Robust coherence analysis for long-memory processes

Abstract

This paper investigates the linear relationships between two time-series in the frequency domain, termed coherence analysis. It is widely used in various fields, including signal processing, engineering, and meteorology. However, conventional coherence analysis tends to be sensitive to outliers. Laplace cross-periodogram and a corresponding robust coherence analysis based on the least-absolute deviation (LAD) regression have recently been developed to improve this shortcoming. In this paper, to extend the scope of Laplace cross-periodogram, we study a robust cross periodogram for long-memory processes and derive its asymptotic distribution. Through numerical studies, we demonstrate the usefulness of the proposed robust coherence analysis for long-memory processes.