Research On Complex Event Recognition With Reverse Security Characteristics

The Internet of Things(IoT)involves a large number of Internet sensors that collect large streams of data as input to various IoT applications,which are widely used to monitor human behavior,prevent electronic fraud,monitor cardiac arrhythmias,and so on.In order to monitor data and search for specified information in real-time in IoT applications,Complex Event Recognition(CER)technology is increasingly used and has gradually transitioned from an emerging technology to play an important role as a key platform for many industries.Complex Event Recognition abstracts raw sensor data streams into event sequences and has become a technique for monitoring event sequences to search for patterns of events of interest to the user,allowing the user to react and take proactive measures.However,as this technology becomes more widely used,how to better balance the privacy protection issues and matching utility has become a hot research topic in recent years.Researchers have proposed many schemes to balance privacy protection and utility.Most of the existing schemes hide individual attributes of events or at the pattern level,and then match the event patterns without setting up mechanisms to protect the complete event sequence,which does not apply to attackers who try to steal the complete event sequence to analyze user behavior.There are also schemes that prevent the adversary from reconstructing the complete event sequence to some extent but are slightly lacking in matching utility.In this paper,we address the above issues and investigate how to achieve efficient matching of more event patterns of interest to the user,but without revealing the user’s complete event sequence.The main contributions of this paper are.1.In this paper,a rule-set-based query model for event sequences is proposed.The model considers the ability to achieve pattern matching of events of interest to the user when the CER system is deployed in a sensitive environment while calculating the probability of an adversary reconstructing the complete event sequence.In this paper,we propose a depth-flexible truncated suffix tree with the rule set Q and string S,which can achieve pattern matching of events of interest to users,followed by introducing de Bruijn graphs,Aho-Corasick automaton to calculate the probability of an adversary to reconstruct the complete event sequence.The analysis and experiments in this paper show that the depth-flexible truncated suffix tree proposed in this paper achieves better compromise in terms of reverse security and matching utility compared with existing similar schemes.2.An adversary background knowledge model is proposed,and the scheme considered in this paper is more generalizable in terms of computing the probability of reconstructing the complete event sequence.It is not only able to take into account the case of disjoint adversary knowledge,but also to give the probability of leakage of the complete event sequence in the case of intersecting adversary knowledge.Then two methods are given to calculate the probability of a complete event sequence reconstructed by the adversary under the background knowledge model of the scheme in this paper,which are Aho-Corasick automaton method and the constructive edge method.