TUTORIAL ON "STOCHASTIC GEOMETRY-BASED MODELING AND ANALYSIS OF 5G CELLULAR NETWORKS: A TUTORIAL" @ IEEE PIMRC 2017

ABSTRACT

For more than three decades, stochastic geometry has been used to model large-scale ad hoc wireless networks, and develop tractable models to characterize and better understand the performance of these networks. Recently, stochastic geometry models have been shown to provide tractable and accurate performance bounds for cellular wireless networks including multi-tier and cognitive cellular networks, underlay device-to-device (D2D) communications, energy harvesting-based communication, coordinated multipoint transmission (CoMP) transmissions, full-duplex (FD) communications, etc. These technologies will enable the evolving fifth generation (5G) cellular networks. Stochastic geometry, the theory of point processes in particular, can capture the location-dependent interactions among the coexisting network entities. It provides a rich set of mathematical tools to model and analyze cellular networks with different types of cells (e.g., macro cell, micro cell, pico cell, or femto cell) with different characteristics (i.e., transmission power, cognition capabilities, etc.) in terms of several key performance indicators such as SINR coverage probability, link capacity, and network capacity. 

For the analysis and design of interference avoidance and management techniques in such multi-tier cellular networks (which are also referred to as small cell networks or HetNets), rigorous yet simple interference models are required. However, interference modeling has always been a challenging problem even in the traditional single-tier cellular networks. For interference characterization, assuming that the deployment of the base stations (BSs) in a cellular network follows a regular grid (e.g., the traditional hexagonal grid model) leads to either intractable results which require massive Monte Carlo simulation or inaccurate results due to unrealistic assumptions (e.g., Wyner model). Moreover, due to the variation of the capacity (both network and link capacities) demands across the service area (e.g., downtowns, residential areas, parks, sub-urban and rural areas), the BSs will not exactly follow a gridbased model. That is, for snapshots of a cellular network at different locations, the positions of the BSs with respect to (w.r.t.) each other will have random patterns. By capturing the spatial randomness of the BSs as well as network entities including network users, stochastic geometry analysis provides general and topology-independent results. When applied to networks modeled as spatial Poisson point processes (PPPs) with Rayleigh fading, simple closed-form expressions can be obtained which help us to better understand the network performance behavior in response to the variations in design parameters. Stochastic geometrybased analysis and optimization of future generation cellular networks is a very fertile area of research and has recently attracted significant interest from the research community.

The aim of this tutorial is to provide an extensive overview of the stochastic geometry modeling approach for next-generation cellular networks, and the state-of-the-art research on this topic. After motivating the requirement for spatial modeling for the evolving 5G cellular networks, it will introduce the basics of stochastic geometry modeling tools and the related mathematical preliminaries. Then, it will present a comprehensive survey on the literature related to stochastic geometry models for single-tier as well as multi-tier and cognitive cellular wireless networks, underlay D2D communication, and cognitive and energyharvesting D2D communication. It will also present a taxonomy of the stochastic geometry modeling approaches based on the target network model, the point process used, and the performance evaluation technique. Finally, it will discuss the open research challenges and future research directions. 

TUTORIAL OUTLINE

1. Overview of 5G Cellular Networks and Spatial Modeling Techniques (20 minutes)

• 5G visions and requirements and enabling technologies
• Key performance indicators (KPIs): SINR outage/coverage, average rate, transmission capacity
• SINR modeling techniques
• Stochastic geometry modeling

2. Point Process and Interference Modeling (30 minutes)

• Point processes (PPP, clustered processes, repulsive processes)
• Campbell theorem and probability generating functional
• Neyman Scott process: Matern cluster process and modified Thomas cluster process
• Laplace transform of the pdf of interference

3. Performance Evaluation Techniques (40 minutes)

• Technique #1: Rayleigh fading assumption
• Technique #2: Region bounds and dominant interferers
• Technique #3: Fitting
• Technique #4: Plancherel-Parseval theorem
• Technique #5: Inversion

4. Modeling Large-Scale Single and Multi-Tier Cellular Networks (60 minutes)

• Modeling downlink transmissions
• Modeling uplink transmissions
• Single-tier networks with frequency reuse
• Biasing and load balancing
• Optimal deployment of BSs
• Large-scale multiple-input multiple-output cellular systems

5. Modeling Cognitive Small Cells in Multi-Tier Cellular Networks (20 minutes)

• Spectrum sensing range and spectrum reuse efficiency
• Spectrum access schemes by cognitive small cells
• Network modeling
• Outage probability (channel outage and SINR outage) analysis for downlink transmissions in cognitive small cells 

6. Modeling Mode Selection and Power Control for Underlay D2D Communication (30 minutes) 

• Biasing-based mode selection and channel inversion power control for underlay D2D communication (Network modeling and stochastic geometry analysis)
• Cognitive and energy harvesting-based D2D communication (Network modeling and stochastic geometry analysis)

7. Open Issues and Future Research Directions (10 minutes)

8. References

TUTORIAL INSTRUCTORS

Ekram Hossain, University of Manitoba, Canada 

Click HERE for more details on all sixteen tutorials scheduled at IEEE PIMRC 2017.