Dynamic Spectrum Allocation In Heterogeneous Wireless Networks

The evolution of wireless communication technology brings us a heterogeneouswireless networks (HWNs) environment. The heterogeneity includes: various radioaccess technologies (RATs) and radio access networks (RANs) co-exist; small basestations (BSs) overlap with macro BS; the types of service are more and more rich,and users distribution varies with time and space; multi-band and multi-mode terminal-s are developing with the cognitive radio technology. However, the spectrum resourceis limited. Traditional fixed spectrum allocation approach results in the paradox of re-source”scarcity” and”waste”. Therefore, this dissertation will study dynamic spectrumallocation (DSA) and the related interference control (IC) in the current HWNs scene.The main achievements and results of this dissertation are listed as follows:1. DSA algorithm based on clonal selection optimization is studied. Since stronginterference will happen in the overlapping area if different BS transmit in the samespectrum band, we first construct interference graph (IG) from BSs distribution topol-ogy. In addition, the channels with various widths for different RANs may overlapwith each other, which is reflected by using channel graph (CG). Note that in orderto achieve the purpose of IC, interfering BSs cannot be allocated the same or over-lapping channels. So based on the constructed IG and CG, the DSA problem is for-mulated as a nonlinear-constrained0-1integer programming. For this problem, theclonal selection optimization-based DSA algorithm is proposed. In the algorithm, wedesign a new coding scheme for the antibody, and add the antibody correction opera-tion for satisfying the constraints. Simulation results show that the proposed algorithmcould improve the network spectrum utility and spectrum utilization compared with thegreedy allocation algorithm.2. Assuming each RAN in the HWNs deploys hexagonal cells for the seamlesscoverage, we design the DSA algorithm under the spatial interference constraint. First,considering that users may randomly locate over the cell, we propose the spatial inter-ference constraint, in which the interference experienced by the user at a point is con-trolled below the level suffered when using fixed spectrum allocation in only a singlenetwork, and the proportion of the cell area where interference is controlled reachesthe required area coverage probability. Then under the spatial interference constraint,we formulate the DSA problem and propose two greedy heuristic algorithms for its solution. Simulation results show that the DSA algorithm could improve the networkspectrum utility for operators and increase the satisfaction rate of spectrum demandsfor BSs.3. Concerned about the HWNs scenario where BSs distribution isn’t regular ashexagonal cells, we further design a DSA algorithm with special interest on guar-anteeing the cell coverage probability. First, under the path loss channel condition,considering users spatial distribution and the ratio of different services, we proposecoverage probability-based IC model. This IC model guarantees SINR requirementsof different services and ensures the coverage performance of BSs. Under such an ICmodel, we formulate the DSA problem and design an algorithm for its solution basedon graph coloring. Compared with existing algorithms, the proposed algorithm can in-crease the network spectrum utility while effectively controlling the interference amongBSs and meeting SINR requirements of users.4. For the case of the shadow fading channel condition, we investigate the cov-erage probability driven DSA algorithm. Ignoring the background thermal noise andconsidering the impact of log-normal shadowing, we first approximate the probabilitydensity function of the interference signal strength and users SIR. Then we analyzethe coverage probability-based IC model under the shadow fading channel condition.Based on the IC model, we solve the DSA problem using the similar approach in the3rd part. Simulation results indicate that this DSA algorithm can increase the spectrumreuse, improve the network spectrum utility, control effectively the interference amongBSs and satisfy the SIR requirements of users under the actual channel condition.5. In spectrum-sharing hybrid macrocell-femtocell networks, the spectrum re-source is divided into resource blocks, and we design a queue-aware resource alloca-tion scheme. For the downlink transmission, BS maintains a separate queue for eachuser. To keep the length of every queue finite, i.e. stabilize the network, we formulatethe resource allocation optimization problem based on the Lyapunov stability theory.Then considering the queue length and the dominant interference, we construct theweighted interference graph, and solve the resource allocation problem based on themaximum weighted independent set algorithm. Through match the resource block-s and the traffic queue, simulations show the proposed queue-aware algorithm canguarantee the queue-length stability, and improve throughput in the finite-queue trafficmodel.