Research On End-to-End Network Analytical Model With Heterogeneous Services Arrival Based On Tandem Queueing Network Theory

With the development of mobile communication technology, communication networkhas been rapidly development. People increasingly high demand for communication network,higher bandwidth, wider network coverage and higher QoS requirements, etc. Wireless accesstechnology emerges with unique advantage one after another. One of those, LTE is thelong-term evolution of3G technology, which has high data rate, low transmission delay andbig system capacity and wide coverage range, etc. LTE will have large commercialapplications. In recent ten years, under the dual drive of technology innovation andapplication requirements, the Machine type communication (Machine to Machine, M2M) hasreceived widespread attention. For M2M services, mobile communication networks are thebest bearing network. LTE is bound to be the most important bearing network of M2Mservices. The characteristics of M2M traffic are different from the existing H2H services(Human to Human), therefore, when huge amounts of M2M services want to access intoexisting and future communication network, there must be some influences on the network.Modeling and performance analysis of communication network are the most importanttheoretical basis and research directions in the field of communication. Existing studies aremostly based on single queueing model, however, because of the communication networkarchitecture is more and more diversified, the traditional single node queueing model isinappropriate to model the complex communication network. As the network operators andsubscribers care more about the end-to-end system performance metrics, in a communicationnetwork analysis model, we not only concern about the single node but also the end-to-endperformance. This paper proposes an analytical model based on tandem queueing theory, wechoose LTE as the wireless access network. Research on the performance analysis model withheterogeneous services, we can get some factors which may affect the QoS of network fromthe perspective of theoretical research, and these results will guide wireless access networksresource allocation scheme.First of all, the characteristics of M2M traffic has been analyzed, we configured thereasonable parameters for different services on the OPNET simulation software, then, wecollected the flow characteristic of specific scenarios, and fitted it into a MMBP processbased on the discrete LAMBDA algorithm, the MMBP source can be used to networkmodeling and performance analysis.Secondly, we simply modeled the LTE wireless network performance analysis model which bearing heterogeneous services simply. According to service characteristics we candivide them into three categories, namely, voice, video and data. For the M2M data servicecharacteristics are different from current data service: frequently sent packet, small amount ofdata and the sent cycle is fixed, therefore, we use two MMBP source processes to simulatethem separately. The dynamic network resources are allocated by LTE system based onservices priority and different rate of basic requirements, so a multiple queueing model withmultiple servers and limited capacity was proposed, each server has different service rate. Forvoice and video services, we modeled queueing system as MMBP/Geo/1/K queueing modelrespectively. For data services, this can be modeled as a2-MMBP/Geo/1/K queueing model,the superposition MMBP source aims to characterize different features of two kinds of dataservices primarily. In the service process of LTE system, we focus on the influence of HARQand channel environment, mapping these factors into the queuing model. By solving thisdiscrete-time Markov model, we can get the steady-state probability, and the performancemetrics of communication system. To evaluate each queuing system respectively, theinfluence of the signal to noise ratio, service rate and cache capacity to the performance of thesystem have been analyzed. At the same time, a performance function was put forward tomake a comprehensive evaluation of network performance.Finally, based on the theory of tandem queuing, the cable network performance analysismodel of communication network was modeled as multiple nodes tandem queueing model.This model can simulate the service process of data packets through multiple routers in theforwarding process. In the model, each node represents a router, because the services arrive arouter may come from multiple routing, thus service arrival process is a superposition source.A decomposition method was used to solve the tandem queuing network, we decomposed theproposed tandem queueing network into several independent sub-networks, analyze thedeparture process of each sub-network and then fit the departure process with another sourcemodel that is used as the input process of next sub-network. So the departure process analysisand parameter fitting method of sub-network with superposition source arrival are the keyproblem. By solving the queuing model, we can get the performance metrics, and analyze theinfluence of different factors on communication network performance. Combined with thecontent of the fourth chapter, we can get the end-to-end performance evaluations ofcommunication network with heterogeneous services arrive.