Research And Simulation Of SWAN QoS Model In Ad Hoc Networks

Ad Hoc network is a special multihop network without a centralized administration. Investigations on Ad Hoc networks have received considerable interest in recent years due to its capability to deploy quickly without infrastructure. Moreover, because of the rising popularity of multimedia applications and potential commercial usage of Ad Hoc networks, supporting QoS for different type of service over Ad Hoc networks has become a key and urgent issue. Characteristics of Ad Hoc networks such as lack of central coordination, dynamically varying network topology, and limited availability of wireless resources also make QoS provisioning very challenging in such networks.This thesis gives a discussion over the question of providing QoS mechanism on Ad Hoc networks, and puts a focus on a simple and flexible QoS model specially designed for Ad Hoc networks--SWAN QoS model. According to the limitation and shortcoming of this model, two improved algorithms are proposed in this thesis. Furthermore, a simulation model of the SWAN and its improved system based on the OPNET platform is setup. Through the means of simulation a serials of data and curve are gathered, and the QoS guarantee supplied by the SWAN and its improved system for the data stream in the network gains validation.SWAN is a stateless network model which uses distributed control algorithms to deliver service differentiation in mobile wireless Ad Hoc networks. Rate control is adopted for UDP and TCP best-effort traffic, and sender-based admission control is applied to UDP real-time traffic.SWAN uses explicit congestion notification (ECN) to dynamically regulate admitted real-time traffic in the face of network dynamic brought on by mobility or traffic overload conditions. A novel aspect of SWAN is that it does not require the support of a QoS-capable MAC. Intermediate nodes (not include source nodes) do not keep any per-flow or aggregate state information in SWAN wireless networks. That is to say, the model is"stateless". As a result, there is no need for the introduction of complex signaling or state controlmechanisms to establish, update, refresh, and remove per-flow state information. Changes in topology and network conditions, even node or link failures do not affect the operation of the SWAN control system. This makes the system simple and robust.However, the SWAN model has its latent disadvantages. Providing QoS guarantee in Ad Hoc networks is a complicated task and should be viewed from the point of systematism. Owing to the particularity of this type of networks, the assured QoS couldn't be completely obtained. The existing SWAN model just gives a referenced quality of service model. The mode of merely relying on this model to support QoS for the entire network without other layers'QoS ensurence lacks of the theory that the element in the Ad Hoc networks QoS system should cooperate with each other. The model seems not enough for the networks which bear high load traffic requiring high QoS requirement. At the same time, it is not wise to ignore other layers'technology; and making SWAN model work well with every layer in the protocol could show its predominance and efficiency better. Based on the above analysis, two improved methods are proposed to the SWAN model in the thesis with the aim to make the QoS system more credible and achieve best performance of the network--SWAN model associated with DSR QoS protocol of IP layer and SWAN model associated with QoS technique of MAC layer. SWAN model associated with DSR QoS protocol of IP layer means that the route request of DSR protocol adds the function of detecting bandwidth along the path to its former operation; this kind of modifying can avoid repetition of the detection. Another improvment is the changing of backoff algorithm in MAC layer. The traffic in the network is differentiated into three types--real-time traffic, degraded real-time traffic, and best-effort traffic, and the priority of these three types is from high to low. High priority traffic is assigned a short backoff time, and thereby has a stronger capability of occupying the wireless channel than the best-effort traffic.This could save the degraded real-time traffic to some extent, and arrive at the goal of smooth the delay of real-time service.At last, OPNET network simulation software is utilized to set up a simulationsystem of SWAN and its improved model on the platform of exsiting Ad Hoc networks. Three aspect of experiment is carried out: (1) looking into the affection of SWAN and its improved model to the performance of Ad Hoc networks. Traffic in the network is divided into two types: best-effort and real-time. Test the operation of the network and observe the changing of the parameters, discuss the infection of the SWAN model to the Ad Hoc networks. (2) testing the SWAN model associated with DSR QoS routing system. Video conferencing and FTP service is deployed in the network. Analyse the performance of the SWAN model associated with DSR QoS routing network when the two service proportion is different. Discuss the different performance brought to the video conferencing service between the source network and the SWAN model associated with DSR QoS routing system. (3) on the base of former experiment, adding the QoS MAC protocol to the model. Observe the ability of video conferencing service in the network again, and validate the feasibility and the efficiency of the two modifying method.To sum up, in the thesis the key technology of providing QoS mechanism on Ad Hoc networks is researched, and among these technology a deep analysis is given to the SWAN quality of service model, including its principle, affection to the network, etc. Aiming at the shortcoming of the SWAN model, two improved methods are proposed to the SWAN model in the thesis: SWAN model associated with DSR QoS protocol of IP layer and SWAN model associated with QoS technique of MAC layer. A simulation system of SWAN and its improved model is set up based on the platform of OPNET software. Through the simulation experiment, a serial of data and curve are collected, and the QoS ensurence to the service in the network supplied by SWAN and its improved model is validated. Due to limited time, this system is not perfect, and the author will improve it during afterward works.