This paper presents an action functional, sample path optimization technique, for formulating and solving nonlinear discrete-time stochastic H∞ estimation problems. These H∞ problems are formulated as minimax dynamic games ...