Efficient Aggregation Of Event Trend With Complex Patterns

With the popularization and continuous development of big data technology,many stream processing systems use multiple queries with Kleene patterns to aggregate event trends from time-series event streams.These Kleene patterns can match one or more events that meet conditions from the time-series event stream.These Kleene queries can provide many useful suggestions for stream processing applications in real-time,helping applications make timely decisions.The traditional time-series event aggregation system calculates the aggregation result of each query separately when performing aggregation calculation on the time-series event stream,resulting in a higher time delay.There is work to optimize multiple aggregate queries and provide a shared plan for multiple queries,but does not support aggregate queries with Kleene pattern.In response to these problems,a new aggregation method SC(Split-Connect)targets multiple patterns with Kleene closures,proposes a sharing scheme under multiple patterns,splits the pattern of aggregate queries into various types,and aggregates the split types first and save the aggregation result,and then use the saved type aggregation result to integrate the types to get the aggregation result of the entire pattern.SC shares a core type that includes Kleene closures,and efficiently calculates the aggregation result of multiple queries.Experiments on the SC method on simulated data show that the SC method can have higher real-time than the current best aggregation method under certain memory resources under multiple patterns queries with the same core type.Applying the SC method to the unlicensed car detection system,the system can efficiently calculate the results of multiple queries,which is of great significance to public safety.