The Internet of Things (IoT) concept is, and will be, one of the most interesting topics in the field of Information and Communications Technology. Covering a wide range of applications, wireless sensor networks (WSNs) can play an important role in IoT by seamless integration among thousands of sensors. The benefits of using WSN in IoT include the integrity, scalability, robustness, and easiness in deployment. In WSNs, data aggregation is a famous technique, which, on one hand, plays an essential role in energy preservation and, on the other hand, makes the network prone to different kinds of attacks. The detection of false data injection and impersonation attacks is one of the major concerns in WSNs. In order to verify the data, there is either the end-to-end approach or the hop-by-hop approach. In the former, the detection of these attacks can only be performed at sink node, i.e., after reception of aggregate, a detection that is inefficient and leads to a significant loss of legitimate data. In this paper, we propose a scheme that provides the end-to-end privacy and allows early detection of the attack through a hop-by-hop verification, thus reducing the need to rely entirely on sink node for verification. Based on an enhanced version of TinyECC, the solution is implemented on MicaZ and TelosB motes. Through simulation and experimental results, we show the applicability of the scheme for WSNs.