Top ten trends of Internet of things technology re

  • Detail

Gartner released ten trends of IOT technology. The last kilometer of artificial intelligence is edge computing

Xinzhiyuan original

: Sanshi, Craig

[Xinzhiyuan guide] edge computing will provide AI capabilities for 10 billion terminals in the future, form an intelligent world of all things' perception, all things' interconnection and all things' intelligence, and open up the last kilometer of AI. In addition, the trend of "edge will eat the cloud" continues to appear, and enterprises with equipment, chips and algorithm capabilities will become the ultimate players of edge intelligence

ai's new opportunity lies in edge intelligence

on November 7, at the seminar and it exhibition held in Barcelona, Gartner, an IT research and market analysis organization, announced its most important strategy in the future - the technology trend of IOT. From now to 2023, IOT will promote digital business innovation

one of the important trends is the transformation from intelligent edge to intelligent lattice

in the IOT field, the transformation from centralized and cloud to architecture is progressing smoothly. However, this is not the end point, because the neat layer associated with the edge architecture will evolve into a more unstructured architecture, including various devices and services connected in a dynamic grid. These structures will realize a more flexible, intelligent and responsive IOT system

in May this year, Huawei released the white paper "giv2025 opens the smart world industrial map", which pointed out that by 2025, eight years later, the number of global iots will reach 100billion, and the number of global smart terminals will reach 40billion

40 billion intelligent terminals, edge computing will provide AI capabilities, and edge intelligence will become the support of intelligent devices. Human beings will be brought into an intelligent world of all things' perception, all things' interconnection and all things' intelligence by the fourth technological revolution based on ICT network and artificial intelligence as the engine

now, on the road of building the world, there are many players besides Huawei. This article makes a simple inventory

Gartner's top ten IOT trends: CIOs should pay attention to AI and chip innovation

Gartner believes that IOT will promote digital business innovation from now to 2023. The following are the 10 most strategic IOT technologies and trends listed by Gartner, which are expected to achieve new revenue sources and business models in the next five years

trend 1: artificial intelligence

trend 2: need to consider society Legal and moral IOT

trend 3: information communication and data agent

trend 4: the transformation from intelligent edge to intelligent lattice

trend 5: IOT with constant resolution throughout the process

trend 6: sensor innovation

trend 7: reliable hardware and operating systems

trend 8: new IOT user experience

trend 9: Silicon chip innovation

trend 10: new wireless networking technology of IOT

tner prediction, In 2019, 14.2 billion connected things will be used, and by 2021, the number will reach 25 billion. This process has generated a lot of data

AI will be applied to all kinds of IOT information, including video, still image, voice, network traffic activity and sensor data. Therefore, CIOs must establish an enterprise organization that makes full use of AI tools and skills in their IOT strategy

at present, most IOT endpoint devices use traditional processor chips, and low-power arm architecture is particularly popular. However, the traditional instruction set and memory architecture are not suitable for all the tasks that the endpoint needs to perform. For example, the performance of deep neural network (DNN) is usually limited by the memory bandwidth, rather than the processing capacity

by 2023, it is expected that the new dedicated chip will reduce the power consumption required to run DNN, and realize the new edge architecture and embedded DNN function in the low-power IOT endpoint. This will support new features such as data analysis integrated with sensors and speech recognition set up in low-cost battery powered devices. Gartner said that CIOs are advised to pay attention to this trend, because chips that support embedded AI and other functions will enable enterprises to develop highly innovative products and services

the three advantages of edge computing will "eat the cloud". Edge computing is the last kilometer of artificial intelligence.

edge computing refers to the computing done at or near the data source, which is different from relying on cloud computing in more than a dozen data centers to complete all work. Edge computing has three main characteristics: low latency, privacy security and flexibility

although enterprises continue to transfer data to the cloud for processing, with the gradual application of edge computing devices, localization management has become more and more common, and enterprises' demand for cloud may face bottlenecks

gartner group predicted in a 2017 report that "the edge will eat up the cloud"

Thomas Bittman, vice president of Gartner, pointed out that because people need to interact with their digital auxiliary devices in real time, it is impossible to wait for a data center miles (or tens of miles) away. Take Wal Mart as an example. Wal Mart retail applications will process data from store cameras or sensor networks locally, and the data delay caused by cloud computing is too slow for Wal Mart

since the development of artificial intelligence, in the vast fields it involves, the problems it still faces are the lack of excellent projects and the lack of scene landing

on the other hand, with the application of artificial intelligence in the edge computing platform and the increasing demand for the coordinated promotion of application landing of edge computing and IOT "end pipe cloud", edge intelligence has become a new form of edge computing, opening up the "last mile" of IOT application

at present, edge intelligence has been applied in the following fields

automatic driving

safety is the most important issue in the automotive industry. In high-speed driving, real-time performance is the primary premise to ensure safety. Due to the delay of network terminals, cloud computing cannot guarantee real-time performance. Vehicle terminal computing platform is the future of autopilot computing. In addition, with the development trend of electrification, low power consumption becomes more and more important for the automotive industry. ASIC chips that can naturally meet real-time and low power consumption will be the future development trend of vehicle computing platforms. At present, horizon robot and Mobileye are the main partners of OEM and Tier1

security, UAV

compared with traditional video surveillance, the main change of ai+ video surveillance is to change passive monitoring into active analysis and early warning. Therefore, it solves the problem that massive monitoring data needs to be processed manually (it also bypasses the problem of hard disk chain dropping at the critical moment). Security, unmanned aerial vehicles and other terminal equipment have high requirements for computing power and cost. With the development of image recognition and hardware technology, the conditions for intelligent security at the terminal are becoming more and more mature. Hikvision, the leader of the security industry, and Dajiang, the leader of unmanned aerial vehicles, have used the myriad series chips of movideous on smart cameras

consumer electronics

Huawei mate10 equipped with Kirin 970 chip and iphonex also embedded with AI chip have led into the intelligent era. In addition, Amazon's echo detonated the smart home market. For the consumer electronics industry, including home electronics, the premise of realizing intelligence is to solve the problems of power consumption, security and privacy. According to market research, smart consumer electronics such as smart appliances, smart, ar/vr devices equipped with ASIC chips have been on the eve of the explosion

the industry is divided into three parts, and those who have terminals, algorithms and computing power take all.

at present, the ecological architecture of the edge intelligent industry has been formed, and there are mainly three types of players:

class I: algorithm players. Cut in from the algorithm, such as providing computer vision algorithm, NLP algorithm, etc

Shangtang technology and Kuangshi technology are relatively successful companies in China that started with algorithms. On October 20 last year, Shangtang technology and Qualcomm announced that they would launch a "algorithm + hardware" cooperation to integrate Shangtang technology's machine learning industry into Qualcomm's chip products for mobile terminals and IOT devices, bringing better edge computing capabilities to terminal devices. In order to meet the needs of different degrees in the actual combat scene, Kuangshi technology is also continuously optimizing the algorithm to meet the requirements of edge computing

the second category: terminal players. Start with hardware, such as intelligent hardware such as, PC, etc

Hikvision, with many terminal devices, has been deeply engaged in the security field for many years, and is a provider of IOT solutions with video as the core. In the process of its development, edge computing and cloud computing are integrated to better solve the practical problems of IOT

the third category: arithmetic players. Start with terminal chips, such as developing AI chips for edge computing

for the field of edge computing chips, Huawei released the shengteng series chip - shengteng 310, which is oriented to edge computing products at the full connection conference in October this year. "Edge computing" will become the core growth point of Pan network, forming the future strategic direction of parallel Huawei terminals; Upgrade pipeline services with AI to help customers build a new ICT infrastructure in the intelligent era

following the pace of Huawei, bitcontinental also released the first low-power edge AI chip bm1880 and a series of products in October, entering the edge terminal market

Jianan Yunzhi is also a domestic technology company focusing on the research and development of high-speed, low-power chips. Jianan Yunzhi released the mass-produced edge computing chip KPU in September, which is mainly used in intelligent terminals, mobile terminals, speech recognition, image processing and other fields. Both companies have entered the edge smart market with the help of the latest EI chips

internationally, Google cloud has launched a lightweight version of TPU - edge TPU, which is used for edge computing and is open to businesses. Amazon, Google's rival to the cloud, was also exposed at the beginning of the year to develop AI chips, mainly to support Amazon's echo and other mobile devices

however, participants who occupy a single category are not the ultimate players. Edge intelligence requires enterprises to have the capabilities of terminal devices, algorithms and chips at the same time. In addition to large manufacturers such as Huawei, new enterprises add extension plans after preloading); Small ant technology in the industry should be regarded as the seed player of the "ultimate player" in edge intelligence

based on its innate advantage of massive data, Xiaoyi technology carries out a large number of training to maximize the accuracy of the algorithm, and then embeds the intelligent chip into the intelligent device to form a closed loop of data, algorithm and computing power. In June this year, Xiaoyi technology officially released three application scenarios of edge Intelligence: smart home, smart travel and smart retail. Using EI to reconstruct computing power, algorithm and business intelligence

Da Shengwei, CEO of Xiaoyi Technology: "the advantage of edge computing is the new Moore's law"

Xiaoyi technology should be regarded as one of the few domestic new enterprises that can achieve "Trinity" and deeply cultivate edge intelligence. Recently, Xinzhiyuan conducted an exclusive interview with Da Shengwei, the founder and CEO of Xiaoyi technology, and Kong Huawei, the partner. The following is the essence of the interview

Da Shengwei, founder and CEO of Xiaoyi technology

Xinzhiyuan: Xiaoyi technology now focuses on edge intelligence. What is the basis for your assertion that "the advantage of edge computing is the new Moore's law" at the Xinzhiyuan AI world2018 AI summit

Da Shengwei: in the field of video, Xiaoyi technology, from the earliest 1million pixels to 20million pixels, mainly overcomes whether humans can see clearly in the process of pixel upgrading. The new Moore's law is mainly reflected in the acquisition and compression of video

now video processing is completely two quadrants,

Copyright © 2011 JIN SHI