k-anonymity based framework for privacy preserving data collection in wireless sensor networks

In this paper, k-anonymity notion is adopted to be used in wireless sensor networks (WSN) as a security framework with two levels of privacy. A base level of privacy is provided for the data shared with semitrusted sink and a deeper level of privacy is provided against eavesdroppers. In the proposed method, some portions of data are encrypted and the rest is generalized. Generalization shortens the size of the data transmitted in the network causing energy saving at the cost of information loss. On the other hand, encryption provides anonymization with no information loss and without saving energy. Thus, there is a tradeoff between information loss and energy saving. In our system, this tradeoff is intelligently managed by a system parameter, which adjusts the amount of data portions to be encrypted. We use a method based on bottom up clustering that chooses the data portions to be encrypted among the ones that cause maximum information loss when generalized. In this way, a high degree of energy saving is realized within the given limits of information loss. Our analysis shows that the proposed method achieves the desired privacy levels with low information loss and with considerable energy saving.