Underwater communication cables transport large amounts of sensitive information between countries. This fact converts these cables into a critical infrastructure that must be protected. Monitoring the underwater cable environment is rare and any intervention is usually driven by cable faults. In the last few years, several reports raised issues about possible future malicious attacks on such cables. The main objective of this operational research and analysis (ORA) paper is to present an overview of different commercial and already available marine sensor technologies (acoustic, optic, magnetic and oceanographic) that could be used for autonomous monitoring of the underwater cable environment. These sensors could be mounted on different autonomous platforms, such as unmanned surface vehicles (USVs) or autonomous underwater vehicles (AUVs). This paper analyses a multi-threat sabotage scenario where surveying a transatlantic cable of 13,000 km, (reaching water depths up to 4000 m) is necessary. The potential underwater threats identified for such a scenario are: divers, anchors, fishing trawls, submarines, remotely operated vehicles (ROVs) and AUVs. The paper discusses the capabilities of the identified sensors to detect such identified threats for the scenario under study. It also presents ideas on the construction of periodic and permanent surveillance networks. Research study and results are focused on providing useful information to decision-makers in charge of designing surveillance capabilities to secure underwater communication cables.

No abstract

No abstract

No abstract

The feasible implementation of signal processing techniques on hearing aids is constrained by the finite precision required to represent numbers and by the limited number of instructions per second to implement the algorithms on the digital signal processor the hearing aid is based on. This adversely limits the design of a neural network-based classifier embedded in the hearing aid. Aiming at helping the processor achieve accurate enough results, and in the effort of reducing the number of instructions per second, this paper focuses on exploring (1) the most appropriate quantization scheme and (2) the most adequate approximations for the activation function. The experimental work proves that the quantized, approximated, neural network-based classifier achieves the same efficiency as that reached by "exact" networks (without these approximations), but, this is the crucial point, with the added advantage of extremely reducing the computational cost on the digital signal processor.