Wireless sensor networks have been widely deployed in many social settings to monitor human activities and urban environment. In these contexts, they acquire and collect sensory data, and collaboratively fuse the data. Due to resource constraint, sensor nodes however cannot perform complex data processing. Hence, cloud-integrated sensor networks have been proposed to leverage the cloud computing capabilities for processing vast amount of heterogeneous sensory data. After being processed, the sensory data can then be accessed and shared among authorized users and applications pervasively.

Various security and privacy threats can arise when the people-centric sensory data is collected and transmitted within the sensor network or from the network to the cloud; security and privacy remain a big concern when the data is later accessed and shared among different users and applications after being processed. Extensive research has been conducted to address the security and privacy issues without sacrificing resource efficiency. Unfortunately, the goals of security/privacy protection and resource efficiency may not be easy to accomplish simultaneously, and may even be sharply contrary to each other. Our research aims to reconcile the conflicts between these goals in several important contexts. Specifically, we first investigate the security and privacy protection of sensory data being transmitted within the sensor network or from the sensor network to the cloud, which includes: (1) efficient, generic privacy preserving schemes for sensory data aggregation; (2) a privacy-preserving integrity detection scheme for sensory data aggregation; (3) an efficient and source-privacy preserving scheme for catching packet droppers and modifiers.

Secondly, we further study how to address people's security and privacy concerns when accessing sensory data from the cloud.

To preserve privacy for sensory data aggregation, we propose a set of generic, efficient and collusion-resilient privacy-preserving data aggregation schemes. On top of these privacy preserving schemes, we also develop a scheme to simultaneously achieve privacy preservation and detection of integrity attack for data aggregation. Our approach outperforms existing solutions in terms of generality, node compromise resilience, and resource efficiency.

To remove the negative effects caused by packet droppers and modifiers, we propose an efficient scheme to identify and catch compromised nodes which randomly drop packets and/or modify packets. The scheme employs an innovative packet marking techniques, with which selective packet dropping and modification can be significantly alleviated while the privacy of packet sources can be preserved.

To preserve the privacy of people accessing the sensory data in the cloud, we propose a new efficient scheme for resource constrained devices to verify people's access privilege without exposing their identities in the presence of outsider attacks or node compromises; to achieve the fine-grained access control for data sharing, we design privacy-preserving schemes based on users' affiliated attributes, such that the access policies can be flexibly specified and enforced without involving complicated key distribution and management overhead.

Extensive analysis, simulations, theoretical proofs and implementations have been conducted to evaluate the effectiveness and efficiency of our proposed schemes. The results show that our proposed schemes resolve several limitations of existing work and achieve better performance in terms of resource efficiency, security strength and privacy preservation.