Modified Allan Variance Analysis on Random Errors of MINS

Allan variance method is a useful tool for analyzing the random errors, but the confidence on the estimate would be lower when the data length became shorter, therefore the modified Allan variance is deduced to analysis the random errors of MEMS inertial sensors (MINS). The definition and limitation of Allan variance are presented first, and then the modified Allan variance is deduced. Allan variance method is compared with modified Allan variance by identifying the simulated 1/f noises, meanwhile the results are illuminated. In the end, the random errors of MEMS inertial sensors were analyzed by the proposed methods. The characteristics of MEMS accelerometers’ and MEMS gyros’ stochastic errors are identified and quantified. The derived error model can be applied further to our attitude and heading reference system of the underwater robot.