When a natural disaster strikes, the immediate loss of communication infrastructure isolates survivors and paralyzes emergency services. Ground-based cell towers are highly vulnerable to physical destruction and power grid failures, as noted by Maurilio Matracia et al. in UAV-Aided Post-Disaster Cellular Networks: A Novel Stochastic Geometry Approach. Even when towers physically survive, they face extreme voice-channel congestion.
Interestingly, Son Doan et al. demonstrated in An analysis of Twitter messages in the 2011 Tohoku Earthquake that while traditional voice networks collapse under load, data-based communication pathways often remain operational, serving as vital "social sensors." However, when the underlying physical backhaul or power supply fails entirely, alternative network architectures must be rapidly deployed to bridge the gap.
The Aerial Dimension: Active vs. Passive Relays
To bypass destroyed ground infrastructure, researchers leverage the "vertical dimension" of 6G networks, deploying Unmanned Aerial Vehicles (UAVs), autonomous blimps, and tethered balloons to act as temporary aerial base stations. For instance, Yasushi Hada et al. in 1A1-F16 Outdoor experimental result of information acquisition using intelligent sensor nodes and an autonomous blimp demonstrated how autonomous blimps can collect data from ground sensors when traditional links fail.
However, flying a fleet of active Aerial Base Stations (ABS) introduces a major challenge: signal interference. When multiple drones emit active signals over a disaster zone, they can jam each other and degrade overall network performance. To solve this, Maurilio Matracia et al. in Comparing Aerial-RIS- and Aerial-Base-Station-Aided Post-Disaster Cellular Networks proposed using Aerial Reconfigurable Intelligent Surfaces (ARIS).
Instead of generating new signals, an ARIS acts as a programmable "smart mirror" in the sky, reflecting and focusing existing signals from surviving ground stations. The authors show in Unleashing the Potential of Aerial RISs in Post-Disaster Scenarios that this passive approach significantly boosts coverage without adding interference or consuming excessive onboard power. Counter-intuitively, the authors also discovered in Reliability in Post-Disaster Networks: A Novel Interference-Mitigation Strategy that intentionally shutting down certain surviving ground base stations can actually improve overall network reliability by eliminating destructive interference.
Decentralized Ground Meshes and Satellite Injection
When aerial support is delayed, consumer devices already present in the disaster zone can self-organize. Zongqing Lu 0002 et al. developed TeamPhone: Networking Smartphones for Disaster Recovery, a system that configures standard smartphones into ad-hoc, opportunistic networks. By utilizing energy-efficient wake-up protocols, these devices can exchange emergency data peer-to-peer without relying on any cellular towers. Similarly, Christian Webersik et al. in Towards an integrated approach to emergency management: interdisciplinary challenges for research and practice advocate for bi-connected ad hoc networks where mobile devices form autonomous local communication hubs.
The primary limitation of a pure ground mesh is its isolation from the wider world. This is where modern satellite integration changes the paradigm. Julian Zobel et al. explored this in On the Impact of Direct-to-Cell Satellite Communication in Decentralized Disaster Networks. They demonstrated that if even a tiny fraction (1–2%) of standard smartphones in a decentralized ground network can establish a "Direct-to-Cell" (D2C) link to low-Earth-orbit (LEO) satellites, they can act as gateways, injecting external connectivity into the entire local mesh and breaking the communication blackout.
Intelligent Resource Management under Constraints
Operating these ad-hoc aerial and ground networks requires solving complex, dynamic optimization problems under strict energy and bandwidth constraints. Drones have limited battery life, and routing data through a constantly shifting topology is highly inefficient.
To address this, researchers apply advanced machine learning techniques. Jie Li et al. introduced FedRDR: Federated Reinforcement Distillation-Based Routing Algorithm in UAV-Assisted Networks for Communication Infrastructure Failures, which uses federated reinforcement distillation to allow UAVs to learn optimal routing paths with minimal communication overhead.
For trajectory and data buffer management, Alireza Shamsoshoara et al. developed a behavioral cloning approach in UAV-Assisted Communication in Remote Disaster Areas Using Imitation Learning, enabling drones to mimic expert human scheduling to maximize data throughput while conserving flight energy. Furthermore, in scenarios where backhaul links are completely missing, the authors of Resource and Placement Optimization for Multiple UAVs using Backhaul Tethered Balloons demonstrated that pairing mobile UAVs with high-altitude tethered balloons provides a stable, optimized backhaul tier that dramatically improves data delivery over large disaster zones.