The transmission of information has been around since the very beginning of human birth. The transmission of information was initially through the babbling between people, the reception and processing of information from the eyes, ears and tongues, the information of food and survival, and the continuation of civilization after that. The spread of images and language has accelerated the rate of human co-evolution, and the footprints of human beings have stepped onto a wider territory.
More than two hundred years ago, the sudden evolution of human brain power, the sudden outbreak of technological civilization, the process of information dissemination and upgrading no longer relies on the footsteps, eyesight and hearing of thousands of years, but is dominated by invisible signal waves. From morse's invention of the cable telegraph in 1838, to the modulation radio broadcast in 1918, and to today, we cannot see the dissemination of information with the naked eye, but information flows in all corners of the world, covering the economy, production, life and other fields, breaking through the barriers of geographical, spatial and other restrictions, affecting everything.
In the era of intelligence, the information network has become very complex, the overall information system has the storage and calculation of information, the signal transmission rate will be affected by the channel width, noise size and signal power, there is a limit, to the transmission of information to bring problems, which also means that the need to explore the overall optimality from the perspective of the system.
The foundation of information theory - Shannon information theory has built the foundation of the modern information building and established a basic theoretical system of information measurement, compression and transmission. At the same time, it also proposes a solution from the semantic level to the limits of communication transmission, so how will the communication technology of the post-Shannon era develop based on semantic changes?
The "limit" of communication transmission
In 1948, a great communication theory was born in the field of information and communication - the famous Shannon's theorem: there is a theoretical limit value for communication channel capacity, which indicates the limit information transmission rate and channel capability in the channel, and is regarded as the "golden rule" of modern communication. Since then, Shannon's theorem has become the basic theory of modern information theory, the information industry has developed rapidly, widely used in the field of communication and data storage, based on Shannon's theorem more than seventy years ago, generation after generation of communication products have emerged, and laid a solid theoretical foundation for the development of communication today.
Entering the era of digital intelligence, we have entered the intelligent society of all things perception and interconnection of all things, and new production and living models are also rising, which have undergone fundamental changes from the Shannon era seventy years ago. With the rapid development of 5G technology, the limits of Shannon's theorem have almost been reached. Many communications experts in the industry have also shown that "no G after 5G" has approached capacity limits in terms of point-to-point channel capacity. In the future development of communication technology, all that remains is the expansion of the number of communication channels.
The development of the communications industry has reached a stage where it is necessary to look directly at the challenges. At the Mobile World Congress, Guo Ping, Huawei's rotating chairman, pointed out: "Shannon's theorem and the von Neumann architecture have encountered great bottlenecks, and we need to explore new theories and architectures. ”
The bottleneck of Shannon's limit and von Neumann architecture also extends the direction of computing and communication evolution in the direction of quantum computing, DNA storage, and semantic communication. Semantic communication is used to solve problems such as the storage and efficiency of information transmission, and the characteristics of quantum computing are used to perform huge computational tasks in parallel.
Exploring these new modes of information transmission and processing can extend the transmission of information to a wider range of unknown areas, such as deep-sea communication, deep-space communication, human body networks and molecular networks. These research directions provide new room for growth and innovation in the future.
Multi-agent ubiquitous joins
To explore and solve the problem of semantic communication at Shannon's limit, it is necessary to answer the basic connotation and core essence of semantic communication. The concept of semantic communication was first proposed by Weaver and Shannon in a classic paper published in 1948, semantic communication is signal-oriented semantic communication, the essence of which is to transmit information expressed by semantic symbols, these symbols can make the recipient understand the meaning of it in seconds.
The purpose of communication between humans is to achieve the intention, and we use voice to communicate every word spoken, not just to let the other person hear but to understand. In the transmission of modern information, the current communication technology is the handling of information signals rather than the transmission of information. The transmission of information is mainly based on the principle of signal waveform or data bit fidelity, and the task of extracting and understanding the information in the signal is entrusted to the person himself. From signal transmission to information understanding is a cost, for the transmission of a large number of signals, transmission efficiency and direct understanding of the information communication methods need to be reconstructed and changed.
The signal receiving and receiving parties of semantic communication rely on the local semantic knowledge base to transmit very little content, and each of them is encoded and decoded separately to achieve the accurate and rapid transmission of a large amount of information. Similar to the black technology transmission form in science fiction dramas and spy dramas, the transmission of a very small amount of content can complete the transmission of information, which not only reduces the time of information transmission and understanding, but also ensures the quality of real-time fidelity communication.
Semantic communication is the deepening of the essence of communication, when the information theorist Weaver proposed this future concept ahead of the time, because of the limitations of technology, the early communication equipment was just a signal conversion machine, did not have the ability to intelligence, could not express and understand semantics. With the continuous development of 5G + AI technology, the intelligent progress of communication equipment, the mode of communication has undergone new changes, but also so that the research and development of semantic transmission has a soil and carrier that can grow, and the new communication mode of task-oriented multi-agent has the possibility of communication and collaboration.
For semantic communication, the core is the semantic knowledge base, which is the cornerstone of communication between the two sides of the communication. The semantic knowledge base is based on universal knowledge and semantic system, which can break the incompatibility problem caused by the inconsistency of information modes in the current M2M intelligent connection.
It can also ensure the service experience of H2M and H2H smart connection interaction and communication, reduce the number of conversions between semantic and physical signals, and reduce possible semantic distortion. Industry insiders will also credibly use semantic communication as a true way of communicating artificial intelligence. It can meet the needs of ubiquitous interconnection between agents.
In the communication industry, semantic communication has also begun to be valued, and the giants and large companies in the communication industry have begun to explore the basic theory. In the pace of these forerunners, in the future, we will also see the real-time normalization of AR and VR, and the transmission of information between a large number of agents will be faster and more accurate.
The ultimate mode of communication
As a new communication technology in the era of intelligence, semantic communication will have broad prospects in the future of intelligent facilities, multi-agent collaboration, human-computer natural interaction, human-machine integration, and urban collaborative security. However, although semantic communication has great potential and value, it is difficult to have both fish and bear paws, and its development still faces challenges.
1. Semantic knowledge base is difficult to share. Semantic knowledge base is the cornerstone of semantic transmission, communication participants need to share the same or similar philosophical worldview and knowledge model, knowledge accumulation and data collection need to be continuously invested, will consume a lot of time and cost. In the transmission of some unknown semantic knowledge, there are also high requirements for the collaborative updating ability of the knowledge bases of both parties. Otherwise, in some cases where the knowledge base of the transmitter and receiver is inconsistent, the common knowledge of the two parties is inconsistent, and the transmitted information cannot be fidelity. For some short-lived plants and machines, maintaining and updating the knowledge base also requires significant communication and storage costs.
2. The limitations of semantic communication technology itself, the context of semantics is difficult to perceive and difficult to identify. Semantic signals contain a wide range of information and rich content. It is easy to be limited by various regions, backgrounds, classes, environments, etc., for example, people in the communication industry will be more sensitive to the expertise in the field of communication, and those who are good at autonomous driving will pay more attention to focusing on driverless information. Semantic information content transmission can easily be limited to the background of both ends of the information transmission. It can be difficult for knowledge identification and processing.
3. The transmission of information involves the contradiction of privacy and security. The two parties to the information transmission are constantly exchanging their own perceived information, and in the case of coordination and updating of the knowledge base, it may cause a potential risk of leakage of local knowledge/privacy data, and data privacy is difficult to protect.
These difficult problems are all obstacles to the development of semantic communication, and the bottleneck solution of Shannon's limit depends on the breakthrough of these limitations. The pattern of semantic communication changes the communication function from mutual signaling to mutual communication. The intelligent mode of human-machine collaboration and human-machine hybridization will also be able to rise to a new dimension. This also promotes efficient cooperation between agents to complete more complex, open, robust and other issues.
Communication is the foundation of the development of modern science and technology industries, and the reform of semantic communication will also lead the communication technology-related industries to a new stage. From the architecture of the communication network to the core neural network chip, including various communication interface specifications and standards, etc. will change with the emergence of semantic communication patterns.
These changes will also affect the changes in various industry interfaces and equipment standards. In the future, there may be such a scene, newly purchased goods, can communicate directly with people, do not need paper graphics or video versions of the instruction manual, and communicate with users has become the most convenient and fast trend. Users no longer need to learn this process of using the product. The machine language for communication between agents will also be born, and the collaborative understanding of multiple agents will also become a reality, and the efficiency of human-machine collaboration and machine-machine collaboration will be greatly improved.
In the past, the transmission of information focused on reproduction and speed, and in the ultimate development of technology, the ultimate pursuit of the communication industry from fast to understanding, people from the steps of analysis and processing of information simplification, liberation, information transmission efficiency from the perspective of content comprehensively upgraded.
In science fiction, the ultimate mode of communication is telepathy. Through some smart devices, as long as the two sides want to transmit information, the information will be transmitted through telepathy, there will be no misunderstanding and distortion, and the transmission efficiency of information is not limited by time, and it can be completed immediately. People can be liberated from inefficient information dissemination and their intellect is fully developed. Complete telepathy in science fiction works is difficult to achieve, but if you look closely at its essence, telepathy and semantic communication are similar in nature, and they are pursuing the fidelity and efficiency of information, in other words, semantic communication is also a low-level telepathic method. Once mature, the liberated efficiency and erased information barriers bring us an unlimited future.