Recently, wireless sensor devices have been widely deployed in various application settings (including environmental research, control systems, etc.). Because of the inherent unreliability of sensor readings, any kind of reasoning in sensor environments needs to carefully account for noise. The key goal of PCET is to build an infrastructure that can automatically infer and reason about the probabilities of triggered events, using a principled probabilistic model for the underlying sensor data. Through such probabilistic reasoning, PCET can incorporate uncertainly factors and make finer-grain decisions on event occurrences. This is achieved through the use of a Bayesian Network to directly model and exploit correlations across different sensors and the definition of a complex-event language, which allows users / applications to create hierarchies of higher-level events. As experimental results verify, PCET simplifies the development process and boosts the efficiency of any system dealing with inherently uncertain data streams.