Chapter 4 - Signal Averaging

Abstract

Data analysis techniques are commonly subdivided into operations in the time domain (or spatial domain) and frequency domain. This chapter discusses processing techniques applied in the time (spatial) domain with a strong emphasis on signal averaging. Signal averaging is an important technique that allows estimation of low-amplitude signals that are buried in noise. The technique usually assumes that: signal and noise are uncorrelated; the timing of the signal is known; a consistent signal component exists when performing repeated measurements; and the noise is truly random with zero mean. In the real world, all these assumptions may be violated to some degree. However, the averaging technique has proven sufficiently robust to survive minor violations of these four basic assumptions. The remaining part of this chapter presents time (spatial) domain procedures to detect features such as signal power, zero-crossings, and peaks, and summarizes several other commonly applied techniques.