The Next Communication Paradigm Driven by AI-Based Semantic Communication

Shannon's Law

Contemporary communication has evolved based on Shannon's Law. This is a theory that defines the limits of network communication, and there have been attempts to maximize the channel capacity of the formula to be discussed later. Examples include technologies such as MIMO (Multiple Input Multiple Output). These technologies are part of the effort to possess higher channel capacity based on Shannon's theory.

Shannon's Law is expressed by the following formula for calculating channel capacity:

These respectively denote Channel Capacity, Bandwidth, and Signal-to-Noise Ratio. To put it simply in words, the Maximum Communication Speed (C) is the total physical frequency resource available to the system (B) multiplied by the Efficiency (log₂(1 + S/N)), which represents how many bits of information can be carried per 1Hz of that resource, depending on the signal quality (S/N).

With the emergence of this law for calculating channel capacity, the communication industry began to concentrate efforts on increasing channel capacity. Thus, for approximately 70 years, innovation in communication has mostly centered on the improvement of channel capacity.

However, in the current era, processing resources have advanced tremendously. Communication has progressed from sending text to the point of transmitting spatial vectors. Limitations are beginning to arise in reliably segmenting and transmitting all this data. For instance, the data generated by an autonomous vehicle can amount to several terabytes per day, and it is nearly impossible for the current communication network to cope with this. Simply laying more cables and installing more antennas reaches economic and physical limits for a solution.

Thus, a paradigm shift is being researched, moving away from the conventional paradigm of accurately transmitting the entire bitstream, and instead seeking to transmit only the context by introducing intelligence into communication. (The concept itself has existed for several decades.) This is a change driven by the recent powerful advancement of intelligent models and the need for communication of even larger data volumes.

This is referred to as Semantic Communication, as it involves the exchange of meaning.

Semantic communication

While conventional communication transmitted the entire data, semantic communication aims to transmit only the core meaning contained within it, namely the context.

This issue was already raised in Shannon and Weaver's communication model, and they separated the maturity of communication into three levels:

1. Technical Problem: How accurately can a symbol be transmitted? (This is the core domain of my theory.)
2. Semantic Problem: How accurately does the transmitted symbol convey the intended 'meaning'?
3. Effectiveness Problem: How effectively does the conveyed meaning influence the recipient's behavior?

To date, the evolution of communication has nearly resolved the technical problem, and it is now undertaking the task of addressing the semantic and effectiveness problems.

The difference between maturity Level 1 and Levels 2 and 3 (Semantic Communication) is typically illustrated using the example of a burning house.

A house is on fire.

In the current communication paradigm, this scene is meticulously converted into data and transmitted as a photograph.

In semantic communication, instead of sending all the data, such as "black smoke is coming out of the window and flames are visible," it is a method of conveying only the core 'meaning' such as "Fire outbreak, immediate dispatch necessary." The purpose of this is to boldly omit unnecessary information and prompt the recipient to take a specific action (dispatch).

If the communication is between endpoints sharing the same knowledge base in the category of firefighting, this can dramatically reduce the amount of data transmission required for situational awareness.

The core encoding/decoding logic of this semantic communication is a communication paradigm but operates on top of the application layer. On the transmitting side, the semantic encoder converts the given data into semantic data, and on the receiving side, the semantic decoder processes it into a form usable by the backend source. The two will take the form of an inference model, or similar, that possesses the same knowledge base, thereby enabling communication that exchanges semantics without the need for massive data transmission.

Naturally, this is guaranteed upon the completeness of the existing communication paradigm. First, the accurate technical transmission of symbols must be possible, and this level of maturity has already been achieved. Now, how well the transmitted symbols convey and interpret the semantics of the information has become the primary challenge, and research is just beginning.

However, unlike the existing syntactic communication system, a communication system based on such semantic context is highly likely to face issues because it relies on AI, etc., for reliability. Even if they possess the identical Knowledge Base, different interpretations may emerge from the black box area of the model.

Retrospective

It is suggested that 6G (Sixth Generation) mobile communication will incorporate semantic communication to establish an intelligent internet system, but there is a question mark regarding why a paradigm that operates on the application layer becomes a research task for mobile network operators. My intuition suggests that mobile network operators are responsible for guaranteeing maturity Level 1, where symbols and bits are transmitted accurately technically, and the point at which semantic communication operates is already the domain of application programs.

Furthermore, there is a question as to whether a new paradigm can exist in communication technology, which must fundamentally prioritize reliability. I also harbor these doubts and personally take a somewhat negative stance.

Nevertheless, the reason I write this article is that I find the next paradigm of mobile communication to be unfolding in a quite intriguing manner. The introduction of satellite internet to expand channel capacity is virtually an established fact with the emergence of Project Kuiper and Starlink, and the attempt to overcome the limitations constrained by Shannon's Law in a new form was quite fascinating.