UBC Theses and Dissertations
Dynamic Weight Mapping Adaptive Routing (DWMAR) for IP networks Xia, Feng
This thesis addresses the difficulties of both Shortest Path First (SPF) and traditional adaptive routing protocols. SPF does not respond to the link flow change so that it cannot find optimized routes for the network. Traditional adaptive routing protocols do respond to link load changes. However, the network flow varies so widely that a fixed global weight mapping function cannot be suitable for all the scenarios. That is why traditional adaptive routing protocols are prone to being unstable and counteract the benefits obtained from the link weight adaptation. Dynamic Weight Mapping Adaptive Routing (DWMAR) is then proposed to provide an optimized and stable adaptive routing protocol. DWMAR proposes an online dynamic weight mapping function which is customized for every link, so that more links will work under an efficient traffic load and have more chances to reach a stable state. To make up for the unstable nature of adaptive routing, DWMAR adopts different adaptive policies on different link loads. Theoretically, DWMAR is a stable algorithm. Finally, the performance of DWMAR is tested in simulations. The results are compared to the non-adaptive SPF algorithm and a fixed weight mapping adaptive routing, Load Sensitive Adaptive Routing (LSAR). DWMAR shows better performance than the other two algorithms in terms of network efficiency. DWMAR also shows its stability in the simulation results.
Item Citations and Data