Self-Evolving Wireless Communications: A Novel Intelligence Trend for 6G and Beyond

TL;DR

This work introduces self-evolving wireless communications as a novel intelligence trend for 6G and beyond, addressing the limitations of traditional adaptive systems. It proposes a three-layer data-information-knowledge architecture where environments are sensed, data and knowledge are fused, and knowledge-driven decisions steer evolution, enabling autonomous learning and policy updates. Through hypothetical modeling and example experiments (e.g., SaE-ELM and SE-QL), the paper demonstrates potential gains in BER performance and convergence efficiency, illustrating how self-evolving modules can adapt to unknown environments and heterogeneous applications. It also outlines promising technologies and key challenges—ranging from environment characterization to security and latency—emphasizing the need for interdisciplinary research to realize robust, scalable, and secure self-evolving networks for 6G and beyond.

Abstract

Wireless communication is rapidly evolving, and future wireless communications (6G and beyond) will be more heterogeneous, multi-layered, and complex, which poses challenges to traditional communications. Adaptive technologies in traditional communication systems respond to environmental changes by modifying system parameters and structures on their own and are not flexible and agile enough to satisfy requirements in future communications. To tackle these challenges, we propose a novel self-evolving communication framework, which consists of three layers: data layer, information layer, and knowledge layer. The first two layers allow communication systems to sense environments, fuse data, and generate a knowledge base for the knowledge layer. When dealing with a variety of application scenarios and environments, the generated knowledge is subsequently fed back to the first two layers for communication in practical application scenarios to obtain self-evolving ability and enhance the robustness of the system. In this paper, we first highlight the limitations of current adaptive communication systems and the need for intelligence, automation, and self-evolution in future wireless communications. We overview the development of self-evolving technologies and conceive the concept of self-evolving communications with its hypothetical architecture. To demonstrate the power of self-evolving modules, we compare the performances of a communication system with and without evolution. We then provide some potential techniques that enable self-evolving communications and challenges in implementing them.

Figures (6)

• Figure 1: Self-evolving communication systems versus other communication systems.
• Figure 2: A conceptual diagram of a self-evolving communication system.
• Figure 3: BER comparison of conventional ELM and SaE-ELM methods in Rayleigh channels.
• Figure 4: Throughput comparison of Q-learning and SE-QL methods in UAV rendezvous.
• Figure 5: Promising technologies corresponding to each module. In the information layer, classification algorithms in environment sensing are used to extract information in different environments or semantic scenarios, algorithms in intelligent waveform generation are used to handle optimization problems of receiving transmitters and waveforms, and algorithms in intelligent decision-making are used to assist in various optimization problems and communication resource allocation problems, etc. The algorithms for the knowledge base are used for the construction and management of the knowledge base.