An Efficient Content-Based Event Matching Algorithm Based on Double-Deck Tree Decomposition

A novel algorithm for event matching of publish /subscribe systems is presented. To use the filtering abilities of attribute names and values synthetically, a double-deck tree structure is proposed. For the first deck, the proposed algorithm builds a tree structure by using the attribute names. All subscriptions are partitioned into the nodes of the first deck tree according to the attribute contained in their constraints. For the second deck, the other tree structure is constructed for each leaf node of the first deck tree based on the attribute values. The subscriptions stored in each node of the first deck tree are also classified in order to be stored in the nodes of the second deck tree. When an event is being processed, the number of the possible matched subscriptions can be cut down by using the attribute values as the criterion. Experimental results show that the algorithm can shorten the matching time and improve efficiency of the event matching.