Haar Wavelet prediction-based Fair Queuing

Since the Internet is developing rapidly in size and in complexity as well, it is very hard to adjust quality of service (QoS) parameters. In order to guarantee service quality, efficient resource reservation is needed. The allocation of resources requires a dynamic reconfiguration mechanism that configures a switch, taking into account the incoming traffic fluctuations. This paper presents a new efficient bandwidth allocation algorithm called Haar wavelet prediction-based fair queuing (HWFQ) which is designed to schedule bandwidth adaptively, fairly and efficiently among all competing flows. The key feature of the new scheduling scheme is to predict the scheduling weights, the HWFQ is first scheduler which have following properties: (1) a new idea of using Haar-wavelet prediction scheme in multi-queue systems; (2) an adaptive on line prediction scheme capable in reliably and precisely estimating the incoming traffic; (3) the proposed algorithm has no need to prior knowledge about the incoming traffic; (4) complexity of HWFQ algorithm is very low; therefore, it simply can be implemented in routers and switches. The performance of the proposed algorithms is evaluated using ns-2 simulator and compared with weighted round robin