Semantic Communications in 6G: Coexistence, Multiple Access, and Satellite Networks

Ishtiaque Ahmed, Yingzhuo Sun, Jingwen Fu, Alper Köse, Leila Musavian, Ming Xiao, Berna ÖzbekIEEE Commun. Stand. Mag 2025 · DOI

TL;DR

This paper explores the integration of Semantic Communication (SemCom) with traditional Bit-based Communication (BitCom) in future 6G networks, focusing on enabling coexistence, efficient multiple access, and applications in satellite networks. It highlights how SemCom can significantly improve spectral efficiency and resource allocation by transmitting meaning rather than raw bits, especially in challenging environments.

What Problem Does This Paper Address?

The exponential growth of wireless users and the increasing demand for data services are pushing current communication systems towards their spectral efficiency limits. Traditional Bit-based Communication (BitCom) struggles to meet these demands, especially in scenarios with limited bandwidth or harsh channel conditions, like those found in satellite networks. This paper addresses the challenge of designing future 6G networks that can efficiently handle diverse communication needs, including both traditional bit-based data and semantically rich information. It specifically investigates how to enable SemCom and BitCom to coexist within the same network, how to manage multiple access for these different types of users, and how SemCom can be leveraged to overcome the unique challenges of satellite communications.

Methods

The paper proposes and analyzes several key approaches:

  • SemCom-BitCom Coexistence: It explores how SemCom, which transmits extracted meaning, can complement BitCom, which transmits raw bits. This is particularly relevant for scenarios where BitCom is superior for exact data recovery at high Signal-to-Noise Ratios (SNRs), while SemCom excels at lower SNRs and for efficient meaning transmission.
  • Multiple Access Techniques: The paper investigates Non-Orthogonal Multiple Access (NOMA) as a promising technique for supporting both SemCom and BitCom users simultaneously. It analyzes how NOMA can be adapted to manage heterogeneous users, allowing for flexible resource allocation and power control.
  • Semantic Rate: A key concept introduced is the "semantic rate," which quantifies the rate at which semantic information is transmitted with a certain accuracy. This rate is modeled using a semantic similarity function, which is approximated using a generalized logistic function fitted to Deep Learning (DL) based SemCom outputs.
  • Satellite Networks: For satellite communications, the paper discusses the use of AI-driven semantic relays and Knowledge Base (KB) alignment to mitigate bandwidth limitations and harsh channel conditions. It also touches upon multi-modal SemCom frameworks for handling diverse data types in these networks.
  • Deep Learning (DL) for SemCom: The paper references DL techniques, such as BERT (Bidirectional Encoder Representations from Transformers), as being crucial for extracting semantic features in SemCom systems like DeepSC.

Datasets

Not explicitly discussed in the paper. The authors mention using simulations generated using MATLAB and averaged over Monte Carlo channel realizations. They also refer to using data regression on DeepSC outputs to approximate semantic similarity functions.

Key Insights and Ideas

  • SemCom as a Complement, Not a Replacement: The paper emphasizes that SemCom is not necessarily a complete replacement for BitCom. Instead, it can coexist and complement BitCom, offering advantages in specific scenarios (e.g., low SNR, bandwidth constraints) while BitCom remains crucial for applications requiring exact data reconstruction.
  • NOMA for Heterogeneous Users: NOMA is identified as a key enabler for integrating SemCom and BitCom users, allowing them to share resources efficiently. The paper suggests that SemCom can help address some of the complexity and fairness concerns associated with NOMA.
  • Semantic Rate as a New Metric: The introduction of a "semantic rate" provides a new way to evaluate communication system performance beyond traditional bit-rate metrics, focusing on the accuracy of transmitted meaning.
  • AI is Central to SemCom: The paper consistently highlights the role of Artificial Intelligence (AI), particularly Deep Learning, in enabling SemCom. AI is crucial for semantic feature extraction, knowledge base alignment, and adaptive transmission strategies.
  • Satellite Networks are a Prime Use Case: The unique challenges of satellite communications (bandwidth, latency, harsh channels) make them an ideal testbed and application area for SemCom, where its ability to transmit meaning efficiently can yield significant benefits.

Key Results

  • Opportunistic Scheme Superiority: In NOMA uplink scenarios, an "opportunistic scheme" that flexibly switches between SemCom and BitCom for the secondary user consistently offers the highest ergodic semantic rates, outperforming SemCom-only and BitCom-only schemes.
  • SemCom Advantage at Low SNRs: The SemCom-only scheme shows performance comparable to the opportunistic scheme at lower bit-rate requirements for the primary user, demonstrating SemCom's effectiveness in low-to-moderate SNR regimes.
  • BitCom Underperformance: The BitCom-only scheme significantly underperforms compared to opportunistic and SemCom-only schemes, especially at lower transmit powers.
  • Semantic Similarity Function: The paper provides a fitted generalized logistic function to approximate the semantic similarity, enabling tractable analysis of SemCom performance.

Limitations and Future Work

Limitations:

  • Interpretability of DL Models: As SemCom systems become more complex with DL, understanding and interpreting their decision-making processes remains a challenge.
  • Performance Trade-offs: Achieving optimal performance trade-offs in multi-modal SemCom systems, where feature selection can impact individual task performance, requires careful optimization.
  • KB Alignment in Satellite Networks: Ensuring that transmitters and receivers have aligned Knowledge Bases (KBs) in large-scale satellite networks is difficult due to wide coverage areas and numerous nodes.
  • Cascaded Fading in Satellite Networks: Severe cascaded fading in satellite networks poses a challenge for reliable communication.

Future Work:

  • AI-Native Physical Layer: Integrating semantic awareness into the physical layer through AI-native air interface designs for enhanced resource utilization in satellite communications.
  • Energy-Efficient Implementations: Developing energy-efficient SemCom solutions is critical for scaling to resource-constrained devices.
  • Standardized Semantic Metrics: The need for standardized semantic metrics for evaluating SemCom performance.
  • Real-time Adaptation: Enhancing real-time adaptation of SemCom systems to dynamic network conditions.
  • Multi-modal SemCom Optimization: Further research into optimizing multi-modal SemCom frameworks for complex tasks and diverse data types.
  • Semantic Relays and KB Coordinators: The paper suggests using semantic relay satellite nodes and KB coordinator satellites to address KB alignment and fading challenges in satellite networks.

Summary updated

BibTeX
@article{ahmed2025semantic,
  author = {Ishtiaque Ahmed and Yingzhuo Sun and Jingwen Fu and Alper Köse and Leila Musavian and Ming Xiao and Berna Özbek},
  title = {{Semantic Communications in 6G: Coexistence, Multiple Access, and Satellite Networks}},
  journal = {IEEE Commun. Stand. Mag.},
  year = {2025},
  doi = {10.1109/mcomstd.2025.3592327},
}
Figure