Improvement And Performance Evaluation Of Dynamic Source Routing Protocol And Its Test-bed For Wireless Ad Hoc Network

Among the numerous routing protocols of the wireless Ad hoc networks, thedynamic source routing (DSR) protocol proposed by Carnegie Mellon University is arepresentative one. However, the control overhead of the protocol's routingmaintenance procedure is too expensive, and the detection of the link failure is notaccurate enough. Therefore, it cannot provide routes with necessarily high bandwidth,low latency or certain reliability for the upper layer traffics. In order to transfer thereal-time multimedia traffic which need high bandwidth and low latency, we improvethe DSR protocol with proposing of a new link-failure detect algorithm. According tothe new algorithm, a DSR pre-node of a link should actively request ACKs from itssub-node at dynamicly changed intervals, and the overtime of DSR's ACK should bechanged with the change of the ACK's RTT. The improved DSR greatly increasesboth the available bandwith for upper layer traffics and the accurate level of thedetection of the link failure.At the meantime, to integrate the Ad hoc networks with other networks or theInternet, we extend the original DSR, with introducing of the concept of the WirelessGateway node and the according new procedures, such as Gateway Discoverytogether with Gateway-failure handling. With the extended DSR protocols, the nodesin Ad hoc network are capable to access the Internet. The extension to DSR greatlyenhances the application of Ad hoc network.In addition, we've developed the test-bed of the new DSR with our improvementand extension, and in order to make it serve multimedia traffics better, we add apriority queue to it. In the queue DSR control packets and UDP packets have thehigher priority. Finally, we made extensive comparing experiments on the test-beds ofour proposal and the original DSR. The results show our betterment to DSR makes theprotocol get a notable performance enhancement and more suit the transfer of thereal-time traffic. We have also done a series of outdoor experiments with the newprotocol's test-bed. In the experiments we captured the real-time videos with audios inthe Ad hoc network and sent them to an external network through our DSR WirelessGateway. The success of the experiments shows that after our improvement andextension to the original DSR protocol it has been able to support the real-timemultimedia traffic's transfer very well and been getting more useful in our daily life.