On stationarity and second-order properties of bilinear random fields

Abstract

One-dimensional indexed bilinear (BL) models are widely used for modeling non Gaussian time series. Extending BL models to multidimensional indexed (spatial) SBL one, yields a novel class of models which are capable of taking into account the important characteristic of non Gaussianity and spatiality behavior. Our main contribution here is to extend various results on BL processes to SBL one. Our attention is thus focussed on the probabilistic structure of some SBLmodels. So, we establish necessary and sufficient conditions for the existence of regular stationary and ergodic solutions in term of their transfer functions. As a consequence, we observe that the second order structure is similar to a spatial ARMA model with some uncorrelated white noise. So, it is necessary to look into higher-order moments in order to distinguish between linear and nonlinear random fields. Our study may be applied to GARCH or ARMA random fields.