Research On Key Issues Of Resource Allocation In Cognitive MIMO Networks

With the rapid development of wireless communications, the demand for spectrum isaccelerating, which makes spectrum a scarce resource. However, the fixed spectrum allocationpolicy currently adopted renders a great portion of the licensed spectrum severely under-utilized,which has become a bottleneck for the deployment of future wireless communications. In recentyears, the rising cognitive radio (CR) technology enables unauthorized users or the so-calledsecondary users (SUs) sharing spectrum with licensed users or the so-called primary users (PUs),thus can improve spectrum utilization and provide bandwidth for new emerging wireless servicesand applications.Multiple-In Multiple-Out (MIMO) brings spatial diversity and multiplexing, can improve thetransmission reliability and spectral efficiency in wireless communication systems, and is one of thekey technologies for future wireless communications. By introducing MIMO into cognitive radio,cognitive radio will be granted etra spatial degrees of freedom besides time and frequency whileaccessing PU’s spectrum. Therefore, it appears particularly important to combine MIMO andcognitive radio, investigate MIMO networks in cognitive radio environment, namely cognitiveMIMO networks.For the purpose of guaranteeing the QoS of PUs, SUs should protect normal signaltransmissions of PUs. In principle, SUs can utilize the licensed spectrum of PUs in the twofollowing manners; one is called opportunistic spectrum access (OSA), SUs only access thelicensed spectrum which is not currently occupied by PUs, and the other one is concurrent spectrumaccess (CSA), SUs coexist with PUs in the licensed spectrum but should restrain the interferencecaused to PUs under a predefined low threshold, i.e., satisfy the interference/interferencetemperature constraint. In this dissertation, we focus on cognitive MIMO networks, in which theSUs are equipped with multiple antennas or virtual multi-antenna arrays are formed through usercooperation. We investigate the resource allocation and management problems in cognitive MIMOnetworks. More specifically, on the premise of guaranteeing the QoS of PUs, we deal with problemsof optimally deploying SUs’ communication resources such as transmit power, beamformingweights and precoding matrices in order to most efficiently utilize the licensed spectrum, e.g., achieve maximum information rate, signal to noise ratio or throughput. The main work andcontributions of this dissertation can be generalized as follows.(1) Optimum transmit covariance matrix design for cognitive MIMO multiple accesschannels (MACs). We consider a cognitive MIMO MAC consisting of multiple multi-antenna SUtransmitters and a multi-antenna common receiver. We investigate, subject to transmit power andinterference temperature constraints, how to design the transmit covariance matrices of SUs suchthat the sum-rate of cognitive MIMO MAC is maximized. By exploiting the partial dualdecomposition technique to relax the interference temperature constraint, the original problem wasdecomposed into more tractable subproblems. An iterative algorithm, in which the dual variableupdate and the iterative water-filling computation were performed alternately, was proposed toobtain the optimum transmit covariance matrices that achieved the maximum sum-rate.(2) Waveform adaptation for cognitive MIMIO interference channels (ICs). We consider acognitive MIMO IC formed by multiple independent SU links. Each link consists of a pair ofmulti-antenna SU nodes. From a non-cooperative game theoretic viewpoint, we investigate theNash equilibrium (NE) problem of SU links trying to maximize individual information rates bywaveform adaptation subject to transmit power and interference temperature constraints. Bytransforming the NE problem into equivalent variational inequality problem, we prove the existenceof the NE and provide sufficient conditions for the uniqueness of the NE. A decentralized iterativewater-filling algorithm (IWFA) with punishing price, called MIMO-CR IWFA, is proposed to solvethe above NE problem; the pricing mechanism is used to satisfy the interference-temperatureconstraint while SU links achieving the Nash equilibrium. The conditions for the convergence ofMIMO-CR IWFA are also provided. Simulation results show MIMO-CR IWFA can satisfy theinterference-temperature constraint perfectly and is fast convergent.(3) Optimal source power control and relay cooperative beamforming in cognitive relaynetworks. We consider an amplify and forward cognitive radio network consisting of two terminalnodes and several relay nodes, in which, to achieve the spatial diversity gain, cognitive relay nodesform a virtual multiple antenna array and forward cognitive source node signal through cooperativebeamforming. Two interference temperature constraints, strict and loose, are considered, the formerconstraint requires the interference generated by cognitive nodes in both broadcast and relay phasesunder a predetermined interference threshold and the latter one requires the average interference generated in the two phases under the threshold. We investigate, subject to interference temperatureconstraint or both interference temperature and transmit power constraints, how to optimizecognitive source transmit power and cognitive relay beamforming weights in order to maximize thesignal-to-noise ratio (SNR) at the cognitive destination.We show the SNR maximization problem subject to interference temperature constraint can beconverted into a generalized Rayleigh quotient problem, the optimal relay beamforming weights areequal to the eigenvector corresponding to the maximum eigenvalue of a matrix associated withchannel coefficients and source transmit power. For the strict case, analytical solutions of theoptimal source transmit power and relay beamforming weights are provided. For the loose case, astringent interval containing the optimal source transmit power is deduced and a steepest descentbased algorithm is used to achieve the optimal solution efficiently.For the problem subject to interference temperature and transmit power constraints, we firstaddress the problem of optimal cooperative beamforming under a given source transmit power. Thisproblem is decomposed into a series of quadratically constrained quadratic programming (QCQP)problems through Dinkelbach type method. By utilizing semidefinite relaxation technique, theQCQP problems are relaxed into more tractable convex semidefinite programming (SDP) problemswhich can be solved by interior point method. Secondly, an evolutionary computation techniquebased on particle swarm optimization is used to achieve the optimum source transmit power.(4) Spectrum sharing with cooperation between PUs and SUs based on orthogonalspace-time block coding (OSTBC). According to OSA model, the secondary user can access thechannel only when sensed idle. We investigate the spectrum-sharing problem in a basic CR scenarioconsisting of a PU link and a SU link, propose a spectrum-sharing scheme in which transmissiondiversity of the PU is formed by the SU actively acting as a cooperative relay of the PU throughOSTBC, and thus the PU’s probability of outage is reduced and the SU’s available idle slots areincreased. We derive the closed-form expression of the SU link’s throughput in our proposedscheme.