Research Of Cluster-Based Resource Allocation Scheme In High Dense Networks

With the increasing demand of high-speed data services, the next generation wireless network is facing severe challenges. However, utilization of extra spectrum resources is a very costly solution. Besides, point-to-point link throughput is very close to the Shannon limits. Little performance gain can be achieved by advanced physical layer technologies. Hence, the most promising way to improve system throughput as well as spectrum efficiency is the evolution of network architecture. Considered as the new architecture to deal with the exploding data traffic in hot spots, high dense networks are widely studied and supported by many scholars.Home enhanced Node B (HeNB) can significantly improve the indoor coverage and service quality. However, the characteristics of dense deployment, overlapping coverage and plug and play cause severe inter-cell interference. It is necessary to study the radio resource management schemes which are suitable for high dense HeNB networks. Although the power consumption of each HeNB is very low, the total power consumption is sharply increased with high dense deployment in a given area. Apparently, reducing the power consumption of high dense networks is the major challenge in the coming decades.This thesis studies the resource allocation and energy efficiency issues in high dense HeNB networks. Specifically, this thesis proposes a two-stage resource allocation scheme and a cluster-based sleep scheme. System throughput and spectrum efficiency has been greatly improved, while the total power consumption of high dense HeNB networks decreases significantly. The main contents of the thesis include:Firstly, after analyzing the interference scenario and features of high dense HeNB networks, we conclude that conventional resource allocation schemes, such as fractional frequency reuse and almost blank subframes, have some deficiencies. Thus, we introduce a cluster-based resource allocation scheme to reduce the processing complexity of high dense networks. And on this basis we propose a modified K-means clustering algorithm.Secondly, this thesis proposes a two-stage resource allocation scheme. This scheme is designed by clustering to maximize the system throughput with less processing complexity. The detailed resource allocation procedure includes two stages. In the first stage, a resource block with the best channel quality is assigned to each HeNB by greedy algorithm. In the second stage, all unsatisfied HeNBs select the supplementary resource blocks for further compensation. Simulation results show that the proposed scheme outperforms the heuristic cluster-based femto-femto interference minimized sub-channels allocation algorithm (HCFM) and achieves a satisfactory performance in terms of average signal to interference plus noise ratio (SINR), the system throughput, and the spectrum efficiency.Thirdly, this thesis proposes a cluster-based sleep scheme. This scheme is designed to minimize the system power consumption on the basis of clustering results, so that as much of HeNBs could keep sleeping. The detailed sleep scheme includes two parts as follows:each cluster head makes sleep decisions independently, and clusters conduct inter-cluster coordination through information interaction. The simulation results show that our proposed cluster-based sleep scheme obtains the least number of average ACTIVE HeNBs under the condition of different amount of users and the user access grant limit, that is, the total system power consumption is lower than the two reference schemes. The result verifies the effectiveness of our proposed scheme in terms of reducing power consumption. Besides, large gains are still achieved in a scenario with dense users. This illustrates that the proposed scheme is appropriate for high dense networks.Finally, the summary is given at the end of this thesis and future research directions in related fields are also pointed out.