Biblio

Underwater communication is vital for a variety of defence and scientific purposes. Current optical and sonar based carriers can deliver high capacity data rates, but their range and reliability is hampered by heavy propagation loss. A vertical Molecular Communication via Buoyancy (MCvB) channel is experimentally investigated here, where the dominant propagation force is buoyancy. Sequential puffs representing modulated symbols are injected and after the initial loss of momentum, the signal is driven by buoyancy forces which apply to both upwards and downwards channels. Coupled with the complex interaction of turbulent and viscous diffusion, we experimentally demonstrate that sequential symbols exhibit a Gaussian velocity spatial distribution. Our experimental results use Particle Image Velocimetry (PIV) to trace molecular clusters and infer statistical characteristics of their velocity profile. We believe our experimental paper's results can be the basis for long range underwater vertical communication between a deep sea vehicle and a surface buoy, establishing a covert and reliable delay-tolerant data link. The statistical distribution found in this paper is akin to the antenna pattern and the knowledge can be used to improve physical security.

Training deep neural networks is a computationally expensive task. Furthermore, models are often derived from proprietary datasets that have been carefully prepared and labelled. Hence, creators of deep learning models want to protect their models against intellectual property theft. However, this is not always possible, since the model may, e.g., be embedded in a mobile app for fast response times. As a countermeasure watermarks for deep neural networks have been developed that embed secret information into the model. This information can later be retrieved by the creator to prove ownership. Uchida et al. proposed the first such watermarking method. The advantage of their scheme is that it does not compromise the accuracy of the model prediction. However, in this paper we show that their technique modifies the statistical distribution of the model. Using this modification we can not only detect the presence of a watermark, but even derive its embedding length and use this information to remove the watermark by overwriting it. We show analytically that our detection algorithm follows consequentially from their embedding algorithm and propose a possible countermeasure. Our findings shall help to refine the definition of undetectability of watermarks for deep neural networks.