Multi-layer Unmanned Aerial Vehicle Networks: Modeling and Performance Analysis

Dong-Sun Kim, Jemin Lee, Tony Q. S. Quek

TL;DR

This paper introduces a novel framework for analyzing multi-layer Unmanned Aerial Vehicle (UAV) networks (MANs), considering diverse UAV capabilities and network configurations. The authors develop analytical models for key performance metrics like successful transmission probability (STP) and area spectral efficiency (ASE), providing insights into optimal network design for enhanced performance.

What Problem Does This Paper Address?

As UAVs become increasingly integrated into wireless communication systems for applications like disaster relief, surveillance, and data acquisition, the need for efficient and reliable aerial networks (ANs) is paramount. Existing research often focuses on single-layer ANs or terrestrial networks with limited UAV considerations. This paper addresses the challenge of modeling and analyzing multi-layer ANs (MANs), where different layers of UAVs possess varying altitudes, densities, and transmission powers. The authors aim to bridge the gap in understanding how these diverse characteristics impact network performance and how to optimize MAN design for maximum efficiency.

Methods

The core of this paper lies in its comprehensive modeling of MANs using stochastic geometry. Key methodological contributions include:

  • Multi-layer Network Modeling: The MAN is modeled as a system of $K$ layers, each with UAVs characterized by distinct densities, altitudes, and transmission powers. Layer 0 represents the terrestrial network.
  • Association Rules: Two types of node association rules are considered: transmitter-oriented (where a transmitter selects the best receiver) and receiver-oriented (where a receiver selects the best transmitter). Association criteria include nearest distance and strongest power.
  • Channel Modeling: Both air-to-air (A2A) and air-to-ground (A2G) channels are modeled, incorporating line-of-sight (LoS) and non-line-of-sight (NLoS) links. The LoS probability is a function of link distance and altitude difference.
  • Interference Analysis: The Laplace transform of interference is derived for both same-layer and cross-layer transmissions, considering LoS and NLoS channels.
  • Performance Metrics: The paper analyzes the Successful Transmission Probability (STP) and Area Spectral Efficiency (ASE) of the MAN.
  • Optimal Density Analysis: Upper bounds for optimal transmitter densities that maximize STP and ASE are derived, offering guidance for network design.

Datasets

Not discussed in the paper. The analysis is based on theoretical modeling and simulation.

Key Insights and Ideas

  • Altitude Matters Significantly: The optimal altitude of UAV layers plays a crucial role in maximizing STP and ASE. There's often a trade-off: higher altitudes increase LoS probability for the main link but can also increase interference.
  • Layered Networks Offer Flexibility: Multi-layer MANs can outperform single-layer ANs, especially when total UAV density is moderate. The optimal strategy can involve concentrating UAVs in lower or higher layers depending on the total density.
  • Association Rules Impact Performance: The choice between receiver-oriented and transmitter-oriented association significantly influences optimal density and performance. For transmitter-oriented association, optimal transmitter density for STP can be zero due to increased interference.
  • Density vs. Interference Trade-off: Increasing transmitter density generally improves ASE by increasing the number of communication links, but it also increases interference, leading to a performance trade-off that necessitates finding an optimal density.
  • Upper Bounds for Design: The derived upper bounds for optimal transmitter densities are valuable for guiding search algorithms and understanding design constraints, showing a decreasing trend with altitude.

Key Results

  • The paper provides analytical expressions for the main link distance distribution, LoS probability, and interference Laplace transform in multi-layer ANs.
  • Numerical results demonstrate the existence of optimal altitudes for UAV layers to maximize STP and ASE.
  • It's shown that the optimal UAV density for maximizing ASE decreases with altitude, and for maximizing STP, it also decreases with altitude in the receiver-oriented case, becoming zero in the transmitter-oriented case.
  • For a given total UAV density, the optimal MAN configuration can be a single layer (lower for high density, higher for low density) or a multi-layer structure, depending on the density level.
  • The upper bound of optimal transmitter density is shown to be a non-increasing function of the altitude difference between layers.

Limitations and Future Work

  • Interference-Limited Assumption: The analysis often assumes an interference-limited environment (σ² = 0) to simplify calculations, which might not always hold in real-world scenarios.
  • Simplified Mobility Model: The paper focuses on static UAV deployments for performance analysis. Future work could incorporate dynamic UAV mobility.
  • Closed-Form Expressions: While significant progress is made, some analytical expressions, particularly for interference, do not have a simple closed-form solution, requiring approximations or numerical evaluation.
  • Specific Association Rules: The study considers specific association rules (nearest distance, strongest power). Exploring other, more advanced association strategies could be a future direction.
  • Resource Allocation: The paper focuses on density and altitude optimization. Future research could delve into more complex resource allocation problems within MANs, such as power control and spectrum sharing.

Summary updated

Abstract

Since various types of unmanned aerial vehicles (UAVs) with different hardware capabilities are introduced, we establish a foundation for the multi-layer aerial network (MAN). First, the MAN is modeled as K layer ANs, and each layer has UAVs with different densities, floating altitudes, and transmission power. To make the framework applicable for various scenarios in MAN, we consider the transmitter- and the receiver-oriented node association rules as well as the air-to-ground and air-to-air channel models, which form line of sight links with a location-dependent probability. We then newly analyze the association probability, the main link distance distribution, successful transmission probability (STP), and area spectral efficiency (ASE) of MAN. The upper bounds of the optimal densities that maximize STP and ASE are also provided. Finally, in the numerical results, we show the optimal UAV densities of an AN that maximize the ASE and the STP decrease with the altitude of the network. We also show that when the total UAV density is fixed for two layer AN, the use of single layer in higher(lower) altitude only for all UAVs can achieve better performance for low(high) total density case, otherwise, distributing UAVs in two layers, i.e., MAN, achieves better performance.
BibTeX
@article{kim2019multi,
  author = {Dong-Sun Kim and Jemin Lee and Tony Q. S. Quek},
  title = {{Multi-layer Unmanned Aerial Vehicle Networks: Modeling and Performance Analysis}},
  journal = {IEEE Trans. Wireless Communications},
  year = {2019},
}
Figure