An adaptive method for structured replication data-centric storage in wireless sensor networks

Data-centric storage solutions are energy-efficient algorithms in wireless sensor networks and will greatly reduce total network load. However, storage nodes will become bottlenecks when the amount of events and queries targeted for a particular node is more than its capacity. This paper proposes a novel approach that brings load-balancing and scalability to the network as well as its ability to adapt itself to network conditions. This adaptability is made possible through a load-control mechanism provided by two thresholds which are responsible for changing hierarchy depth and the number of storage replicas as the frequency of events fluctuates in the network. Thus, the proposed method will not only reduce storage cost when the event frequency is high but also prevents hotspot problem in the root storage node. Moreover, it will reduce query traffic by merging replica nodes as the event frequency goes back to its normal state. Also, a simulator based on discrete-event system simulation methodology is developed to evaluate the proposed method on the basis of storage traffic, query traffic and the number of dead sensors. The result of evaluation shows that the proposed method has performance improvement over other compared methods.