Research On Physical Layer Interference Management In Cognitive Heterogeneous Networks

With the broadband orientation of the next generation mobile wireless networks and the development of new high-speed data services such as the Internet of Things,the demand for frequency spectrum resources is experiencing explosive growth.As a non-renewable resource,the resource of the licensed spectrum is increasingly saturated.Its scarcity has gradually become a major bottleneck restricting the development of new wireless technologies.Next-generation wireless communication system takes dynamic spectrum sharing as a research hotspot,triggering a wave of researches on efficient spectrum sharing technologies.In order to further improve the spectrum efficiency and meet the requirements of high-speed data services,cognitive radio based heterogeneous network are generally considered as one of the key trends and research interests for next generation mobile communication,Industrial 4.0,Internet of Things and smart grid.However,the hierarchical architecture and cognitive spectrum sharing will introduce co-tier and cross-tier interference to the existing macrocell networks,which poses new requirements and challenges on interference management.Instead of using spectrum sensing and handoff to avoid interference,the interference management is studied based on MIMO and spatial procoding.More specifically,this thesis studies the downlink centralized coding,uplink distributed coding and user grouping algorithm,then proposes novel solutions.Firstly,for the interference in cognitive tiered network downlink,we propose a centralized leakage based precoding algorithm with dimensionality reduction.According to the characteristics of cognitive radio,an interference channel angle estimation method which requires no feedback from the macrocell is designed using the idea of Multi-Channel Ratio(MCR).With the obtained interference channel,we eliminate the cross-tier interference to the macrocell by dimensionality reduction at a rather low complexity.For the co-tier interference within the femtocell,by fully exploiting the generalized Rayleigh quotient,we employ the leakage based criterion to decouple the optimization problem.The proposed algorithm can be further enhanced with matrix simultaneous diagonalization to balance the channel gain of multiple streams.In order to improve the access capabilities,a hybrid beamforming based on non-orthogonal multiple access(NOMA)is designed to allow more users to access the network.Compared with the traditional precoding algorithms,the proposed scheme are self-organized and does not require the primary user's cooperation.Simulation results demonstrate that the proposed precoding schemes can provide significantly higher network capacity,while causing no quality of service deterioration on the macrocell.Secondly,the distributed interference suppression coding methods in cognitive heterogeneous network uplink is studied in depth.In this thesis,we propose a distributed interference alignment serial detection algorithm with the help of channel duality and null space iteration theorem.The proposed algorithm utilizes LQ decomposition to design a linear precoder to eliminate the cognitive interference to macrocell,then eliminates the co-tier and cross-tier interference by iterating on the forward and reciprocal channels.A low complexity serial detection method is proposed based on null space iteration for multi-stream interference.In order to further improve the network capacity under the low or medium signal-to-noise ratio,an improved cognitive interference alignment is also proposed by relaxing the non-interference constraints.In addition,the convergence proof and feasibility analysis are provided.Compared with the existing algorithms,the proposed schemes exploit null space iteration to reduce the complexity and can be implemented in a distributed way,which simplifies the network architecture.The network capacity and spectrum efficiency are improved,without interfering the primary users in macrocell.Finally,for the interference caused by the increasing number of users and the waste of spectrum resources by existing user grouping methods with fixed threshold,this thesis studies the user grouping problem based on graph theory and clustering to manage interference.For hybrid beamforming NOMA network,we consider both intra-beam and inter-beam correlations,and propose a NOMA user grouping method based on spectral clustering.By exploiting the multi-user diversity and favorable propagation,the idea of overlapping user grouping is first proposed.We proposes to use generalized Fubini-Study function as the improved similarity measurement,then propose two overlapping user grouping approaches based on heuristic search and spectral clustering.The theoretical analysis of the algorithm performance is also presented.The proposed approaches can allocate the users with favorable propagation into multiple subgroups,increase the number of users served at each group by overlapping,while the intra-group interference is kept at a low level.Compared with the existing non-overlapping user grouping methods,the proposed approaches can obtain higher network capacity,and the spectral clustering based user grouping algorithm strikes a good balance between performance and computational complexity.