| This dissertation proposes, develops, and evaluates a digital short term correlation filter which is an effective and practical signal processing algorithm for determining rate information from the sense axis output of a vibratory MEMS-based gyroscope. The vibratory gyroscope exploits the Coriolis effect, creating a secondary vibration that produces an amplitude-modulated sense signal. Due to its small mass, a vibratory MEMS-based gyroscope outputs a small signal that is very susceptible to electrical and mechanical thermal noise. The thermal noise defines the resolution of the sensor.; The short term correlation filter simultaneously demodulates the gyroscope's sense signal and significantly improves its signal-to-noise ratio. Furthermore, the short term correlation filter requires considerably less computational load than a conventional digital correlation filter, which makes it more suitable for real-time applications. The key to the short term correlation filter is to store the input samples in a series of buffers and then, using pipes, to process the information on a buffer-by-buffer basis rather than on a sample-by-sample basis. This approach greatly reduces the processing power required of the DSP, making the filter practical for a virtually real-time application. Within the bounds of decimation error, the short term correlation filter performs as well as a conventional, digitally-implemented correlation filter, demodulating the sense signal and increasing its signal-to-noise ratio in direct proportion to the number of samples used in the processing. Both simulation and experimentation results show the performance of the short term correlation filter in reducing the effects of additive white Gaussian noise. |
