Analysis and Modeling of an IMU and a FOG of an Underwater Vehicle Using Allan Variance

Allan Variance (AVAR) is a method to study the underlying noises residing in a stochastic random process. Devised by David W. Allan to characterise the stability of oscillators, it has been adopted later to investigate MEMS (Micro-Electro-Mechanical Systems) stability over time. Having a good knowledge about the noises that influence a sensor helps to devise better filters and have better estimates of the physical quantities we want to measure. In this regard underwater navigation is a sensitive topic of research: in the absence of a Global Navigation Satellite System (GNSS) signal, the underwater position of an Autonomous Underwater Vehicle (AUV) is estimated through sensor fusion; but in this environment, having the necessity of a small and watertight space, electronic devices are strongly coupled with each other, worsening their overall performance. AVAR can help study these interconnections, understanding how sensors are influenced and what is their behavior in complex structures as an underwater vehicle.