Machine Learning-Driven Solutions for Enhanced Underwater Communication

​

A.) Summary figure of ML-driven technologies to improve underwater acoustic communication for a variety of applications. These technologies enhance image and video transmission while adapting for underwater conditions that may interfere with signal transmission. B.) Schematic for integration of JSCC, ASVTuw, and Doppler effect compensation technologies in underwater image and video reconstruction

Invention Summary:

Image and video transmission are key elements for underwater exploration. Underwater acoustic communication (UAC) is an established method where transmitted sound waves bouncing off underwater objects are detected and collected to form a visual representation of an object’s characteristics. Traditional methods for underwater communication are often computationally intensive and struggle to adapt to the dynamic and unpredictable conditions of the environment, creating a need for innovative solutions that can transmit information more efficiently.

Rutgers researchers have developed machine learning-powered solutions to tackle the challenges of underwater acoustic communication, improving image clarity, video streaming, and data transfer in tough environments.  Three technologies provide solutions for overcoming challenges:

1. AI-Powered Image Transmission

Docket # : T2023-049

1. • Problem : Traditional underwater image transmission methods struggle with low bandwidth and signal distortion.
   • Solution : Rutgers researchers developed a deep learning-based joint source-channel coding (JSCC) approach that optimizes image coding and transmission based on channel conditions.
   • Key Benefits :
     • Adaptive error protection based on Channel State Information Feedback (CSI)
     • Efficient bandwidth usage and adaptability to variable underwater acoustic channels
2. Adaptive Video Streaming for Underwater Networks

Docket # : T2023-050

2. • Problem : Delivering high-quality video to multiple uses with varying needs and conditions is challenging in underwater networks.
   • Solution :  The Adaptive Scalable Video Transmission (ASVTuw) leverages machine learning to adjust video coding and transmission schemes based on channel states and user requirements.
   • Key Benefits :
     • Avoids resource wasted by preventing the transmission of redundant SVC sub streams
     • Effectively satisfies diverse user video quality requirements
3. ML-Based Doppler Compensation

Docket # : T2023-051

• • Problem : The Doppler effect causes frequency shifts and distortion in underwater communication.
  • Solution : Machine learning (ML) tracks and compensates for the Doppler effect. The ML-based tracker estimates motion and tracking-aided ML-based compensator corrects for interference caused by multipath Doppler effect.
  • Key Benefits :
    • Enables more accurate and efficient signal demodulation
    • Improves signal accuracy in the presence of multipath interference

These innovative methods are crucial for overcoming challenges in underwater exploration.

Market Applications:

• Ocean monitoring
• Underwater exploration
• Search and Survey
• Defense and Security

Advantages:

• Improve underwater image and video quality and clarity
• Adaptable to underwater conditions
• Limits resource waste

Publications:

• ASVTuw : Adaptive Scalable Video Transmission in Underwater Acoustic Multicast Networks (2022)
• Deep Joint Source-Channel Coding for Underwater Image Transmission (2022)
• ML-based Joint Doppler Estimation and Compensation in Underwater Acoustic Communications (2022)

Intellectual Property & Development Status: Provisional application filed. Patent pending. Available for licensing and/or research collaboration. For any business development and other collaborative partnerships, contact:  marketingbd@research.rutgers.edu