Search for: All records

Traditional methods for coding underwater acoustic communications are bound to be surpassed by methods optimizing for source-channel coding jointly. However, the complexity of joint-optimization has thwarted successful breakthroughs in this area. We, therefore, present a novel approach, where we model the coding problem as the translation problem of the input sequence to another ‘language’, depending on the estimated channel conditions. We use Long Short-Term Memory (LSTM)-based sequence-to-sequence models to enable this and explain our approach in detail. 

Scalable Video Coding (SVC) has been widely used in video transmissions. However, inappropriate SVC structures may lead to received video quality lower than user’s requirement or resource waste, especially in underwater time-varying channels. In this work, an adaptive cross-layering solution is proposed and validated for video transmissions in underwater acoustic multicast networks, namely Adaptive Scalable Video Transmission (ASVTuw). In ASVTuw, the transmitter collects over time the information about the channel states and the users’ video quality requirements to adaptively select the SVC video structures and transmission schemes, using Machine Learning (ML). At-sea experiments were conducted to collect the required acoustic data. The collected data were then used in MATLAB simulations to validate the ASVTuw. The results show that the usage of ASVTuw avoids resource wasting from transmitting redundant SVC substreams and satisfies the multicast users’ video quality requirements effectively with higher flexibility compared with the existing noncross-layering designs.