify the front is garbage or obstacles, resulting in unsatisfactory cleaning effect and route planning. With the development of 3D vision technology, breaking the limitations of the previous 2D vision. 3D vision technology can obtain depth information, capture the object 3D data, empowering intelligent terminal live detection identification, gesture recognition and other capabilities to achieve “understanding” of the world.

3D vision applied to the door lock, compared to the traditional fingerprint-based identification and 2D identification door lock is more reliable. Of course, security and control is only the most basic level of smart home applications, with the future of the smart family on the terminal intelligence of the increasingly high demand, 3D vision technology can also be applied to a deeper level of family life. 3D vision-generated skeletal recognition technology can capture the movement of people to achieve 3D physical games, 3D physical fitness and other quality of life; combined with AR to help parents achieve a more immersive child The combination of AR helps parents to achieve more immersive early childhood education, truly educate and entertain [3].

(C) Senseless experience The real smart home should be automatically adjusted according to the user”s living habits to provide a senseless experience, that is, the smart home is integrated into your life and mine, but will not cause interference. For example, smart air conditioners can monitor the temperature, humidity, light and environmental changes around the thermostat through built-in sensors, determine whether there are people in the room, record the user”s temperature adjustment habits, and thus decide whether to turn on the temperature adjustment device. Not the user to tell it should be adjusted to how much temperature, but so that the user can not afford to adjust the temperature, because the temperature has been automatically adjusted by the intelligent air conditioning to help users, the user simply do not realize what needs to be adjusted. Another example is the smart refrigerator, to be able to tell the user what day a certain kind of food is put in, how much is inside, when will expire, with ideas and data analysis capabilities. This kind of smart home “thought” is through the data analysis of user needs, and find ways to meet these needs.

To achieve a senseless experience, on the one hand, we need to make use of the multimodal fusion application of vision temperature smell body voice, etc., and on the other hand