Research On Data Storage Problem Of Wireless Sensor Networks In Challenging Environments

Many emerging sensor network applications are deployed in challenging environments that are remote and inaccessible. In such scenarios, it is difficult to have a long-term deployment of high-powered base stations in the monitoring f ield. And sensor nodes including sink nodes may fail due to resource constraints and the impact of external environments. The failures of sensor nodes will contribute to data loss, which will seriously affect the availability of sensory data. Therefore, how to design effective storage strategies to improve the reliability of sensory data has become a hot research topic.This paper firstly studies the data storage problem caused by resource depletion of sensor nodes. Since storage(or energy) constraints and data retrieval costs are ignored in current storage strategies with no dat a redundancy, two non-redundant data storage algorithms are proposed to solve the data redistribution and retrieval problem with data priority(DRRP). One is a centralized data storage algorithm(22DPE) based on minimum cost flow. The other one is a distributed data storage algorithm(22PD) based on cooperative communications among sensor nodes. Both of these two algorithms minimize the total energy consumption of data redistribution and retrieval under limited nodes’ energy and storage capacity.For unpredictable node failures cased by external environments, simple non-redundant storage strategies are unable to ensure the data reliability, so coding mechanism can be introduced to realize fault-tolerant data storage. In order to overcome the slow traversal and low decoding efficiency of existing data redundant storage solutions based on codes, another more applicable data storage algorithm(QRNCDS) based on both minimum spanning tree and quasi-randomized network codes is designed in this paper. QRNCDS disseminates k source data packets in the network according to the minimum spanning tree traversal mechanism. Each node that receives source data packets performs encoding operation according to the quasi-randomized matrix theory and eventually stores only one encoded data packet.The theoretical analys is and simulat ion show that the proposed three storage algorithms can not only reduce the energy consumption of communication, but also improve data persistence and extend the data preservation time.