Research On The Event-driven IoT-aware Collaborative Business Process Modeling And Message Scheduling Technology

The Internet of Things(IoT)service is an important component of the next generation of information technology,which has received extensive attention in industry and academia.IoT services connect the physical world and digital space through a variety of business process applications hosted on physical devices,especially smart sensors.These smart sensors continuously collect external information and upload it to the data center in a specific data format.Firstly,the distribution of sensor is discrete and its perception environment is complex.Many cooperative patterns coexist in heterogeneous sensor networks,which makes the direct modeling of business process applications very difficult.In addition,massive raw data is encapsulated as a unified event,which is then routed,forwarded and processed.Thus,an efficient event scheduling algorithm is particularly important.Finally,due to the whole business process application is often in a distributed execution environment,we urgently need a deployment algorithm,which takes into account both load balancing and access efficiency.In this paper,we propose a resource-oriented direct modeling method,a dynamic consistency hash-based deployment algorithm and a dynamic priority first response-based event scheduling algorithm.These approaches have important practical value and theoretical guiding significance in constructing event-driven collaborative business process system.The main contributions of this paper are as follows:(1)In this paper,we propose a resource-oriented direct modeling approach.This method discusses how to directly model an IoT-aware event-driven collaborative business process application from two aspects:modeling specification and concrete executable method.we extend the traditional Petri net model and describe three collaboration modes:event unicasting,event broadcasting and service collaboration.Based on this,a formal modeling specification for sensor task nodes is proposed.In addition,to make the system executable,we integrate the paradigm of the IoT device at the business process level,and extend the graphical and machine-readable models of business process application based on BPMN 2.0 specification.Thus,an IoT device could be fully parsed as a special business process resource in the basic process element library.When a user needs to create an IoT-aware event-driven collaborative business process application,s/he could directly model it by dragging the corresponding process element icon.(2)To solve the problem of deploying event-driven collaborative business process applications in the heterogeneous IoT environment,a resource deployment algorithm based on Dynamic Consistency Hashing(DCH)is proposed in this paper.Because IoT applications are often in a distributed execution environment,a cross-sensor business process would be decoupled into multiple process fragments.The DCH algorithm first constructs a virtual hash ring of 232 nodes,then maps various resources,including process fragments and server IP,to the hash ring based on the given rules,and finally deploys resources into the target sever that is first encountered in clockwise direction in the hash circle.Through theoretical analysis and simulation,it could be found that DCH algorithm not only effectively improves the resource access rate,but also better guarantees the system load balance when the status of servers change.(3)This paper presents an event scheduling algorithm based on Dynamic Priority First Response(DPFR).In an actual IoT execution environment,the available computing resource is often limited.Thus,the performance of the whole system is often determined by the selected event scheduling algorithm.In order to get the system response as fairly as possible,the concept of Response Ratio(RR)is proposed in this paper.It could be expressed as the ratio of response time to execution time in numerical terms.If the waiting time is the same,the system first responds to the perceived events with shorter execution time;if the execution time is the same,the system first responds to the perceived events with longer waiting time.Furthermore,a series of simulation experiments are carried out to verify the effectiveness of the proposed algorithm.The results show that DPFR algorithm could effectively improve the throughput and CPU utilization of the system and reduce the loss rate of perceived events compared with the traditional First Come First Response(FCFR)and Static Priority First Response(SPFR)scheduling algorithms.