On July 10th, the "Digital Intelligence Empowerment driving the future" innovation forum was held in Shanghai. Zhang Ping, academician of the Chinese Academy of Engineering, delivered a keynote speech on "AI Enables Future Mobile Communication Systems".
Zhang Ping said: "As we all know, artificial intelligence is developing rapidly, and it is experiencing the third wave triggered by deep learning, making great progress in four aspects: data, algorithms, computing power and terminal platform. ”
First, it provides strong learning capabilities based on big data. The advantages of mobile communication big data are large data volume, comprehensive audience, and objective analysis results; the disadvantage is that the data types are unstructured and highly real-time. These characteristics lead to the inefficiency and inability of traditional data analysis and processing methods when dealing with big data. The advantage of AI is that it can automatically extract advanced features from data with complex structures and internal correlations, greatly reducing expensive labor costs. At the same time, the larger the amount of data, the stronger the learning ability of AI.
Second, AI algorithms improve the actual effect of communication models. Communication system designs typically take a modular optimization approach. However, there is non-Gaussian or nonlinear noise and distortion in the actual wireless channel state, so there will always be an error between the theoretical value and the real data. AI can theoretically learn any hidden structure and hidden parameters to fit arbitrarily complex functions, which provides a new means for the perception of the wireless channel environment and the portrayal of network state space. At the same time, algorithms such as deep learning can solve difficult problems in communication systems that cannot be modeled or solved by traditional methods. This provides new ideas for the design of a new end-to-end architecture for future mobile communication networks.
Third, the development of distributed parallel computing and cloud computing has brought powerful computing power. The computing power used in AI training tasks is growing exponentially, and parallel computing based on graphics processing units (GPUs) enables deep learning to process massive amounts of data in milliseconds. Cloud computing services enable the public to obtain computing resources on demand anytime, anywhere with the help of the Internet, realizing the intensification and large-scale application of computing power. The development of cloud-network integration has brought incomparably powerful computing power to terminals. The development of distributed parallel processing technology and cloud computing has continuously improved computing power, which helps to improve the accuracy and timeliness of mobile communication network analysis and management, and provides strong software and hardware support for the application of artificial intelligence in mobile communication.
Finally, the edge network and the terminal platform are intelligent. 5G networks introduce edge computing architectures, and edge devices include smart devices of different magnitudes and computing nodes on the end side, which will have the ability to learn. Intelligent terminal devices no longer simply refer to terminal devices in the traditional sense such as smart phones, and various products such as automobiles, home products, and robot agents will become terminal devices with intelligent functions in the era of interconnection of all things. The combination of mobile communications and artificial intelligence will drive a new round of industrial revolution, and eventually completely change the production mode and business model of all walks of life.
Author: Zhen Qinglan
Editor-in-Charge/Layout: Fan Fan
Proofreader: Shu Wenqiong
Review: Shen Qing
Producer: Liu Qicheng