https://doi.org/10.1140/epjp/s13360-025-06520-9
Regular Article
DeepSemComm: a deep learning-based semantic communication framework with case studies in text and biomedical signal transmission
School of Computer Science and Artificial Intelligence, SR University, 506371, Warangal, Telangana, India
Received:
29
April
2025
Accepted:
3
June
2025
Published online:
26
June
2025
Semantic communication systems have garnered considerable attention in recent years due to their potential to enhance communication efficiency by conveying meaningful information rather than raw data. Traditional communication systems, built around the paradigms of bit-level transmission, suffer from everyday problems, including inefficient spectrum utilization and loss of semantic context in noisy backgrounds. Although many breakthroughs have been made in semantic communication, the main challenges that need to be addressed are real-time adaptation, robustness in noisy conditions, and scalability to multiple modalities (e.g., text, speech, physiological signals, such as ECG). Although existing expertise has succeeded in particular fields, it falls short of acknowledging issues regarding communication, such as cross-linguistic conversation, a variety of transmission data, real-time transmission, and operations in resource-constrained contexts. Others cannot guarantee a consistent approach with robustness and efficiency across different modalities. Herein, to this end, we develop a deep learning-based semantic communication framework, DeepSemComm, focusing on improving communication efficiency, robustness, and adaptability for various applications and data types, from the perspective of data (such as text or ECG signals). It combines transformer-based semantic encoding and decoding architectures, offering improved noise resilience and real-time adaptability. Specifically, a framework for cross-linguistic communication is proposed within the system, enabling global-level applications. Through experimental results, we demonstrate that DeepSemComm achieves higher communication efficiency, better semantic retention, and stronger noise resistance compared to existing methods. In noise settings, the framework demonstrates significant improvements over text- and ECG-based communication tasks, achieving up to 30% improvement in semantic preservation accuracy. The framework demonstrates its usefulness for multilingual communication or healthcare systems by processing such diverse communication passages. Following recent progress in nonlinear wave physics and symbolic computation, the framework features an emergent behavior reminiscent of principles realized in soliton dynamics and integrable systems in the presence of distortion. This cross-disciplinary basis increases immunity to noisy transmission of semantics in deep encoders, which act as nonlinear semantic transformers.
Copyright comment Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
© The Author(s), under exclusive licence to Società Italiana di Fisica and Springer-Verlag GmbH Germany, part of Springer Nature 2025
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
