Recent advances in wireless communications and electronics have enabled the development of wireless sensor networks. Spatial and temporal data correlation makes Wireless Sensor Networks (WSNs) unique compared to other ad hoc networks. In-network aggregation as a power-efficient mechanism for collecting data in wireless sensor networks has been emphasized with the development of other WSN network protocols. Communication errors during data aggregation may have an untrivial impact on aggregation algorithms. In this paper, we firstly propose a general architecture to seamlessly integrate data aggregation into wireless sensor networks. Secondly we present one approach to provide fault tolerant to data aggregation algorithms by exploiting the inherent correlations that exist between sensor nodes. Our approach enhances the reliability of aggregation algorithms instead of using traditional forward error correction (FEC) or autonomic resend request (ARQ) methods.