2025_programme: A Novel GRU-CNN Model for Self-Interference Cancellation in IBFD Underwater Communication Systems

Back


  • Day: June 19, Thursday
      Location / Time: D. CHLOE at 15:10-15:30
  • Last minutes changes: -
  • Session: 21. Underwater Communications and Networking
    Organiser(s): Charalampos Tsimenidis, Paul Mitchell, Konstantinos Pelekanakis
    Chairperson(s): Charalampos Tsimenidis, Paul Mitchell, Konstantinos Pelekanakis
  • Lecture: A Novel GRU-CNN Model for Self-Interference Cancellation in IBFD Underwater Communication Systems [Invited]
    Paper ID: 2191
    Author(s): Manar Alhamnday, Bilal Jebur, Charalampos Tsimenidis, Shahid Mumtaz
    Presenter: Charalampos Tsimenidis
    Abstract: The in-band full duplex (IBFD) scheme has attracted immense attention in the underwater communication society due to, its ability to improve communication throughput. However, self-interference (SI) remains a major problem for the performance and efficiency of IBFD systems, as the transmitted signal from a device leaks into its own receiver, necessitating advanced cancellation techniques to enable effective communication. Recently, machine learning (ML), particularly deep learning (DL), has presented promising techniques for self-interference cancellation. In this work, we propose a deep learning-based hybrid model, termed GCCN, which synergistically integrates Gated Recurrent Units (GRU) with Convolutional Neural Networks (CNN) specially designed to enhance self-interference cancellation (SIC) in IBFD underwater communication systems. Our proposed model leverages the temporal processing skills of GRUs for sequential data, along with the spatial feature extraction strengths of CNNs for collecting complex patterns in data, to effectively mitigate SI. Experimental results demonstrate that the GCCN model outperforms traditional methods and standalone deep learning models in terms of interference cancellation performance and improves the bit error rate (BER), achieving significant improvements in signal quality and system throughput.
      Download the full paper
  • Corresponding author: Dr Bilal Jebur
    Affiliation: Ninevah University
    Country: Iraq

Back