Authentication by Location Tracking in Underwater Acoustic Networks

Physical layer message authentication in underwater acoustic networks (UWANs) leverages the properties of the underwater acoustic channel (UWAC) to identify the transmitting device. However, as the device moves, its UWAC changes, and the authentication mechanism must track such changes. In this paper, we propose an authentication mechanism that works in two steps: first, we estimate the position of the transmitting device, and then we predict its future position based on the previously estimated locations. Next, the position prediction error is used to verify the authenticity of the transmission. The position is estimated using a convolutional neural network (CNN) that takes as input the sample covariance matrix (SCM) of the estimated UWACs. The predictor is implemented via a Kalman filter or a long short term memory (LSTM)-based recurrent neural network (RNN). Numerical results obtained using the Bellhop ray tracer under various environmental conditions (water salinity, pH, and temperature) confirm the effectiveness of the proposed approach and show that the Kalman filter-based predictor outperforms RNN when a precise measurement and evolution model are available. Conversely, when such a model is not provided, the RNN performs better than the Kalman filter.