Abstract: Unattended wireless sensor network (UWSN) have attracted more and more interests in recent research community. In UWSN, sensed data are stored locally for a long term, instead of being sent to a central sink immediately. It is motivated by certain applications that only digest information (e.g. historical information), not real-time data, are of interest. The digest information can be extracted on-site upon request and real-time data are avoided to be forwarded away in order to mitigate the communication overhead. As UWSN always confront many security risks and adversaries that result in nodes random failure or node compromise, such stored data need to be survived to the collecting moment. Therefore, the security problem arises: how to maximize the data survival till the data are collected, or to maximize the valid data upon data retrieval. In particular, the involved defense strategies need to be efficient due to the resource constraints. A family of strategies is proposed to improve the data survival in this paper. Some observations are proofed such as location entropy based hopping limited data moving strategy. The proposed advanced strategy makes use of such observation to achieve efficiency and takes the advantage of computational secret sharing to achieve both fault tolerance and compromise resilience. The analysis of the performance and security are also presented extensively.