Orthogonal vs Skewed Inertial Sensors Redundancy: A New Paradigm for Low-Cost Systems

Abstract

Systems integrity is of primary concern in numerous applications and inertial navigation systems is one of them. In the early days, orthogonal redundancy of sensors was used for its simplicity. From the 70's, optimal skewed redundant structure of sensors have been preferred because it reduces the number of sensors performing a given task. It is believe that the advent of low-cost sensors, including micro-electro-mechanical systems (MEMS), will make a change in the paradigm of skewed redundancy back to the implementation of orthogonal redundancy. Based on the information theory, it is shown that given a fixed number of sensors, the volume of information is the same for optimal (skewed or orthogonal) configurations. Also, the loss of information upon fault isolation in the orthogonal configuration is less damaging for the overall system performance than its skewed counterpart. Finally, cost figures for orthogonal configuration of sensors is also challenged, giving a relative importance to geometric complexity of skewed configurations, hence making the former structure more appealing for real system design.