On the Topological Aspects of UAV-Assisted Post-Disaster Wireless Communication Networks

Maurilio Matracia, Mustafa A. Kishk, Mohamed-Slim AlouiniIEEE Communications Magazine 2021 · DOI · OpenReview

TL;DR

This paper investigates how the physical "topology"—the size and location—of a disaster zone dictates the effectiveness of aerial backup networks. By modeling various fleets of drones and high-altitude platforms using stochastic geometry, the authors demonstrate that the optimal network configuration depends heavily on whether a disaster strikes a dense urban center or an isolated rural area.

What Problem Does This Paper Address?

When natural or man-made disasters strike, terrestrial communication infrastructure (cell towers) is often the first thing to fail. In the critical "72-hour window" for search and rescue, the lack of connectivity can be fatal for victims and paralyzing for first responders.

While researchers have studied how to optimize drone trajectories or transceiver designs, they often overlook the topological context. Specifically:

  1. Extension: How does the physical radius of the "dead zone" change which aerial platform we should use?
  2. Location: Does it matter if the disaster happens in the middle of a dense city versus the outskirts?

The authors argue that we cannot design a one-size-fits-all aerial recovery plan without accounting for these spatial variables.

Methods

The researchers used stochastic geometry to model the locations of both surviving ground towers and newly deployed aerial nodes as Point Processes (PPs). They compared two main scenarios:

  • Homogeneous (HPPP): A simplified model where ground towers are spread uniformly across an infinite plane. This isolates the effect of the disaster's size (radius).
  • Inhomogeneous (IPPP): A more realistic model where tower density follows a 2D Gaussian distribution—dense in a "town center" and sparse further out. This allows them to test how the disaster's distance from the city center affects performance.

They evaluated three types of aerial platforms:

  1. Low-Altitude Platforms (LAPs): Standard drones and tethered drones (flying at ~100m–500m).
  2. Tethered Balloons: Providing more stability and power at medium altitudes.
  3. High-Altitude Platforms (HAPs): Airships or gliders in the stratosphere (~17km up) that cover massive areas.

Datasets

The paper relies on synthetic simulations rather than static real-world datasets. They performed $10^5$ iterations of their Point Process models to calculate ergodic capacity (the average maximum data rate) under various conditions. Key parameters included a 100 MHz bandwidth, a -5 dB SINR threshold, and specific transmit powers ranging from 1.585W (drones) to 20W (HAPs).

Key Insights and Ideas

  • The Interference Paradox: More drones are not always better. In small disaster zones, a large fleet of drones creates significant "self-interference," actually lowering the network's capacity.
  • The HAP "Dead Zone": For disasters with a radius smaller than 1 km, High-Altitude Platforms (HAPs) are actually detrimental. Because they fly so high (17 km), they act more like distant interferers than useful base stations when compared to surviving ground towers on the periphery.
  • The 3-Sigma Rule: In non-uniform cities, the steepest increase in network performance occurs when the disaster is located approximately three standard deviations away from the city center. At this distance, the "noise" from the dense urban core drops off enough for the aerial backup to shine.
  • Rural Simplicity: In isolated areas where ground towers are sparse, the optimal strategy is often to deploy just a single drone. This provides a short serving distance to the user while keeping aerial interference at zero.

Key Results

  • Extension Matters: As the disaster radius ($\rho_d$) grows beyond a few hundred meters, ground infrastructure becomes insufficient, and the need for aerial support becomes mandatory.
  • Altitude vs. Power: The optimal number of aerial nodes ($n_A$) decreases as the altitude ($h$) and transmit power ($p_t$) of the platforms increase.
  • HAP Threshold: HAPs only become the superior choice when the disaster radius exceeds several kilometers, allowing their wide coverage to outweigh their high path loss.
  • Capacity Gains: The study quantitatively shows that drones provide the best capacity for small, localized failures, while balloons and HAPs are necessary for regional catastrophes.

Limitations and Future Work

The authors acknowledge several hurdles that still need to be cleared for 6G emergency systems:

  • Deployment Time: While drones are fast, HAPs and tethered balloons require significant setup time that might not be available in an immediate crisis.
  • Backhaul Constraints: The paper assumes the aerial nodes can connect back to the core network, but finding reliable "backhaul" links in a destroyed zone remains a major technical challenge.
  • Dynamic Optimization: Future work should look at "on-the-fly" optimization where the number and placement of UAVs are adjusted in real-time as the disaster assessment data (like hazard maps) comes in.
  • Intra-region vs. Extra-region: There is a need to better distinguish between communication within the disaster zone (relays for responders) and communication out to the rest of the world (backhaul).

Summary updated

Abstract

In the context of sixth-generation (6G) networks, emergency management systems based on wireless communications have recently gained increasing interest. Hereby, fundamentals and open problems of post-disaster communications are discussed, especially focusing on their topological aspects. The motivation behind this choice is that whenever a natural or man-made disaster occurs, there is a high chance that the terrestrial communication infrastructure is compromised, and therefore alternative networks need to be deployed efficiently in order to enable the majority of the civilians and first responders to communicate. In this article, we first provide a brief review of existing aerial ad hoc networks for post-disaster communications. Next, we shed light on some new aspects of this problem, which are related to the topology of the network supporting the impacted area. Finally, with the aid of selected simulation results, we show how the cellular infrastructure requirements for a disaster-struck region significantly depend on its location and extension.
BibTeX
@article{matracia2021topological,
  author = {Maurilio Matracia and Mustafa A. Kishk and Mohamed-Slim Alouini},
  title = {{On the Topological Aspects of UAV-Assisted Post-Disaster Wireless Communication Networks}},
  journal = {IEEE Communications Magazine},
  year = {2021},
  doi = {10.48550/arxiv.2109.13671},
}
Figure