Joint source-channel coding for underwater multimedia transmission

  • * Register (or log in) to the Neural Network to add this session to your agenda or watch the replay

  • Date
    25 November 2025
    Timeframe
    15:30 - 17:00 CET
    Duration
    90 minutes

    Underwater acoustic communications present unique challenges for multimedia transmission, particularly in supporting oceanic research, the blue economy, and ecological monitoring through sensor networks. While traditional coding methods have served well, the limited bandwidth and variable channel conditions of underwater acoustics demand innovative communication solutions that optimize source-channel coding jointly.

    This session presents two groundbreaking approaches: first, an image transmission system that reconceptualizes the coding problem as a language-translation task, where input sequences are transformed based on estimated channel conditions using Long Short-Term Memory (LSTM)-based sequence-to-sequence models. Second, a novel analog video transmission scheme is introduced, specifically designed for small submersibles and Internet-of-Things (IoT) devices, that not only achieves reliable multicast communication but also doubles the video framerate while maintaining transmission fidelity under bandwidth constraints. The system’s potential for analog implementation makes it particularly power efficient – a crucial factor for underwater deployments.

    Through experimental evaluation using real-life acoustic Channel Impulse Responses (CIRs) the speaker will demonstrate the effectiveness of these approaches in revolutionizing underwater acoustic multimedia communications.


    Session Objectives: 

    By the end of this session, participants will be able to:

    • Identify challenges in underwater multimedia transmission and the need for joint source-channel coding.
    • Explore innovative image and video transmission techniques that are revolutionizing underwater acoustic multimedia communications.
  • Are you sure you want to remove this speaker?