Efficient Complex Event Processing Over Concurrent Event Streams

We study the problem of Complex Event Processing from the event streams.Existing methods solve this problem based on the assumption that the events in the stream do not occur concurrently. Due to the application of concurrent data acquisition techniques, the events could occur concurrently, i.e., there could exist events with the same timestamps in the stream.However, the existing methods generally utilize the finite state automaton-based technique to discover complex events, which requires the events to have distinct timestamps so that the event sequences can be processed by the automaton, leading to the complex events not being discovered correctly under the scenario of concurrent events. In this paper, we propose an efficient method to discover complex events over concurrent event streams. We follow the technique route of using the finite state automaton to represent the complex event, while we design an efficient matching method for the automaton so that complex events can be correctly discovered over the concurrent events. Experimental results show our method can accurately discover complex events over concurrent event streams without sacrificing query efficiency.