Resource Management Schemes Based On Interference Alignment With Clustering In Dense Small Cell Networks

The world-wide explosive growth in the number of mobile users as well as the massive amount of mobile data traffic has posed new challenges to the current wireless communication networks.As the data traffic which calls for higher rates,e.g.,online games and high definition videos,mainly happens in the indoor areas,e.g.,home,where the coverage of macrocell base station is always poor,and the quality of received signal will be worse,and the users also have bad experiences of data traffic with higher rate requirements.Due to the low-power and low-cost advantages of small cell base stations(SBSs),they will be densely deployed in indoor areas to improve the spectrum efficiency and address the aforementioned challenge.Therefore,such a promising and economical solution will be greatly welcomed by the mobile operators,and it will also be an inevitable tendency in future wireless communication networks.Nevertheless,the severe interference among SBSs and small cell user equipments(SUEs)incurred by the dense deployment of small cell networks will significantly degrades the network performance.As a result,how to efficiently management the interference in dense SCNs is one of the most critical issues that needs to be addressed.Interference alignment(IA)is an efficient and new interference management technique.However,exploiting IA in dense SCNs is limited by its feasibility condition,implementation complexity and feedback overhead incurred by the estimation of perfect channel state information(CSI).IA with clustering can address the aforementioned issues of IA.However,only performing IA with clustering is still unable to mitigate the relatively strong inter-cluster interference in dense SCNs,which will degrade the sum rate of all SUEs in dense SCNs,can't sufficiently guarantee SUEs' differentiated QoS requirements,or make all SUEs have satisfactory connectivity requirements.As a result,resource allocation should be further performed based on IA with clustering to meet the aforementioned demands.This dissertation mainly focus on allocating the limited subchannel resources based on IA with clustering in dense SCNs with massive amount of SUEs,which aims at improving the sum rate of all SUEs,maximizing the number of QoS guaranteed SUEs,and maximizing the sum connectivity of all SUEs in dense SCNs.As all the corresponding optimization problems are NP-hard,we propose efficient algorithms with low complexity and notably reduced feedback overhead to solve them.Besides,we briefly introduce how to apply our research results to the corresponding demands in military communications.Specifically speaking,the main contents of this dissertation are summarized as follows.1.We exploit IA with clustering for efficient subchannel allocation in full connected ultra-dense mall cell networks(SCNs),which completely eliminates the co-tier interference between SBSs and SUEs and maximizes the sum rate of all SUEs.Our problem is formulated as a combinatorial optimization problem which is NP-hard.Therefore,we propose a two-phases efficient solution with low complexity and notably reduced feedback overhead.The first phase groups all SUEs into disjoint clusters only according to the path losses between SBSs and SUEs,which also notably reduces the feedback overhead as the estimation of global perfect CSI is avoided;and the second phase allocates subchannels to the formed clusters performing IA.In this way,the intra-cluster and inter-cluster interferences are eliminated by clustered IA and subchannel allocation,respectively.Low-complexity algorithm is proposed to solve the corresponding subproblem in each phase.Furthermore,we analyze the computational complexities of proposed solution as well as the optimal solution.Simulation results validate the effectiveness of proposed scheme.2.We propose an efficient subchannel allocation scheme based on IA with similarity clustering in partial connected dense SCNs underlaying a macrocell,which aims at maximizing the number of QoS guaranteed SUEs performing IA through allocating the limited subchannel resources.The corresponding problem is formulated as a combinatorial optimization problem which is NP-hard.So a low-complexity efficient solution is proposed which includes three phases.In the first phase,similarity clustering for SUEs is performed through graph partitioning,which ensures that both QoS requirements for SUEs and path losses between SBSs and SUEs are similar in each cluster.In the second phase,cluster sizes are further adjusted to meet the feasibility condition of IA in each cluster.The first two phases are performed only according to the QoS requirement for each SUE and the path losses between SBSs and SUEs,which also avoid estimating the global perfect CSI,so that the feedback overhead is notably reduced.In the third phase,to maximize the number of QoS guaranteed SUEs,efficient subchannel allocation for the formed clusters is performed by proposed algorithms with notably reduced computational complexities.Besides,we respectively analyze the computational complexities of proposed solution and the optimal solution.Numerical results show that the proposed three-phases solution not only outperforms other related schemes but also achieves a performance close to the optimal solution.3.In partial connected dense SCNs,we present a subchannel allocation scheme based on IA with partial clustering to maximize the sum connectivity of all SUEs,guaranteeing that each SUE achieves satisfactory connectivity requirement.The corresponding problem is NP-hard.Consequently,we propose a graph-based low-complexity solution consisting of four phases: finding all the first category of stand-alone vertices,using min-cut criterion to recursively partition the obtained subgraph through Stoer-Wagner algorithm,exploiting local exhaustive search to remove vertices from the clusters where the number of vertices do not meet the feasibility condition of IA,and utilizing coloring theory to allocate subchannels to the formed clusters as well as the second and third categories of stand-alone vertices.Each of the aforementioned phases can be performed by a proposed algorithm with lower computational complexity.Besides,the proposed solution notably reduces the feedback overhead incurred by the estimation of CSI.This is because performing the first three phases only require to estimate the path losses between SBSs and SUEs instead of the global perfect CSI.Finally,the computational complexities of both the proposed solution and the optimal solution are analyzed.