Is Edge Computing a New Air Outlet?
Technical challenges and business rules are "mountains" in front of edge computing.
On February 27, 2019, Horizon Robotics announced that it had obtained an investment of about 600 million dollars The round B financing jointly led by SK Hynix and several Chinese first tier automobile groups (together with their funds) has been valued at US $3 billion after this round of financing.
Horizon said in its release that it will continue to move towards the vision of "becoming a global leader in edge AI chips and computing platforms". This 300 word ambition involving "edge computing" has attracted more than 20000 people's attention. The Baidu search index with the keyword "edge computing" jumped to 15032 on March 6, while it hovered around 1000 before.
Since the implementation of application pilots in security monitoring, smart city, smart home and other industries in 2015, edge computing has finally caught fire. Wang Chuannan, senior researcher of Tencent Youtu AI application research, explained that the first 10 years were a period of rapid development of cloud computing. Through cloud computing, local computing resources were cloud based, and on-demand use and on-demand payment were achieved. However, with the popularization of cloud computing and the cloudization of scenarios, problems such as high latency, huge data volume, and insufficient bandwidth have emerged, so edge computing has emerged as the times require.
The emergence of edge computing can advance computing and decision-making, obtain the most original data, analyze and make decisions on the edge nodes in real time, and transmit the filtered data to the cloud to form a synergistic effect with the cloud. In short, edge computing is a computing model that performs computing tasks at the edge of the network. Compared with cloud computing model, it can respond to user needs more quickly, reliably and energy-efficient.
Heating input
Wang Chuannan said, "Although the edge is still in its infancy, Hong Kong server However, it is obvious that a large number of companies are making additional investment in edge product research and development and programs, and edge products are constantly emerging in different scenarios. "
In addition to Horizon, major companies have already deployed edge computing. In the field of Internet of Things, Tencent Cloud released its IoT Suite, an Internet of Things platform based on edge computing. Alibaba Cloud's ability to expand on the edge is the Internet of Things edge computing platform Link IoT Edge. In the field of AI application, companies such as Broadview, Shangtang and iFLYTEK have launched AI+hardware products to enable traditional hardware through AI. Real time calculation is more needed in the direction of Internet of Vehicles and automatic driving to reduce reaction delay. Real time intelligent analysis and decision calculation in the local vehicle has become a standard configuration. Tesla has developed AI chips for automatic driving scenarios.
A related practitioner said frankly that, generally speaking, as long as chip related companies are concerned about edge computing 。 Yang Ge, founder of Xinghan Capital, also said that in terms of investment direction this year, the field of Internet of Things will be one of its key focuses 。 Edge computing is the core of the Internet of Things. "The Internet of Things needs a certain amount of intelligence at each end, which is edge computing," Zhou Xiang, vice president of the Yuncong Institute of Science and Technology, told Geek Park.
When AI became the outlet, according to the statistical data of "Zero2IPO", there were 112 companies obtaining financing from 2016 to the beginning of 2017 alone. When the UAV becomes the air outlet, Wang Mengqiu, founder of Hover Camera, said that every investment institution would like to add drones to its portfolio. Yang Ge believes that with the development of the Internet of Things, the number of start-ups related to edge computing will definitely increase this year, which is inevitable.
Wang Chuannan, who is technically oriented, does not believe that edge computing will become an overheated outlet like AI: "Edge computing in cloud edge collaboration is essentially a technical solution, which is to solve problems encountered in the actual scene. Whether it is" overheated "depends on whether there is demand. Cloud computing provides infrastructure for the development of the Internet of Things and smart cities, but it still cannot meet the higher diversified requirements of low latency, big data, and low bandwidth. Edge computing can effectively solve these problems, thus vigorously promoting the development of edge computing. "
Gao Tianyao, vice president of investment of Lenovo Star, once commented on the "new outlet", saying that the capital market never lacked new hype concepts, such as the Internet of Things, Internet finance O2O、 Sharing economy and so on, wave after wave. "Maybe the original intention is good, but excessive exaggeration will lead to some bubbles, so that entrepreneurs or capital that should not have appeared will enter a certain field."
The shackles of moving forward
China Economic Information Service mentioned in the 2017-2018 Annual Report on the Development of the Internet of Things in China released in September 2018 that the number of Internet of Things devices reached 8.4 billion in 2017, exceeding the global population. The following is the rapid growth of data. Statista, a data statistics company, predicts that there will be about 31 billion Internet of Things devices connected in 2020.
Under such premise, whether edge computing can become the next outlet or not, it is inevitable to solve some problems existing in edge computing and accelerate the construction of cloud network integration.
From the technical point of view, Wang Chuannan believes that the current edge computing faces four challenges: algorithm, architecture, operation and maintenance, and security: the algorithm side needs to optimize the network model for the edge node's limited computing power; The engineering side needs to design a set of AI multi terminal unified architecture for different edge device types; Edge node devices are deployed dispersedly. How to manage tens of thousands of scattered terminal devices also poses new challenges to the way and efficiency of operation and maintenance; At the same time, because edge devices are closer to data sources, the types and quantities of data have soared, and the emerging attack methods, especially attacks against physical devices, have brought new challenges to the security of devices and data.
Professor Shi Weisong of Wayne State University in the United States proposed six challenges facing edge computing in 2016: programmability, naming, data abstraction, service management, privacy and security, and performance index optimization.
"For ordinary users, security is the first concern," said a related practitioner. Before that, Amazon Echo had engaged users in conversations for violating the rules, which caused global discussion on privacy rights in the Internet of Things.
The edge computing model can complete a part of data processing at the edge of the network. Compared with the cloud computing model, it can prevent user privacy from being stolen on the transmission link or the cloud computing center. However, as Wang Chuannan said, it is not difficult for attackers to steal information from Internet of Things devices at present, and the highly dynamic nature of the network edge also increases the vulnerability of the network.
Fortunately, some academic papers pointed out that the academic community and the industry have made relevant achievements in the issues of programmability, naming, service management and privacy security. Relevant statistics show that ICDCS, the top conference in the computer field, included 6.7%, 9.3% and 12.7% of articles on edge computing in 2015, 2016 and 2017, respectively. Indeed, the academic community will point to better solutions as the research success increases year by year.
In addition to these problems, the implementation of edge computing is also limited by business rules. Yang Ge is very sorry. He thinks that for the same reason, the proportion of many academic achievements actually applied to business is very low. This may also explain why edge computing was first introduced in the fields of security and automatic driving. An investor said that the real development of edge computing also needs to carry the entire industry that can endow it with B-end capabilities, just like the "marriage" between the horizon of Hillhouse Investment and Belle.
In the financing announcement, Horizon finally wrote: "Empowering everything with artificial intelligence to make everyone's life safer and better.". At present, it seems that there is still a long way to go before these two visions finally become reality.
