Tracking objects underwater is prone to error since no equivalent system to GPS exists for underwater applications. Accurate positioning is vital for conducting surveillance. However, it is hard for a system to keep an accurate prediction while moving without a GPS source. Noise in sensor measurements causes drift which quickly leads to significant errors unless these errors can be successfully identified and removed. This study researches how to track the position of a portable platform that, at least in theory, should have the capability to be dropped into the ocean and track its own position. The initial position is known, but as soon as the node is dropped into the water, then this position can be seen as old. When dropped, the node starts moving with the current until coming to a complete stop somewhere on the ocean floor. A prediction of the position is given by combining multiple sensors in an extended Kalman filter aided by zero velocity updates. Once the node becomes stationary, smoothing is applied with prior information to improve the initial sensor bias estimates. This is done over several iterations allowing the initial biases to converge. The results indicate that this method could improve the prediction of the position estimate on the ocean floor.