The Research Of Critical Techniques Of Scalable Collaborative Virtual Environment

Collaborative virtual environment(CVE)is a research focus in currently computer science field which,combines some technologies such as computer supported collaborative work,artificial intelligence,multimedia and network.CVE enhances the function and application of computers as interaction and communication tools.Today,CVE has been applied in the fields of military simulation,entertainment,education,medicine and so on,but it is still a challenge to construct a scalable CVE because of its complexity and supporting technologies' limitation.In the paper,we focus on the research of critical technologies of scalable CVE and propose some solution proposals from three aspects of architecture,interest management and multicast.The purpose of our work is to provide supporting technologies for constructing a scalable CVE system.On the basis of Open Grid Services Architecture(OGSA)we discuss the communication model and the data distribution method of scalable CVE firstly.Then a CVE architecture based on grid services is devised.Utilizing grid service the CVE architecture separates CVE's function realizations from physical resource to achieve the purpose of constructing a scalable,functional expansible and heterogeneous CVE.In addition,the paper focuses on the research of virtual world management service and some related services which are key issues to solve the scalability of CVE.At last,a CVE case based on Globus Toolkit is developed to validate the feasibility of our CVE architecture and methods.Interest Management is one of the critical technologies of scalable CVE,but the existing Interest Management methods are not suitable for filtering dynamic data of large scale CVE.A novel Interest Management method oriented to filtering dynamic data of large scale CVE is devised.First,the method rapidly converges objects' interest aggregates to small range using grids to partition space.Then the method calculates the awareness intensity using space interaction awareness and collaboration awareness mechanisms.Based on the calculation of awareness intensity,interest levels and multi-level Interest Management are proposed.The efficiency of data's filtration and network resource's utilization is indicated by the simulation experiment's results.Multicast is also one of the critical technologies of scalable CVE.A variety of issues,both technical and commercial,have hampered the widespread deployment of traditional IP multicast in the global Internet Recently,as a viable alternative to IP multicast,application level multicast has been widely researched and used.In section 4 of the paper,we discussed application level multicast in detail.To satisfy one to many communication requirements in CVE,a single source application level multicast method called SSALMHMF is devised.SSALMHMF uses a hybrid multicast framework(HMF)which combines IP multicast with application level multicast to utilize network resource efficiently.Then how to construct and maintain mesh between multicast service nodes (MSN)in HMF is discussed.After that a multiple object optimization multicast routing mode (MOSSMRM)for mesh and a routing algorithm based on intelligence ant colony(IACSMR)are devised.A simulation experiment compared the routing algorithm with some other heuristic algorithms and the result shows that IACSMR is efficient.To satisfy many to many communication requirements in CVE,a centralized multi-source application level multicast method called CMALMM is devised.We discuss the problem of multi-source application level multicast(MALMP)and present a multi-source application level multicast routing model(MALMM).Then a heuristic routing algorithm(MALMRA)is devised for MALMM.In addition,we analyze MALMP's theoretical lower bound based on integer programming relaxation.Finally,we compare the performance of MALMRA's solution with MALMP's theoretical lower bound through a variety of simulation experiments.Experiment results demonstrate the effectiveness of MALMRA for solving MALMP.Because the scalability of centralized multi-source application method is not well,referring DHT routing mechanism of large scale peer to peer overlay network,we devise a distributed multi-source application level multicast based on random neighbor selection mechanism (DMALMM).DMALMM not only solves the problem of multi-source application level multicast efficiently but also has good scalability.Virtual space teleconferencing(VST)is a special CVE that has abroad application perspective. Adapting mesh-first and centralized strategies,a multi-source application level multicast method called ESMVST for awareness driven virtual space teleconferencing on Internet is devised.We focus on the application level multicast routing model VSTMMR and propose a heuristic routing algorithm MDMM.The result of experiment shows MDMM algorithm's performance and ESMVST method's feasibility.