In this study, we focus on design efficient security techniques to maximize chances of data survival in wireless sensor networks, which involve disconnected or unattended operation with periodic visited by the sink, we refer to such networks as UWSNs. Data security in such UWSNs poses a number of challenges when applied in security-sensitive environments. First, sensors must accumulate data for a long time until it can be off loaded to a periodic sink. The adversary has lots of time to mount various attacks that aim to learn, erase, or modify potentially valuable data collected and held by sensors. Second, there is no ever-present sink, thus real time detection dose not help and the adversary can reach its goal and remain undetected. To address these security problems, we present CADS, a novel Co-operative and Anti-fraud Data Storage scheme for UWSNs by integrating the techniques of secret sharing and Discrete Logarithm Problem (DLP). We first propose a share generation and distributed scheme to achieve reliable and fault-tolerant initial data storage by providing redundancy for original data components, we then utilize discrete logarithm problem to ensure the integrity of the distributed data shares. The proposed scheme enables individual sensors to verify all the related data shares simultaneously in the absence of the original data in each round. Security analysis and simulations show that the proposed scheme has resistance against node capture attacks and outperforms existing security scheme in terms of data survival quantity and false negative probability.