Transportation of data between nodes in a sensor network is expensive, as wireless radio transmission depletes finite battery capacity. In addition, wireless data transmission is prone to errors, like static, making reliable data exchange between sensor nodes even more expensive. This paper describes a novel transport scheme that allows sensors to predict data from other sensors. Thereby, communication can partially be omitted, which in return results in reduced radio traffic, less energy consumption, and thus improved network lifetime. In addition to that, simple techniques to ensure reliable communication become much more affordable. The proposed scheme seamlessly integrates into innetwork data aggregation. The prediction mechanism is based on the evaluation of polynomials derived from simplified Kalman filters.