Discussion Overview
The discussion centers around the selection of an appropriate filtering method for processing digital signals from an accelerometer, with a focus on achieving near real-time noise reduction. The context includes considerations of hardware limitations, specifically a 32-bit, 400MHz microprocessor without a floating point unit, and the nature of the noise affecting the signal.
Discussion Character
- Exploratory
- Technical explanation
- Conceptual clarification
- Debate/contested
- Mathematical reasoning
Main Points Raised
- One participant suggests using an LMS filter due to its O(n) complexity but notes its slower convergence compared to a Kalman filter, seeking opinions on the best filtering approach.
- Another participant inquires about the bandwidth of the valid signals, the spectra of the noise, the signal-to-noise ratio, and the possibility of noise cancellation using a second pickup.
- A participant clarifies that the primary noise source is vibration from a person holding the accelerometer, with minimal electrical noise, and provides specific bandwidths for the x, y, and z axes along with the sampling rate.
- One participant questions the desired adaptability of the filter, suggesting various options such as moving notches or variable gain, and emphasizes the importance of understanding the application context to determine the best approach for noise reduction.
- A participant expresses the intention to estimate overall acceleration rather than achieve exact measurements, indicating a focus on filtering the signal sufficiently for quantization to determine device orientation.
Areas of Agreement / Disagreement
Participants have not reached a consensus on the best filtering method, with multiple competing views on the approach to noise reduction and the adaptability of the filter. The discussion remains unresolved regarding the optimal solution.
Contextual Notes
Limitations include the dependence on the specific application context, the nature of the noise, and the constraints of the processing hardware. The discussion does not resolve the mathematical or technical details of the proposed filtering methods.