[Paper Review] Mobility-Aware Analysis of 5G and B5G Cellular Networks: A Tutorial

Hina Tabassum, Mohammad Salehi+1 more|arXiv (Cornell University)|May 7, 2018

Advanced MIMO Systems Optimization67 references18 citations

TL;DR

This tutorial presents a comprehensive mobility-aware performance analysis framework for 5G and beyond 5G (B5G) cellular networks, leveraging stochastic geometry and mobility models to evaluate key metrics like handoff rate, coverage probability, and sojourn time. It introduces two analytical approaches—trajectory-based and association/handoff-based—demonstrating their application in random and non-random network topologies, with validation via simulations and numerical results for hexagonal and square lattices, as well as Poisson point process-based deployments.

ABSTRACT

Providing network connectivity to mobile users is a key requirement for cellular wireless networks. User mobility impacts network performance as well as user perceived service quality. For efficient network dimensioning and optimization, it is therefore required to characterize the mobility-aware network performance metrics such as the handoff rate, handoff probability, sojourn time, direction switch rate, and users' throughput or coverage. This characterization is particularly challenging for heterogeneous, dense/ultra-dense, and random cellular networks such as the emerging 5G and beyond 5G (B5G) networks. In this article, we provide a tutorial on mobility-aware performance analysis of both the spatially random and non-random, single-tier and multi-tier cellular networks. We first provide a summary of the different mobility models which include purely random models, spatially correlated, and temporally correlated models. The differences among various mobility models, their statistical properties, and their pros and cons are presented. We then describe two main analytical approaches for mobility-aware performance analysis of both random and non-random cellular networks. For the first approach, we describe a general methodology and present several case studies for different cellular network tessellations such as square lattice, hexagon lattice, single-tier and multi-tier models in which base-stations (BSs) follow a homogeneous Poisson Point Process (PPP). For the second approach, we also outline the general methodology. In addition, we discuss some limitations/imperfections of the existing techniques and provide corrections to these imperfections. Finally, we point out specific 5G application scenarios where the impact of mobility would be significant and outline the challenges associated with mobility-aware analysis of those scenarios.

Motivation & Objective

• Address the critical challenge of modeling user mobility in dense, heterogeneous, and ultra-dense 5G/B5G networks to improve network performance and QoS.
• Characterize mobility-aware metrics such as handoff rate, handoff probability, sojourn time, and direction switch rate in spatially random and structured cellular deployments.
• Provide a systematic comparison of mobility models (random, spatially/temporally correlated) and analytical techniques (trajectory-based and association/handoff-based) for performance evaluation.
• Identify limitations in existing analytical methods and propose corrections to improve accuracy in modeling mobility impacts on network performance.
• Explore the implications of mobility in emerging 5G use cases such as UAVs, V2X, and network slicing, and propose adaptive solutions using SDN and machine learning.

Proposed method

• Classify mobility models into purely random (e.g., random walk, random way point), spatially correlated (e.g., pursue, column), and temporally correlated (e.g., Gauss-Markov, Lévy flight), analyzing their statistical properties and suitability for network modeling.
• Develop a trajectory-based analytical approach using stochastic geometry to model user mobility in single- and multi-tier cellular networks with base stations (BSs) distributed as a homogeneous Poisson Point Process (PPP).
• Implement an association/handoff-based analytical method to evaluate handoff dynamics and coverage probability, focusing on signal strength-based handoff decisions and user mobility patterns.
• Calibrate analytical models using numerical results and simulations across different network tessellations (hexagonal, square lattice) and deployment scenarios (single-tier, multi-tier).
• Integrate SDN-based mobility management techniques, including automatic neighbor relation (ANR) and decoupled control/data plane architectures, to reduce signaling overhead and improve handoff efficiency.
• Apply data-driven deep learning models (e.g., DNNs, autoencoders, RNNs) to predict mobility patterns and network performance metrics where analytical modeling is intractable, using training and test datasets for model optimization.

Experimental results

Research questions

• RQ1 How do different mobility models (random, spatially, temporally correlated) affect key performance metrics such as handoff rate and coverage probability in 5G/B5G networks?
• RQ2 What are the relative strengths and limitations of trajectory-based versus association/handoff-based analytical approaches in modeling mobility-aware performance in random and non-random cellular topologies?
• RQ3 How does user mobility impact network slicing, especially in terms of handoff frequency and signaling overhead across different network slices (e.g., eMBB vs. mMTC)?
• RQ4 To what extent can SDN-enabled control plane decoupling reduce handoff signaling and improve QoS for high-mobility users in ultra-dense networks?
• RQ5 Can deep learning models accurately predict mobility patterns and network performance metrics in complex, dense 5G/B5G environments where analytical modeling is infeasible?

Key findings

• The handoff rate and coverage probability in 5G/B5G networks are significantly influenced by mobility patterns, with high-mobility users experiencing substantially higher handoff frequencies than static or low-mobility users.
• The PPP-based model provides a pessimistic lower bound for SINR coverage probability, while the hexagonal grid model offers an optimistic upper bound, validating the use of stochastic geometry for tractable performance evaluation.
• Decoupled control and data plane architectures reduce handoff frequency by up to 50% for high-speed users, significantly lowering signaling overhead and improving QoS.
• Mobility-aware caching strategies improve cache hit probability by accounting for random contact durations between mobile users and BSs, particularly in video streaming scenarios.
• Data-driven deep learning models, such as RNNs and DNNs, can effectively predict mobility patterns and network performance metrics when analytical models are intractable, though they require large training datasets and careful architecture tuning.
• Network slicing in 5G/B5G networks demands adaptive handoff thresholds and multi-RAT coordination to manage diverse mobility requirements across slices, especially for URLLC and mMTC services.