Signals are inherent to nature and most man-made processes that influence human existence. They are attributed to numerous sources and typical examples include acoustic, vibration, biological and financial systems. Before it is possible to process a signal, it is first necessary to measure or collect the signal data using an accurate measurement method that does not distort or limit the signal content in any way.

After the measurement, signal processing is then applied to the measured data in order to extract the information that is required. At this point, the person carrying out the analysis has to choose the most appropriate signal processing tool from a range of options.

In order to select the right tools, it is not only necessary to have a clear understanding of the characteristics of the measured signal i.e. signal class, but also the signal processing technique and its limitations. Most people are familiar with classical time and frequency domain methods which often become the method employed because of habit and availability, even though they may not be the most suitable choice.

For periodic, linear time invariant type systems, Fourier transformation is very effective in breaking down a complex signal into its individual frequency components, identifying the characteristics of individual components, their relationships and contribution. Various filtering and signal averaging methods are usually available to reduce noise and remove unwanted frequency components. Similarly, it is possible to identify the characteristics of physical elements in a system or process by measuring an input/output relationship or transfer function. For periodic signal measurements, reasonable analysis accuracy can usually be achieved without the optimisation of time-record length and windowing.

Many signal processes do not generate the type of signals described above; these include Evolutionary Broadband Signals and Transient Signals. Processing these signal types requires a much greater understanding of signal processing methods and a range of less known techniques, in order to extract useful information from the process and provide better sensitivity between events in the time and frequency domains. In this case, processing methods may include short time Fourier analysis, time frequency representation, adaptive filtering such as Kalman, auto regressive signal modeling, Prony residue techniques and Wavelet analysis.

An extension of these methods is to use the sensitivity of a signal processing technique to identify certain characteristics in the signal by the process of decomposition, then reconstructing the signal minus the components that confuse the signal identity.

Another area of importance in signal processing is Parametric Modeling. In this case, it is possible to derive the mathematical parameters that describe the signal, system or process. Using information such as impulse or frequency response or input/output relationships, it is possible to derive the coefficients of a linear system and model its behavior.

Further areas of application of signal processing include the generation of complex signals and sequences as sources of excitation for specific processes. Alternatively, complex adaptive algorithms are included in control loops to enable real-time control of a process; these include algorithms such as LMS, Filtered X, Filtered U and hybrid adaptations. Typical examples of control systems include active noise and vibration control and positioning guidance systems.