Online Time Interval Top-k Queries in Wireless Sensor Networks

Abstract

Abstract-Motivated by many applications, top-k query is a fundamental operation in modern database systems. Technological advances have enabled the deployment of large-scale sensor networks for environmental monitoring and surveillance purposes, efficient processing of top-k query in such networks poses great challenges due to the unique characteristics of sensors and a vast amount of data generated by sensor networks. In this paper, we first introduce the concept of time interval top-k query that is to return k highest sensed values from the sensory data generated within a specified time interval. We then propose a filter-based algorithm for time interval top-k query evaluation, which is capable to filter out nearly a half unlikely top-k data from transmission in comparison with a well known existing solution. We also develop a novel online algorithm for answering time interval top-k queries with various ks and time intervals one by one through maintaining a materialized view that consists of historical top-k query results. We finally conduct extensive experiments by simulations to evaluate the performance of the proposed algorithms on real sensory datasets. The experimental results show that the proposed algorithms outperform existing algorithms significantly to prolong the network lifetime.

Top-k Queries in Wireless Sensor Networks Leveraging Hierarchical Grid Index

“…Paper [2] proposed a online time interval top-k query which is to return k highest sensed values from the sensory data generated within a specified time interval. The topology of the sensor network is a spanning tree T rooted at the base station.…”

Content-Based Wake-Up for Top-k Query in Wireless Sensor Networks

“…[22] uses an approach termed tiny aggregation (TAG) to exploit data aggregation, by which each intermediate node along a constructed route compares its own data with the received data, forwarding only the data that are likely to be included into topk data set. There are also solutions based on arithmetic filters, e.g., filtering approach (FILA) [23], exact top-k (EXTOK) [7], and their enhanced versions, (e.g., [24], [25], [26]), where each sensor node transmits its sensing data only if its value is within a filter. These schemes are useful for repeated top-k monitoring where sensed data is temporally correlated.…”

“…A more exact approach is to carry out a top-k value-set query, where the sink collects top-k values in the sensing field [7]. To illustrate, assume that there are 5 nodes A, B, C, D, E, which respectively have the sensing data of 16, 25, 30, 32, 30. In this case, the information to be collected with the top-3 node-set query is C, D, E and their readings of 30, 32, 30 while that with the top-3 value-set query is 25,30,32,30 and their corresponding ID of B, C, D, E.…”

A General Framework for Efficient Continuous Multidimensional Top-k Query Processing in Sensor Networks

“…Instead of using a shortest-path tree for routing, XP [24] employs the cluster tree as the routing infrastructure, in order to aggregate many data points locally before further transmissions. Chen et al [7] strived to return k highest data points generated within a specified time interval.…”