Large file transfers or real time video underwater transmission at a low cost are becoming more necessary in the context of underwater observatories. Hence there is a need for increasing the current data rate capabilities both in shallow and deep underwater. Existing solutions (acoustic, mechanical or laser) suffer from different specific limitations: reduced data rate, insufficient positioning accuracy, high cost, etc. In this work supported by the French Europôle Mer organization, a high data rate underwater wireless radio system is being developed. We have chosen to modify some parts of the standard Wi-Fi architectures to keep costs low while obtaining performance that is currently unavailable under water. However, it is very well known that radio waves are strongly attenuated when they propagate in water. For this reason only wireless short range applications are expected to benefit from such innovations. Because modelling is necessary for electromagnetic (EM) design, we have first compared and selected an appropriate EM model of the sea. The most important parameters are the frequency, the salinity and temperature which varie with the geographic location, and sea weather variations. The attenuation computed in the band around 2.4 GHz differed, depending on the specific EM model that was used. A new underwater radio antenna is now under development. We have considered further specific constraints: easy integration into both remotely operated and autonomous underwater vehicles, insensitivity to variations of sea water electrical characteristics, resistance to very high hydrostatic pressures compatible with oceanic trench exploration, bio-pollution resistance, and low sensitivity to relative positioning of the transmitter and receiver. Polarization properties have been carefully chosen to improve the latter while simplifying the design. We are currently performing additional design optimizations and manufacturing an advanced antenna prototype in order to confirm the different simulation results and performance.