Monitoring Exact Top-K Values in Wireless Sensor Networks Using Dominating Set Trees

Top-k queries form an important class of aggregation queries in wireless sensor networks. Unlike previous proposals we consider the exact top-k values query problem, i.e., where one seeks to nd exactly all k highest unique values in the network regardless of how many nodes report it. Previous proposals also did not pay due attention to the underlying logical tree topology used for data aggregation and forward- ing. In this context, this paper presents two main contribu- tions: (1) we propose the use of a particular tree topology, based on Dominating Sets, which is well suited to explore the network's physical topology for processing top-k queries eciently; and (2) we propose EXTOK, a ltering-based al- gorithm for processing the exact top-k values query, we also prove its correctness and investigate its performance with respect to a number of parameters, including network link failures. In all examined cases, EXTOK performs consis- tently well while eectively exploiting the proposed logical tree topology and it is also resilient to link failures.