Research On Quality Of Service For Multi-Service Transmission Over Mobile Hotspots

With the rapid development of the Internet service and the maturity of the broadband wireless access technology, both the requirement of the information from the people and the information obtaining method are changed. The original method by which only fixed terminal or a single mobile node connects the network can not fully satisfy the need of the people. On the public transport or in a personal area network, the people wants to set multiple inter-connected mobile terminals as one unit which can moves within the network and do the information interchange with the mobile Internet. These mobile terminals and the mobile router which is responsible for the information re-laying constitute a mobile hotspot network. Mobile hotspot network can not only achieve high utilization efficiency of the wireless resource, but also can maintain the connection in some extremely severe wireless environment. As a result, mobile hotspot network has recently attracted lots of attention from the academic world and industry. However, the introduction of the mobile router has change the one-hop transmission into two-hop or multi-hop transmission, which makes it harder to fully guarantee the Quality-of-Service (QoS) requirement. Therefore, how to effectively satisfy the differentiated QoS require-ment in the mobile hotspots is a scientific problem and has practical application value.Based on the theoretical methods such as queueing theory, Markov chain and s-tochastic Petri net, the dissertation has done several research work on the QoS of multi-service transmission over the mobile hotspots. The main innovations of this thesis are listed as follows:1) The video transmission over the mobile hotspots is modeled by using queueing theory. We have obtained several packet-level performance metrics such as packet delay, jitter and packet loss probability. The impact of wireless local area network (WLAN) transmission over the backhaul link is also investigated. Further, we have proposed an accurate and simple discrete-time Markov-modulated arrival process in order to model the video service which arrives at the mobile router by going through the WLAN.2) Compared with the normal streaming services, real-time service such as video con-ference has a more strict requirement on the packet delay. A delay-based proactive packet discarding transmission scheme is presented for the transmission of real- time service. The proposed scheme can achieve better system performance such as lower packet delay and jitter by avoiding the limited wireless resource wasted on the data which is useless for the upper-layer application.3) The connection between mobile router and base station (wireless access point) is called backhaul link. Considering the network selection issue for the backhaul link in a heterogeneous wireless network, the dissertation has proposed a stochastic Petri net model which can be generalized applied to model different network selection algorithm. We have also analyzed several system-level performance metrics such as new user call blocking probability, handover dropping probability and number of users within the network.4) The application of stochastic Petri net and Markov chain can hardly avoid the "state space explosion" problem. Considering this issue, the dissertation has presented a network decomposed-based stochastic Petri net for the modeling of network selec-tion algorithm. Corresponding iteration algorithm for the Petri net model is also provided. The numerical result has shown that our proposed method can reduce the computation complexity while guarantees the accuracy. The presented method can be further applied for the analysis of lager scale network.5) In order to satisfy the differentiated QoS requirement of different service, the dis-sertation has presented a cross-layer design for multi-service transmission over the backhaul link. In the physical layer, the corresponding adaptive modulation and coding schemes are selected based on the different service requirement on the pack-et delay and packet error probability; In the data link layer, the services are granted with different priority according to the service importance and the non-preemptive prioritized packet transmission scheme is adopted to guarantee the QoS. Based on the analytical result from the Petri net model for the network selection algorithm, we have further evaluated the impact of different network selection algorithm on the QoS of the mobile network.