Research On Key Problems Of Statistical Time Characteristic For Embedded Internet Based On Macroscopic Topology

The Internet has been growing rapidly in recent years, more and more embedded real-time systems have been accessed the Internet, which make man’s perception and control ability get unprecedented expanding. The embedded system can’t make sure it’s time security when it accesses to the Internet because of high complexity and uncertainty of time delay in the Internet. Every step in the embedded system has a time limit range, so the system shall crash once time transfinite break out. Facing the mass and complex Internet at present and the deterministic time limits of embedded system, the interaction of embedded system and Internet faces huge challenges.There are two kinds of ways for solving the above problems. One method is analysis and solving problem through the micro perspective, such as Redesign of routing protocol, Design of congestion control strategy and Access gateway selection etc. Another method is parsing the structure of Internet through macroscopic topology, which getting the corresponding delay characteristics to guide real-time system to access the Internet. The first point is for the certain level of network delay. There are many factors leading to uncertain time delay in the Internet, for example, the packet processing delay, conflict transmission delay and the routing service delay. So the results dropped from the first method can not be directly applied in practical. The second method is based on statistical theory. Through the analysis of massive Internet data, the complex delay characteristics of the Internet macroscopic topology were abstracted. According to the characteristics, topology modeling method can be designed and providing the basis and guidance for real-time Internet analysis.Firstly, the measurement technology of CAIDA macroscopic topological measurement project and the acquisition method of topological data are used in this paper, so that we can detect, combine and pretreat the data which providing the correct data for later chapters accurately. Second, we renormalized the Internet by the core-thresholding coarse-grained process and Study the self-similar fractal characteristics of Internet routing topology. The self-similarity of Internet topology is analyzed in terms of degree distribution and degree correlation and clustering. The qualitative and quantitative analysis of degree distribution exponent of different KC networks, Scale invariance of degree-related characteristics of Cumulative average value of neighbor nodes and fractal dimension indicates that there is self-similarity between part and whole, and part of routing level network, so as to indicates that the Internet routing topology has fractal feacture, that the fractal complexity of network topology has an impact on delay, which Provides the necessary foundation for the analysis of delay characteristics in this paper.Secondly, work in this paper is mainly to analyse the behavioural characteristics of network delay, by using the abundant data collected by CAIDA ARK project. It is found that the network delay is typically distributed by multi-modal and heavy tailed. The behavioural characteristics of network delay has little change over time and strong self-similarity, which show that network delay affected by the fractal complexity of network topology also has fractal characteristics.Then three factors which may affect the network delay were studied, such as the traveling diameter, the actual geographic distance and the max link delay. And it is shown that the behavior distribution of traveling diameter closes to normal distribution and network delay has nothing to do with traveling diameter. About 90% of max link delay taken up one third of network delay is also named bottleneck delay. The max link delay is the factor affecting the network delay. Meanwhile, analysis on the evolutionary of bottleneck delay in the last five years shows that both the average of network delay and bottleneck delay increase year by year and distribute geographically.Thirdly, Based on analysis of the structural complexity of Internet macroscopic topology, the Internet macroscopic topology was studied by k-core decomposition. A hierarchical model LM based on k-core delay characteristics was proposed.The experiments shows that this model can reproduce on Internet on most properties and has got topology complexity close to that of Internet. The analysis shows the effectiveness and rationality of LM model. Then the paper introduced the application of combination of LM model and NS-2 platform, which provide the platfoem and mothed to the application of Embeded Internet.Finally, considering the dominant delay has a great influence on the behavior of spread network behavior, the paper furtherly study on the behavior characteristics of dominant delay, explain the cause and the influencing factors of dominant delay, So as to provide the basis for connecting embedded systems to the Internet. First of all, comparing transmission behavior on different traveling time and interval of network, it was found that the path, which is largely different from each other on traveling time, has similar proportion of dominanting delay for traveling time. It indicates that there is no necessary relation between the dominant delay’s influence and traveling time. But the dominant delays themselves differ greatly on numerical values, which causes great difference among the similar diameter’s traveling time and directly manifested in the multi-modal distribution of Internet traveling time. Then, by mapping the nodes from IP level to AS autonomous domain, the further analysis of dominant delay on AS autonomous domain is taken which discovers that dominant delay tends to appear inside the AS autonomous domain and it explains the reason why dominant delays seldom occur between the AS autonomous domains. The dissertation discusses the main reason which causes the dominant delay by mapping the IP node to its geographical location. The results show that fractal complexity of Internet topology mainly affects the scale of the dominant delay. Then interactive condition of embeded system connected to the Internet is given in this paper.This article analysis the characteristics of Internet delay from the Internet topology.The conclusions are of guiding significance to the uncertain delay problems of embeded system accessing the Internet.