Distance-based design and evaluation of heterogeneous cellular networks

• Main Author: Sinnen, Meshal
• Other Authors: Li Kwok Hung
• Format: Theses and Dissertations
• Language: English
• Published: 2019
• Subjects: Engineering::Electrical and electronic engineering::Wireless communication systems
• Online Access: https://hdl.handle.net/10356/84431; http://hdl.handle.net/10220/50106
• Institution: Nanyang Technological University

In recent times, heterogeneous cellular networks (HCNs) have been proposed as an attractive solution to cater to the increased demands in mobile user capacity and could have widespread application in cellular networks deployed under 5th generation (5G) mobile standards. Deploying small cells to coexist with macrocell base transmitting stations (BTSs) from traditional cellular networks enables increased area spectral efficiency (ASE) and throughput through the enhanced reuse of available bandwidth. However, deployment of HCNs is a complex operation and ongoing research aims to solve some of the problems relating to it. In particular, the aggregation of carrier interference can create coverage holes if cell planning is not optimized. While increased ASE may still be achieved by simply deploying small cells in large numbers, the coverage of the system can suffer due to overwhelming interference at mobile terminals, with reduced transmission efficiency per BTS. Therefore, it is important to consider strategies to mitigate interference and increase BTS efficiency when designing BTS location and resource allocation. Stochastic geometric models such as the Poisson Point Process (PPP) have been researched in detail to model the random nature of highly densified HCN BTS locations and uniformly distributed users. In most of this research, the emphasis is placed on the overall spatial performance metrics such as coverage and spectral efficiency. However, using the PPP to model users and BTSs in an HCN ignores practical factors such as the correlation between BTS and user locations, as well as the separation that usually exists between BTSs, where the latter is desirable for interference mitigation. In this thesis, we propose that these factors should be analyzed and accounted for when planning the locations of BTSs in two-tier HCNs and allocating resources for each BTS. Following a comprehensive literature review of stochastic geometric models, we introduce a novel framework that models user correlation to BTSs based on the Poisson Cluster Process (PCP). Furthermore, we incorporate a cross-tier separation rule between tier-1 macrocell BTSs and tier-2 picocell BTSs, by modelling the latter as a Poisson Hole Process (PHP). The first part of the analysis is a comprehensive study of distance distributions and the association of different user groups to BTSs from each tier, as well as the utilization of BTSs under constrained resources. These theoretical models are reused for the analysis of coverage, with some detail on how this affects the rate and ASE. Using numerical results, we analyze the effect of centralizing BTSs to user hotspots and demonstrate what the optimal separation between tier-1 and tier-2 BTSs must be in situations when the network has enough resources for users, and conversely, when resources are limited. The HCN model is extended by incorporating a co-tier separation rule that prohibits tier-2 BTSs from existing within a minimum distance from each other. The proposed model is analyzed for the case of uniform users, such that we measure the effectiveness of deploying tier-2 BTSs to fill coverage holes that exist naturally. We begin the analysis by focusing on distance PDFs and association probability, while showing that a minimum radius of association can be guaranteed based on the setting of co-tier and cross-tier separations. This is followed by a comprehensive analysis of coverage probability, and its relationship to ASE. Limits on both separation categories are explored for optimizing spatial coverage and ASE, while maintaining a viable tier-2 BTS density to form an HCN. The discussion highlights trade-offs that exist between coverage and ASE according to different separation settings and suggests the use of thresholds on each performance parameter to determine the optimal combination of co-tier and cross-tier separation. We showcase how this proposed planning technique works better as a rule of thumb in deploying HCNs when compared to benchmark results and include practical measures of throughput to support this claim. The work is concluded with a summary of contributions and recommendations for future work. One major conclusion is that careful placement of tier-2 BTSs is needed in relation to hotspots modelled by a PCP, and that the coverage of an HCN is maximized when the BTSs are near the centers of hotspots. Furthermore, optimally separating tier-2 BTSs from each other and tier-1 BTSs results in optimized coverage and ASE for the HCN. The work can be extended towards incorporating the mixed user distribution for the analysis of the HCN modelled according to both co-tier and cross-tier separations. Extension towards the use of alternate path loss and fading models such as the ones used to measure path loss indoors are also recommended.