Research On Dynamic Resource Allocation For Underlay Cognitive Radio Networks

With the rapid growth of wireless applications and mobile devices, there has been anincreasing demand for spectrum resources. Due to the scarcity of radio spectrum, theconflict between scarce spectrum resources and the increasing demand of people has beenvastly more severe. Since cognitive radio technology can well solve the issue of spectrumshortage, it has been widely concerned and has been a hot topic of wireless communicationfield.Cognitive radio is an intelligent wireless communication technology which canimprove spectrum utilization efficiency. In cognitive radio networks, there are two kinds ofusers: secondary user (i.e., unlicensed user or cognitive user) and primary user (i.e.,licensed user). Cognitive radio technology allows secondary user adjusting its transmissionparameters (i.e., transmit power, carrier frequency, modulation mode) for accessing andusing idle licensed spectrum bands, which can achieve spectrum sharing betweensecondary user and primary user as well as improve spectrum efficiency. There are two keytechnologies in cognitive radio: spectrum sensing and resource allocation. According to thespectral analysis of surrounding environment, cognitive device can know the activity ofprimary user and find when and where there is idle spectrum resource. Spectrum sensingaims to find available spectrum resource from licensed bands as more as possible.Additionally, resource allocation (i.e., power control) can achieve spectrum sharing. Basedon the sensing results, secondary transmitter or cognitive base station adjusts transmitpower to satisfy communication quality without affecting the normal communication ofprimary user simultaneously. In this thesis, resource allocation strategies are proposed tosolve the aforementioned problems from different levels. Specifically, distributed resourceallocation and robust resource allocation for underlay cognitive radio networks areintensively studied. The main contributions of this dissertation are summarized as follows.First, distributed resource allocation problems for multiuser underlay cognitive radionetworks are investigated to improve transmission efficiency, reduce message exchangeand computational burdens. On the one hand, a globally distributed resource allocationalgorithm with average interference temperature constraints is proposed under SINRconstraints of secondary users, maximum transmit power constraints and interferencetemperature constraints of primary users while both the convergence of algorithm and therange of parameter are also analyzed. The proposed method is superior to the traditionalmethod in the way of total throughput of system and SINR of secondary users. On theother hand, in order to solve the problem of each user’s fairness caused by average interference constraint, a mixed resource allocation strategy via position information ofuser is presented for different network scenarios. According to the size of cognitivenetwork, the improved algorithms with the average/weighted interference temperatureconstraint can achieve the switching control to adapt dynamic resource allocation ofdifferent network sizes.Then, robust resource allocation approaches with bounded uncertainties are studiedfor achieving the seamless connection of cognitive radio system subject to parameterperturbation. To overcome the effect of estimation errors, quantization errors and channeldelays for nominal optimization model in wireless communication network, robust totalthroughput maximization and robust total transmit power minimization problems arestudied. For the issue of throughput maximization, based on the worst-case approach, theoriginally semi-infinite programming optimization problem with the consideration ofchannel and interference uncertainties is transformed into a feasible and convex formsolved by using Lagrange dual methods. The impact of uncertainties on the transmit powercontrol and system throughput is analyzed, showing that the performance of scheme can besignificantly improved in terms of preventing the interference received by primaryreceivers from exceeding the interference temperature levels. However, the communicationquality of primary users with the non-robust scheme is destroyed under uncertainparameters. Additionally, in order to save energy and improve the operating life ofcognitive radio network, the robust transmit power minimization problem is alsoinvestigated under channel uncertainties modeled by Euclidean ball-shaped uncertaintysets. Simulation results demonstrate that the performance of the designed method has goodrobustness in term of SINR of secondary receiver.Finally, robust resource allocation problems under probability constraints (i.e.,distribution uncertainty) are studied to improve the fault-tolerant ability of wirelesscommunication system under uncertainties. The statistical approach as another method tosolve uncertain optimization problem is very necessary, since the aforementioned robustresource allocation algorithms require to perfectly know the upper bound of uncertaintywhich is impossible for some practical environments due to the random nature ofcommunication system (i.e., random channels, random accessing users). Therefore, twodifferent stochastic optimization problems with probabilistic SINR and interferenceconstraints are presented to solve robust resource allocation problems for underlaycognitive radio networks. Firstly, by the assumption of perfect knowledge of distributionmodels of errors (i.e., exponential distribution), a robust probabilistic distributed powercontrol algorithm for the total transmit power minimization problem is proposed under theconstraints that the satisfaction probabilities of both interference temperature of primaryusers and SINR of secondary users exceed some thresholds. The robust optimizationproblem is converted into a convex one solved by convex optimization theory and iterativeupdate schemes with forgetting factors. Simulation results show that the convergence of the proposed algorithm is faster than that of traditional sub-gradient method and gradientprojection algorithm. The performance of the proposed scheme is superior to the robustresource allocation algorithm under worst-case method in terms of robustness and energyconsumption. In addition, another robust resource allocation algorithm based on imperfectknowledge of statistical models of uncertainties is studied while the statistical model ofuncertainty is difficult to obtain. Based on a distribution-free method, a robust resourceallocation scheme with outage probability constraints of primary users and secondary users(i.e., probabilistic interference constraints, probabilistic SINR constraints) is proposed toachieve total transmit power minimization of secondary users. Based on the minimaxprobability machine, the originally robust optimization problem is reformulated as asecond cone programming problem resolved by interior-point method. An adaptive updatealgorithm is designed to estimate actual mean and covariance matrix of uncertainparameters. The simulation results demonstrate the effectiveness of the proposed schemeby comparison of the robust worst-case resource allocation approach and the robustresource allocation approach via Gaussian distribution model.