A Simple and Efficient Tree-Based Algorithm for the Distributed Trigger Counting Problem

Abstract

The distributed trigger counting (DTC) problem is defined as raising an alarm and notifying a user when the total number of received triggers reaches a predefined value w in a distributed system of n nodes. DTC algorithms can be used for environmental surveillance with sensor networks and global snapshots. In this paper, we propose a simple and efficient algorithm for the DTC problem. The proposed algorithm is based on a tree structure of degree n and height 2. The proposed algorithm operates in three phases depending on the remaining number of triggers. We prove the correctness of the proposed algorithm: the probability of not notifying a user even though the total number of received triggers reaches w is 0. Experimental results show that the proposed algorithm has lower message complexity than the best previous algorithms: CoinRand and TreeFill. MaxRcv (the maximum number of received messages per node) of the proposed algorithm is also smaller than CoinRand and TreeFill when the number of nodes is not very large.