Research On Disaster-Tolerant Data Storage In Wireless Sensor Network

Wireless sensor networks (WSNs) are data-centric networks, the collection,transmission and storage of data is the main task of WSNs. How to store a largeamount of sensed data is an important issue in WSNs.The sensor network consists of a large number of cheap sensor nodes withlimited computing power, the scarcity of resources and the dynamic nature of thenetwork structure may result in data loss. How to store the sensed data to improveefficiency of data query and recovery rate of loss data has become the key problemin WSNs. Effective storage strategy can not only meet the efficiency of data queryand balance the data load for each node in the sensor network, but also ensure thatthe data can be successfully recovered while a disaster happened.This dissertation focuses on data storage strategy in WSNs, especially datastorage and recovery under spatial failure pattern. First we describe the four storagestrategy in WSNs that is centralized storage, local storage, distributed storage, andnetwork database. Then we discuss two kinds of data storage strategy under spatialfailure pattern: data replication and erase code.Based on data replication, we propose an optimal matching strategy whichassigns a redundant storage node for each sensed node. The algorithm cansuccessfully recover data in fixed spatial failure pattern, while the cost of storage isoptimal. In random spatial failure pattern, we propose a storage strategy based onant colony algorithm, which can reduce the time complexity compared with theoptimal matching algorithm.The simulation shows the two disaster-tolerant storage strategies can reducethe total energy consumption of network and prolong the lifetime of networks forsuccessful data recovery.