Future 5th generation (5G) wireless networks aim at supporting 1000 times higher spectral efficiency as compared with current network deployments. To meet this requirement, network densification is regarded as one of the dominant themes for wireless evolution. Network densification is realized by deploying small access nodes over the conventional macrocell networks, generating heterogeneous networks (HetNets). Nevertheless, deploying HetNets faces many challenges. One of the key challenges is the incursion of the interference between different tiers, namely, the cross-tier interference. A great amount of research has been launched on exploring effective methods to manage the interference in HetNets. Unfortunately, the optimal spectrum access scheme, which is one of the most effe...[ Read more ]

Future 5th generation (5G) wireless networks aim at supporting 1000 times higher spectral efficiency as compared with current network deployments. To meet this requirement, network densification is regarded as one of the dominant themes for wireless evolution. Network densification is realized by deploying small access nodes over the conventional macrocell networks, generating heterogeneous networks (HetNets). Nevertheless, deploying HetNets faces many challenges. One of the key challenges is the incursion of the interference between different tiers, namely, the cross-tier interference. A great amount of research has been launched on exploring effective methods to manage the interference in HetNets. Unfortunately, the optimal spectrum access scheme, which is one of the most effective interference management methods for HetNets, remains unknown. In this thesis, we shall investigate the optimal spectrum access scheme for macro-femto HetNets.

Firstly, we propose a general cognitive spectrum access framework, where the idea of cognitive cross-tier spectrum access is introduced into the conventional spectrum access scheme to improve the area spectral efficiency (ASE) of macro-femto HetNets. With this proposed framework, the optimal probability for femtocells to access the idle macro-tier spectrum can be determined. This cross-tier spectrum access probability provides the degree of freedom (DoF) to flexibly offload data traffic between the macrocell and femtocell tiers. The maximum ASE obtained with this optimal cross-tier spectrum access probability is much larger than that achieved in previous works. Moreover, the proposed general spectrum access framework provides high feasibility since it is applicable for different base station models.

Next, two kinds of optimal cognitive spectrum access schemes for macro-femto HetNets are determined. Specifically, each optimal spectrum access scheme consists of three parts: spectrum deployment, cognitive cross-tier spectrum access and co-tier spectrum access. Corresponding to these three components, there are three DoFs that we can manipulate to maximize the ASE of macro-femto HetNets. We determine the optimal F-ALOHA based overlay and underlay cognitive spectrum access schemes for macro-femto HetNets by optimizing the three DoFs, where the spectrum overlay and underlay are two spectrum sharing techniques. With the derived optimal three DoFs, the ASE of macro-femto HetNets can be maximized. Moreover, many meaningful results are obtained after determining the optimal spectrum access schemes. As a simple example, we find that cochannel deployment is the optimal spectrum deployment for sparse macro-femto HetNets while partial orthogonal deployment is optimal for dense networks.

To explore the dominate factor in the ASE of macro-femto HetNets, we further investigate the parameters involved in the expression of ASE based on the derived optimal three DoFs in the proposed overlay and underlay spectrum access schemes. Analytical results show that the factor which dominates the ASE of macro-femto HetNets is the number of active femto-links per unit area. In addition, the maximum ASE of macro-femto HetNets can be obtained with the optimal number of active femto-links per unit area, which is closely related to the femtocell intensity. Therefore, to maximize the ASE of macro-femto HetNets, the optimal number of active femto-links supported in each unit area should be achieved. Simulation results shall validate all the analytical results and demonstrate the effectiveness of our proposed optimal cognitive spectrum access schemes.