Research Of Data Distribution Model And Technology On Complex Network

With rapid growth of peer-to-peer (P2P) network technology and deployment, it becomesthe major topology in the world of data distribution. Though P2P dramatically improves theutilization of network bandwidth, it also causes the network congestion from time to time.How to improve the application level network topology and transmission model to get betternetwork usage becomes one of the hotspot in data distribution area.Based on in-depth analysis of complex network modelling and network topology, anoptimized method of complex network modelling is proposed. By simulating different datadistribution models over the network just built, an Occupancy Cost Model is proposed tofocus on the occupancy of linkage between nodes. The model provides a mean forquantitative assessment of the effectiveness of data distribution. To reduce transmission timeover the edges, an optimized network transmission protocol is proposed on high-latencyhigh-throughput network. Finally it proposes a Dynamic Cluster Net, a structural optimizednetwork topology, which aims at regulating the data flow between nodes in the network.Major tasks include:1. Research the modelling of complex network topology. Combining the structure of thehierarchical complex network, e.g. Internet at the autonomous system level, and thecharacteristics of preferential attachment in node growth of theoretical model, this paperpresents a dynamic rate based multiple layers Internet topology modelling with betteraccuracy in multiple measurements. According to different scales of the complex networksimulation modelling, the results show that three parameters of the simulation modelling, thenode degree distribution, clustering coefficient, and the average path length, are highlyconsistent with the typical complex network. Moreover, the results also match manycharacters of the complex network, such as scale-free and small world.2. Propose a quantitative Occupancy Cost Model for evaluating the performance of datadistribution. The model abstracts realistic network into points and edges which constitutetheoretical network topology. Network data flow and weighted edges are fed into the model asmain parameters to evaluate the balance of variety of data distribution patterns. Applying thismodel, the data distribution method with theoretical minimum occupancy cost can be deduced.This model points out the direction of optimization of data distribution. In addition, thecalculation and comparison of several typical data distribution methods’ network occupancycost reveal how peer-to-peer network transmission method takes advantage of network topology information to dynamically adjust the data delivery to produce a more balanced dataflow.3. Improve network transmission protocol by proposing a reliable high-throughputnetwork transmission protocol, ACE-Based Reliable UDP (ABRU), in high-latency networkto shorten the time for data transmission. Based on thorough analysis of TCP transportationcontrol methods, a quick start, high-throughput and reliable transport layer protocol isdesigned and implemented by adjustable parameters over UDP data packets to enable simplelightweight connection establishment, smooth D-AIMD congestion algorithm, flexible initialcongestion window and slow-start threshold. The protocol is implemented with adaptivecommunication Environment (ACE) interface. Experimental results show that under certainconditions, ABRU can achieve a higher rate than TCP transportation.4. Propose a Dynamic Cluster Network featuring regional dynamic clustering. In theform of static partition and dynamic clustering nodes organization, a local preferred resourceexchange network is established to achieve effective clustering node services. Forasynchronous and synchronous video-on-demand request, all participating nodes cachefragments of the video contents. Then the nodes provide video streaming service to the nodesin the cluster by delivering fragments. This not only reduces servers’ stress and backbonetraffic, but also achieves higher overall service capacity. Good scalability and quality ofservice guarantees are also supported to enable large-scale streaming media services.To recapitulate, this paper researches the key issues in distributed data transmission andmakes some contributions. It has important theoretical meanings and application values inimproving the network utilization.