Research On The Key Problems Of HLA Simulation In Grid Environment

At present, high level architecture(HLA) is the mainstream of application in distributed simulation technology.Grid is a bottom-layer support framework for high performance computing and data processing in distributed environment and it is able to manage a large number of heterogeneous resources of geographically and organizationally distributed.It is the demand for more development to build the HLA simulation in grid environment and research the key problems.These researchs have a profound theoretical significance and great value of the engineering and have become research focus of simulation community at the international and domestic.In this paper, the main work and innovative results are as follows:(1) Proposes a Grid-HLA framework and Proxy-based architecture.View of the current HLA/RTI limitations exist and the the feasibility of combining grid technology, this paper proposed a distributed simulation framework Grid-HLA by migrating HLA to the grid and designed a Federate-Proxy-RTI architecture system.The system enables different simulator(Federation) interoperability.The Grid-HLA simulation service discovery, implementation process and key components have been studied.The interface of architecture comply with the standard HLA interface characteristics and can achieve reusable and have load-balancing effect.(2) Proposed the PASMM algorithms and TDMM algorithm of independent simulation task scheduling on heterogeneous platform.For the independent task scheduling of simulation grid on heterogeneous platform, this paper proposed the priority based on average-standard Min-Min (PASMM)algorithms through the priority analysis of simulation task.Through the trust analysis of recommendation evidence,this paper proposed trust driven Min-Min (TDMM)heuristic algorithm. PASMM algorithm can achieve the unification of low span and short average waiting time.TDMM algorithm can eliminate hot spots of services, improve the dishonest node identification rate and the overall system performance.(3) Proposed the STDCMS algorithm of dependence simulation task scheduling on heterogeneous platforms.For the dependent task scheduling of simulation grid on heterogeneous platform, this paper presents a new scheduling algorithm STDCMS based on select replication of the parent task and cluster merging. Not only consider the task of copying other parent in order to reduce communication time, but also as little as possible to copy the parent task, rather than blindly copying all the ancestors of the cluster. This algorithm can produce more than TDS and OSA shorter scheduling algorithms. At the same time, the algorithm merger the tasks of lesser load processor under without increasing the length of scheduling, thereby reducing the number of processors.(4) Proposed three kinds of algorithms or strategies about interest mangagement:a region matching algorithm of combined adaptive extrapolation model and the R-tree; ontology-based grid service matching algorithm; multicast address allocation strategy of HLA hierarchyOn the basis of the study the interest technology of HLA data distribution management, this paper proposed a region matching algorithm of combined adaptive extrapolation model and the R-tree. The algorithm performance is analyzed and compared through experiment to test the effectiveness of the proposed method; This paper proposed a ontology-based grid service matching algorithm.This method is effective use of the knowledge of ontology.With an acceptable service to match the time, a higher success rate and a strong obtain capacity of grid services information, it provides a viable mechanism for matching the appropriate grid service; This paper proposed a multicast address allocation of HLA hierarchy structure and described the data forwarding algorithm. The results show that:the maximum level of nodes in the multicast reception of the message overhead can be significantly enhanced.(5) Proposed a dynamic adjustment method of Lookahead associated with time synchronization, LBTS optimization calculation method and the improved algorithm of Frederick.On the basis of an analysis of HLA time management,this paper proposed a Lookahead dynamic adjustment algorithm based scale factor and adjusting window; this paper Proposed a optimize design and implementation to separate of query from calculation of LBTS. This method can take the initiative check into a passive update, greatly reducing the number of LBTS communications.This paper explores the rules of deadlock and proposed an improved Frederick algorithm and solve the deadlock problem of the original Frederick algorithm.