A Model Of QoS Guaranteeing Based On QoS Information Statistics In Wireless Ad Hoc Network

The wireless Ad Hoc network is a centerless and self-organization network, which provides many merits, such as flexible networking and autonomous operating, thus being widely applied in military communication, earthquake rescue and other circumstances where networks cannot be easily established but require rapid networking. With the development of the wireless Ad Hoc network technology, the communication service is expanded from traditional data service and voice service to multimedia service with Qo S requirement. Because of the time variability of the wireless Ad Hoc network' topology and link state, the nodes' limited energy, and so on, it becomes more complex for the wireless Ad Hoc network to achieve Qo S guaranteeing. Qo S guaranteeing of the wireless Ad Hoc network requires the cooperation of each protocol layer. Therefore, the suitable Qo S guaranteeing model for the wireless Ad Hoc network is the key for the research of the wireless Ad Hoc network' Qo S guaranteeing.Firstly, the thesis summarizes the existing Qo S models in the Ad Hoc networks, such as Int Serv model, Diff Serv model, EPDS model, SWAN model and its improved modelPSWAN model, and analysis their operation mechanism, suitable situation, advantages and disadvantages. At the same time, by insulting many literatures, the thesis summarizes the Qo S guaranteeing technologies, such as IFS differentiation and real-time ECN, and analysis their operation principle in detail.On the basis of the above work, the thesis puts forward QIS model, a Qo S guaranteeing model based on Qo S information statistics. QIS model is a distributed, stateless Qo S model, which does not require nodes to exchange any control information. The basic idea of QIS model is to employ the end-to-end Qo S information carried by the data-replying packet, thus establishing the local Qo S information statistics table. When sending a packet, the node refers to the local Qo S information statistics table and employs the adaptive on-demand weighted algorithm to realize packet admission management, packet scheduling, congestion control, and so on, to ensure different packets' end-to-end Qo S demand. Compared with other Qo S models, the admission management module of QIS model can provide admission management based on the packet. At the same time, admission management can be employed by each node in the transmission path, so QIS model can adapt to the Ad Hoc network better. Packet scheduling module combines priority scheduling, probability scheduling with the remaining Qo S demand scheduling, so QIS model can ensure the Qo S demand of the high priority packets and prevent the low priority packets from starvation. Simultaneously, QIS model can somewhat decrease the loss rate. Moreover, QIS model expands the types and priorities of the packets to achieve the packets' service differentiation. In addition, QIS model is a fully distributed model, because all the judgments and decisions are completed in the node. Meanwhile, QIS model employs Qo S information statistics and does not need to send any additional control packets to guarantee packets' Qo S demand, which does not increase the control overhead of the network.Finally, the thesis analyzes the QIS model by setting up a simulation platform in the wireless Ad Hoc networks with the OPNET simulation software. Because QIS model achieves Qo S guaranteeing hop by hop and is based on the Qo S information statistic, the simulation results show that, compared with EPDS model and PSWAN model, the QIS model can realize the packets' service differentiation reasonably, and guarantees the delay, packet loss rate and other Qo S requirements, without increasing the control overhead, both in the plat Ad Hoc network scenarios and the hierarchical Ad Hoc network scenarios.