Queueing Network Modeling And Performance Analysis Based On Two-dimension Arrival Model

In the recent years, Internet of things is an attractive concept in the wordwide, which isregarded as the most promising business growth point.3GPP predicted7class and35kinds ofpossible M2M services, some of them are used commercially now. In the future, the numberof M2M terminals will reach trillion magnitude, the large scale implement of these serviceswill certain bring influences on the network and other services. The random traffic of M2Mservice will increase the source burstiness and concurrency, a huge number of heterogeneousTCP flows and long real-time network traffic makes service even more complex. Existingnetwork traffic models are unable to describe the M2M service, so it has a significance tostudy models and radio resource management theory for existing network service.In the M2M service communication transaction, flow is the basic service unit of machinetye communication, machine is sensitive to the flow characteristics of service. Therefore, inorder to get a comprehensive description of M2M service, accurately description of thecharacteristics of M2M service, association of modeling parameters of packet-level andflow-level have a certain significance.The research on queuing network modeling and performance evaluation for M2M smalldata service under packet-level, traffic arrival is based on OPNET network simulation. First,we use the discretization method to model small M2M data traffic flow as general distributionof Gm, packet arrival is modeled as Poisson distribution; we describe the M2M small dataarrival process from packet-level and flow-level, and study the self-similar characteristics, thescale expansion between the packet arrival and flow arrival and the center-biasedcharacteristics; By using the embedded markov chain approach, M2M small data queuingmodels are established from packet-level and flow-level. We also construct a continuous-timetwo-dimensional markov chain and get the QoS parameters; According to the packet loss rateand network throughput, we analysis the advantages and disadvantages of flow-level accessand packet-level access, furthermore, we research on parameters configuration of M2M smalldata packet-level flow-level joint access control.The research on queuing network modeling and performance evaluation for M2M videosurveillance service under packet-level, traffic arrival is based on OPNET network simulationdata collected and we study the queuing model for M2M video surveillance service under packet-level and flow-level arrival. First, we analysis the self-similar characteristics of M2Mvideo surveillance service; Next, by using LAMBDA algorithm, M2M video surveillanceservice flow arrival is modeled as MMPP distribution and packet arrival as Poissondistribution; we explore the scale expansion and the center-biased characteristics of M2Mvideo surveillance service under packet-level and flow-level, M2M video surveillance servicequeuing model under packet-level and flow-level arrival is established and continuous-timethree-dimensional (first dimension for the number of packages, the second dimension forflow number, the third dimension for the MMPP state of source) markov chains are proposedand the QoS of service are obtained; Last, according to the packet loss rate and networkthroughput, we analysis the advantages and disadvantages of flow-level access andpacket-level access, furthermore, we research on parameters configuration of M2M videosurveillance service packet-level flow-level joint access control.The research on queuing network modeling and performance evaluation for M2M andH2H mixed service under packet-level and flow-level arrival is based on theory ofsuperposition of continuous time source. We use Konecker theory and Descartes rules toestablish the M2M and H2H mixed service packet-level, flow-level arrival model. Flowarrival is modeled as MAP distribution and packet arrival as Poisson distribution. We studythe scale expansion and the center-biased characteristics between packet arrival and flowarrival. M2M and H2H mixed service queuing model under packet-level and flow-level isestablished and continuous-time three-dimensional(first dimension for the number ofpackages, the second dimension for flow number, the third dimension for the MAP state ofsource) markov chains are proposed and both the joint probability distribution and the QoS ofservice are obtained. Last, we explore three problems for the network resource managementwith mixed service, which are, first is research for and access parameters configuration ofM2M and H2H mixed service from packet-level and flow-level based on the packet loss rateand network throughput, second is research on the traffic equilibrium problems for M2Mservice and H2H service in the network according to two service network throughput curve,third is the best network bandwidth allocation scheme under specific network environment byintroducing the concept of performance function.