Study On Network Resource Allocation Mechanism And Algorithm With Guaranteed QoS

The increasing number of users and emerging of large bandwidth multimedia applications make resource allocation become more and more important. In wireline networks, selfish users always try to use more bandwidth. This severely affects the QoS (Quality of Service) of other users and decreases the overall network's performances. Therefore, new bandwidth allocation algorithms are needed to solve the unfair bandwidth allocation problem. In wireless networks, most of the licensed spectrum is underutilized. The FCC (Federal Communications Commission) suggested improvements on spectrum usage to efficiently allocate spectrum to unlicensed users without impacting the primary licensees. This has stimulated a swarm of researchers to work on the dynamic spectrum allocation.First, we study the bandwidth allocation in wireline networks with selfish users. A new algorithm based on VCG (Vickrey-Clarke-Groves) auction-based mechanism is proposed. The dominate strategy incentive compatible property and the Nash Equilibrium solution's existence and uniqueness are proven. Based on this, we introduce a discrete stochastic learning algorithm to help users to make correct decision. The convergence of the algorithm is also analyzed.Second, we mainly concern the spectrum allocation in underlay cognitive radio networks. Both the interference temperature constraints and the QoS of the unlicensed users are considered. The first approach is to develop a joint admission control and power control algorithm. In order to decrease the interference temperature violation probability, three removal criterions are proposed under three different conditions. It can satisfy both the QoS constraints for accepted unlicensed users and the interference temperature constraints for primary users in dense networks. The second approach is to formulate the spectrum allocation problem between primary users and unlicensed users as a non convex optimization. By introducing variable substitution and transformation, we derive a distributed random access and power control algorithm that can achieve global optimal solution to the original problem. Convergence of the algorithm is proven theoretically.Numerical simulation results are provided to demonstrate that our proposed algorithms meet the design purposes. It's shown that the proposed allocation algorithms can guarantee the QoS of users and improve the overall system performance.