Analogue Signal Predictor: Frequency-Domain Analysis

Predictors are employed to estimate the future states of a system that have many engineering applications, e.g. speech signal processors, process controllers, simulators of mechanical movements, etc. Common predictors are based on digital methods and consequently need digital processing equipment. In some cases, e.g. prediction on a real time scale and fast control, digital techniques appear rather complicated and expensive. Alternatively analogue electronics [Kuo, 1995] seem to be advantageous, for example to control fast chaos [Blakely et al., 2004]. Very recently we have described an extremely simple analogue predictor using active RC filters [Meškauskas et al., 2005]. The predictor is convenient for short-term predictions, especially in cases when the system model is not available. The time-domain characteristics have been considered using the Laplace transform. The hardware experiments have been carried out for simplicity at very low frequencies (several tenths of Hertz). It is the purpose of the present paper to extend the analysis of the analogue predictor [Meškauskas et al., 2005] to the frequency domain and to perform numerical simulations of the circuit at high frequencies.