THE DETERMINATION OF THE NUMBER OF TERMS IN A MULTICHANNEL SINUSOIDAL REGRESSION
Hideaki Sakai · Journal of Time Series Analysis · 1993
Abstract. This paper treats a problem in multichannel harmonic analysis when the signals are sums of sinusoids whose periods are divisors of the series length and the disturbance noises are white Gaussian. The presence or absence of the harmonics is determined by an information‐criterion‐like method. The criterion proposed here is a multichannel generalization of a Bayesian information criterion for the scalar signal case independently derived by Quinn and the author. The criterion is based on the distribution of the maximum of N/2 independent χ2d7 random variables where N is the series length and d is the number of channels. Some simulation results are also presented.