Research On QoS-aware Routing Protocols For Ad Hoc Network

As the developing of wireless technology and improving of terminal devices, Ad Hoc has become one of the most important areas in wireless communication. In order to fulfill requirements of different applications, it is necessary to provide QoS on Ad Hoc instead of just best-effort service. However, characteristics of Ad Hoc network such as dynamic time-varying topology, limited resources and competed wireless channel bring a great many difficulties and challenges.QoS-aware routing is the most essential part of the whole QoS system.Its goal is to determine suitable routes for flows with the knowledge of available network resources and application requirements. Known protocols, which vary in the way of guiding search of routes, can not scale well because they need to conserve additional network information or are based on table-driven routing protocols.Therefore, this dissertation focused on devising a protocol, which could effectively guide route search based on maintenance of less topology information without consumptive flooding.A routing protocol named MBQR (Multipoint-Based QoS Routing) which is based on restricted flooding and intelligent search was presented in chapter 3. Network nodes collect local topology information within 2-hop range distributedly. Upon this, they choose MPR (Multi-Point Relay) sets which could provide better QoS support from 1-hop neighbors according to certain strategies. When there are application datas to transport, MPRs are used to restrict flooding and guide search intelligently in order to find an appropriate route quickly with less costs. The simulation results showed reduction in latency and improvement in bandwidth of established route compared to QAODV which is based on simple improvement of AODV. Meanwhile, MBQR was more adaptive to dynamic environment.Resources estimation and routing maintenance were also included in the routing framework, resulted a revised protocol called MBQR-MC (Multi-Constraint MBQR) in chapter 4. In this edition, available bandwidth of a node is estimated using existing hello algorithm by counting bandwidth consumption of nodes within the interference range. As for one way end-to-end delay, it is evaluated considering clock difference in MANET. Node stability, which reflects real network situation, is also introduced to improve MPR selection. MBQR-MC was proved to evaluate resources effectively and chose more stable MPRs, which actually reduced packets loss during simulation. Detection of QoS violation is as important as traditional route break detection, which was unfortunatelly not discussed in most QoS-aware routing protocols. MBQR-MC embraced an intact route recovery framework and implemented quick route recovery based on QoS violation detection in destination and redundant routes. Simple simulation showed that, MBQR-MC detected QoS violation when the current path could not satisfy application's requirements and swiched to an appropriate one.