Exact Distribution of the Generalized Shiryaev-Roberts Stopping Time Under the Minimax Brownian Motion Setup

TL;DR

This paper derives exact formulas for the distribution of the GSR stopping time in a Brownian motion change detection problem, providing insights into detection performance under different regimes and parameters.

Contribution

It provides closed-form survival functions for the GSR stopping time in both pre- and post-change scenarios, solved analytically via spectral methods.

Findings

Closed-form survival functions for GSR stopping time derived

Distributional characteristics depend on drift magnitude, threshold, and headstart

Analytical results enable precise numerical characterization of detection delay

Abstract

We consider the quickest change-point detection problem where the aim is to detect the onset of a pre-specified drift in "live"-monitored standard Brownian motion; the change-point is assumed unknown (nonrandom). The object of interest is the distribution of the stopping time associated with the Generalized Shryaev-Roberts (GSR) detection procedure set up to "sense" the presence of the drift in the Brownian motion under surveillance. Specifically, we seek the GSR stopping time's survival function (the tail probability that no alarm is triggered by the GSR procedure prior to a given point in time), and distinguish two scenarios: (a) when the drift never sets in (pre-change regime) and (b) when the drift is in effect ab initio (post-change regime). Under each scenario, we obtain a closed-form formula for the respective survival function, with the GSR statistic's (deterministic) nonnegative headstart assumed arbitrarily given. The two formulae are found analytically, through direct solution of the respective Kolmogorov forward equation via the Fourier spectral method to achieve separation of the spacial and temporal variables. We then exploit the obtained formulae numerically and characterize the pre- and post-change distributions of the GSR stopping time depending on three factors: (a) magnitude of the drift, (b) detection threshold, and (c) the GSR statistic's headstart.