Zhejiang Lab

Terminal-Edge-Cloud Collaboration: The Secret to Faster Computing in the Era of 5G-Data Explosion

As the digital economy has become a global trend marking a new era of economic revolution, novel technologies such as artificial intelligence (AI), the Internet of Things (IoT), and 5G have been facilitating economic and social development through integrated tactics on the perception, transmission, and computation of data. In this era of data explosion, what we need is ubiquitous computing that can be performed anytime and anywhere, so that we can acquire the information we want.

Recently, Doctor Gao Feng, a senior research fellow at Zhejiang Lab’s Research Center for Intelligence Integration, gave a lecture entitled "Terminal-Edge-Cloud Collaboration – A Novel Computing Model" to colleagues at the Lab. He believes that edge computing can provide efficient and strong real-time data processing services, owing to its proximity to data sources. Therefore, a terminal-edge-cloud collaborative operating system that incorporates edge computing can provide efficient, stable, and safe computing support for ubiquitous computing through task deployment and resource management.

Terminal-Edge-Cloud Collaborative Operating Systems Based on Edge-Cloud Integration

IoT and 5G have brought massive connections and explosive data growth. Consequently, it is undeniable that cloud computing is not adequate for the computing needs required by the Internet of Everything, because it is performed far away from devices and users, taking a long time to process data.

Edge computing has thus been developed to address the challenge caused by insufficient computing power. Gao said, "Edge computing is a distributed computing paradigm. It transfers applications, data resources, and other related computing operations to edge nodes for processing." That is to say, edge computing takes place more closely to users, devices, and data sources, so that data can be processed and analyzed at the nearest location. It can therefore provide real-time and agile computing support for intelligent devices.

Does this mean that edge computing can replace cloud computing? Gao believes that, "The relationship between edge computing and cloud computing is complementary, rather than alternative." Cloud computing is more powerful in non-real-time and long-cycle data processing, while edge computing is more applicable to real-time and short-cycle data processing. With their complementary advantages and different strengths, they are good partners that can be used together to process the massive amounts of data that IoT creates.

Although edge computing demonstrates prominent advantages, it is undeniable that massive amounts of IoT data will cause tremendous pressure on bandwidth and data storage. As predicted by the research firm IHS, there will be 54 million driverless cars in the world by 2035. For every second, the sensors and cameras on these driverless cars will generate 1GB of traffic information. At that time, inefficient network architecture and inadequate computing power will become major obstacles to the development of digitization and intelligence.

Strong and Fast Intelligent Computing Facilitated by Terminal-Edge-Cloud Collaboration

All kinds of questions concerning intelligent computing power ultimately point to the same problem: that is, "How to provide efficient, safe, and user-friendly computing support for algorithms used in IoT and intelligent devices?"

Gao believes that the terminal-edge-cloud collaborative operating system is an ideal solution. "A terminal-edge-cloud system is a loosely coupled distributed system composed of multiple interconnected computing nodes on the device, at the edge, and in the cloud computing environment." That is to say, a terminal-edge-cloud system can leverage the complementary advantages of terminals, edges, and cloud computing centers, ensuring the efficient use and transmission of data through strategies such as task deployment and resource management strategies. In addition, the three sides can be dynamically adjusted and coordinated to realize task migration and balance the computation load among them.

Then, how does a terminal-edge-cloud system provide efficient and safe computing support for the IoT? Gao pointed out that, "First of all, we can set a task deployment model in the system and reasonably allocate resources such as data or tasks to terminal or edge devices. Secondly, components in the system can share infrastructure and wireless resources through technologies such as network virtualization and slicing, so as to largely improve the utilization efficiency of resources. Thirdly, in a terminal-edge-cloud environment, as computing can be performed on any of the three sides, data control and data ownership are separated from each other, which can ensure data security."

Efficient and Safe IoV Systems Based on Terminal-Edge-Cloud Collaboration

Since the invention of the automobile, concerns over driving safety have never stopped. Traffic accidents often occur within seconds, and any delay in response can be catastrophic. Therefore, an Internet of Vehicles (IoV) system should demonstrate low latency and high efficiency to ensure driving safety.

To create a more secure and stable driving environment, we can introduce the terminal-edge-cloud collaborative architecture into an IoV system. According to Gao, "An IoV based on terminal-edge-cloud collaboration can attain outstanding performance in aspects such as environmental perception, data exchange, task deployment, and content distribution."

Firstly, an IoV based on terminal-edge-cloud collaboration can collect and integrate different types of data through videos, millimeter-wave radars, on-board unit sensors, and other devices, so as to achieve the collection of heterogeneous and multi-source perception data. Secondly, the direct data exchange between the on-board units of the preceding and host vehicles can provide more efficient computing support for autonomous driving. In addition, as data processing tasks are transferred from on-board units to roadside units, which means the migration of computing tasks to edge devices, vehicles can greatly reduce their computing consumption and improve their battery efficiency. Lastly, high-capacity and high-precision maps that are originally distributed from the cloud can be distributed directly from roadside units. Vehicles can therefore save the time and improve the efficiency of data transmission. Also, traffic information is directly sent to on-board devices by roadside units closer to data sources, so that the autonomous driving system can operate in an environment with lower latency and higher accuracy.

The terminal-edge-cloud collaborative operating system developed by Zhejiang Lab’s Research Center for Intelligence Integration is an in-depth combination of new perception, transmission, and computing technologies. It is expected to achieve tech innovation in terms of technological architecture and platforms in fields such as intelligent transportation and smart cities, so as to better facilitate the development of the digital economy.