<p>This paper discusses the distributed filtering problem for discrete-time linear systems over sensor networks with measurement saturation and switching topologies. In a sensor network, a set of sensor nodes with saturated nonlinearity is deployed to sense and measure a plant, and the measurement is transmitted to the corresponding filter. Then, the local estimator in the filtering network collects the sensor measurements and combines them with the estimation obtained from neighboring nodes in accordance with the time-varying communication topology. During the process of identifying the communication topology, issues such as unrecognized (unavailable) signal delay and transmission losses may arise. Thus, Bernoulli random variables are used to characterize the filtering behavior for uniformly obtaining the current topology information. When the communication topology information is known, a distributed filter performs state estimation by combining both its own and neighboring nodes’ estimations. In contrast, when the communication topology is uncertain, the distributed filter degenerates into a topology-independent estimator, which performs the state estimation task entirely through self-estimation. Using the Lyapunov stability theory, sufficient conditions are obtained to ensure the exponential mean-square stability and <i>H</i>_∞ robustness of the filter error dynamic system. The developed distributed estimation method with partially topology-dependent characteristics is given in the form of a linear matrix inequality. Finally, two numerical examples are used to illustrate the effectiveness of the proposed design approach.</p>