The utility of binary-sequence-based signals

It is frequently implied that there are advantages to the use of signals based on binary sequences for the characterization of linear time invariant (LTI) systems. However, this appears to neglect the role of the interpolation (or reconstruction) filter which is necessary in order to translate discrete sequences into analog signals and to discount the utility of analog signals that are generated by other means. Whereas a discrete sequence may offer desirable or optimal characteristics in comparison with other discrete sequences, these may easily be destroyed by the imposition of the interpolation filter. Similarly, even when these characteristics are substantially preserved in the interrogation signal, they appear consistently to be either mediocre or sub-optimal in comparison to properties of signals that are generated without consideration of binary sequences (e.g., chirps or multi-sine signals). The relative superiority of any signal with respect to the alternatives for interrogation of an LTI system depends on the goals of the characterization and relies on the assumption that some form of nonlinearity will ultimately limit signal amplitude. There does not appear to be any set of goals and limiting nonlinearity that are uniquely suited to binary-sequence-based signals.