Source-linked AI summary
A Drone-based Networked System and Methods for Combating Coronavirus Disease (COVID-19) Pandemic
Adarsh Kumar, Kriti Sharma, Harvinder Singh, Sagar Gupta Naugriya, Sukhpal Singh Gill, Rajkumar Buyya
TL;DR
COVID-19 created a need for drone-based healthcare operations in settings where connectivity limitations or infection risks complicate conventional response. The paper proposes and evaluates a multilayered UAV architecture using wearable, movement, and thermal-image data, reporting broad operational coverage and effective indoor patient identification while noting simulation and hardware-scope limitations.
Problem
The paper addresses the need for drone-based healthcare systems that collect and analyze pandemic data across monitoring, sanitization, medication, and thermal-imaging operations.
Method
The proposed architecture gathers wearable-sensor, movement-sensor, and thermal-image data, then uses multilayered edge, fog, and cloud processing for analysis and decisions.
Results
1200 kilometers can be covered in 2293 to 18900 minutes with 3 to 30 drones, and indoor thermal image-based patient identification is found very effective.
Takeaways & Limitations
The system supports coordinated multi-drone pandemic monitoring, data processing, and operational responses across indoor and outdoor healthcare scenarios.
Takeaways & Limitations
Simulation cases assume ideal drone movement and compatible indoor drones, whereas real deployment requires environmental effects and compact-drone design to be considered.
Abstract
from arXiv · showhide
Coronavirus disease (COVID-19) is an infectious disease caused by a newly discovered coronavirus. It is similar to influenza viruses and raises concerns through alarming levels of spread and severity resulting in an ongoing pandemic worldwide. Within eight months (by August 2020), it infected 24.0 million persons worldwide and over 824 thousand have died. Drones or Unmanned Aerial Vehicles (UAVs) are very helpful in handling the COVID-19 pandemic. This work investigates the drone-based systems, COVID-19 pandemic situations, and proposes an architecture for handling pandemic situations in different scenarios using real-time and simulation-based scenarios. The proposed architecture uses wearable sensors to record the observations in Body Area Networks (BANs) in a push-pull data fetching mechanism. The proposed architecture is found to be useful in remote and highly congested pandemic areas where either the wireless or Internet connectivity is a major issue or chances of COVID-19 spreading are high. It collects and stores the substantial amount of data in a stipulated period and helps to take appropriate action as and when required. In real-time drone-based healthcare system implementation for COVID-19 operations, it is observed that a large area can be covered for sanitization, thermal image collection, and patient identification within a short period (2 KMs within 10 minutes approx.) through aerial route. In the simulation, the same statistics are observed with an addition of collision-resistant strategies working successfully for indoor and outdoor healthcare operations. Further, open challenges are identified and promising research directions are highlighted.
1. Introduction
The paper proposes an AI-enabled drone healthcare system for COVID-19 operations, combining real-time implementation and simulation. Its architecture supports data collection, analysis, sharing, decision-making, and collision-resistant multi-drone operations.
- Pandemic motivation: 24.0 million infections and 824,162 deaths by August 2020 motivated drone-based support for monitoring and pandemic response.The passage also describes close-contact transmission and recommended protective measures.
- Existing systems: Existing drones support monitoring, vigilance, thermal scanning, medication, food supply, and alerts, but centralized data collection and analysis remain challenging.The paper identifies AI, thermal imaging, sanitization, and social-distancing measurement as enhancement opportunities.
- Architecture: The architecture exchanges drone information with edge, fog, and cloud servers for computing, data sharing, and analytics.Its contributions include multilayered architecture and control-room statistics generation.
- Proposed system: The proposed system uses AI for data collection, analysis, statistical visualization, information sharing, and decision-making across COVID-19 operations.Operations include sanitization, medication, monitoring, and thermal imaging.
- Evaluation: Real-time implementation and simulation evaluate sanitization, monitoring, vigilance, face recognition, thermal scanning, collision avoidance, and indoor and outdoor operations.The implementation was conducted in Delhi/NCR, India, while simulation considered multiple drone scenarios.
2. Related Work
Related work shows broad use of drones for pandemic surveillance, sanitization, medical delivery, and thermal monitoring. The paper identifies gaps in drone-network design, collision-free routing, AI integration, and pandemic-specific evaluation.
- Pandemic applications: Drones have been used for surveillance, social-distancing and mask alerts, sanitization, monitoring, thermal scanning, governance, vigilance, and supply delivery.Examples include deployments in China, the United States, and India.
- Healthcare applications: Drone systems have supported medicine delivery, sanitization, monitoring, analysis, patient identification, and thermal body-temperature measurement, including in resource-scarce rural areas.One reported real-time approach was tested with at least 1000 patients.
- Comparative analysis: Comparative analyses summarize existing drone-based approaches for pandemic, disaster, healthcare, and other application domains.Table 1 compares the proposed approach with an existing pandemic or disaster system, while Table 2 compares systems across applications.
- Open challenges: The literature identifies limited practice in healthcare drone-network construction, resource-constrained QoS improvement, collision-free routing, multilayered flight, and pandemic-operation testing.It also lists solar charging, AI-based operations, and drone-level federated learning as challenges.
3. A Drone-based Architecture for Smart Healthcare System
The proposed smart-healthcare architecture integrates AI, IoT-related networks, and cloud, fog, and edge computing for drone-based COVID-19 monitoring, control, and analytics. It includes thermal imaging, wearable-sensor monitoring, local decision-making, and secure data exchange.
- The architecture integrates machine learning, deep learning, IoT, IIoT, IoMT, IoD, and cloud, fog, and edge computing.These components support data collection, instruction-based services, analysis, storage, and processing.
- The figures present the proposed architecture, wearable-sensor person monitoring, and information-capture, processing, and security flows.The captions identify these three views of the smart-healthcare system.
- Thermal Imaging System: Thermal imaging uses drone cameras to capture images, measure social distancing and density, and support temperature-related monitoring.The camera scanner detects objects and produces thermal-image displays, with image clarification when needed.
- Edge Network and Computing System: Edge computing performs data modeling, aggregation, preprocessing, and initial decisions while forwarding necessary data to fog or cloud systems.Local processing helps maintain real-time operation and reduces transferred data, although transfer costs increase with network scalability.
- Security: The architecture uses secure tunnels and end-to-end encryption so authenticated devices can receive drone-collected data.The described implementation uses the Lightbridge application for encrypted data sharing.
4. Algorithms and Operational Strategies
The operational strategy divides geographic areas into drone-sized zones and combines federated learning with edge aggregation for pandemic monitoring and coordinated operations. Each drone learns from its zone while aggregated observations support collective decisions.
- Zone-Based Operations: The system divides a geographic area into square zones of side length τ, with each zone planned for coverage by one drone.Drones monitor individual zones, while edge computing collects average observations across zones for collective decisions.
- Federated Learning: Federated learning lets drones self-learn from zone-specific COVID-19 experiences before securely sharing training data with edge, fog, or cloud servers.The approach supports collaboration, data sharing, and processing across multiple drones without requiring every raw observation to be centralized.
- Monitoring Operations: The algorithm combines patient monitoring, thermal scanning, image identification, and body-temperature measurement in zone operations.These observations support monitoring and operational responses such as sanitization and vigilance.
- Algorithm Parameters: The algorithm defines τ as a zone’s length and width and δ as the interval between successive drone zone scans.The kth patient in the ith zone is represented in the algorithm’s notation.
1. For each 𝑁𝑙 : 2. For each 𝑍𝑖
The operational loop repeatedly collects COVID-19 and wearable-sensor observations for each zone and scanning interval, then initiates sanitization and medication when a patient’s temperature increases over time.
- For each scanning interval δ, the algorithm collects COVID-19 scans, thermal images, temperature, and other wearable-sensor measurements.The loop then advances the scanning interval and measures the zone quantity specified by the algorithm.
- When a patient’s body temperature increases with time, the algorithm starts sanitization and medication before continuing the monitoring loop.The intervention is conditional on the stated temperature trend.
12. End For 13. Share 𝑄𝑍𝑖
The paper presents single- and multi-layer drone movement strategies for transferring drones between zones while supporting COVID-19 monitoring, distancing, and operational tasks. It combines collision avoidance, zone-transfer mechanisms, social-distancing algorithms, and distance measurement methods.
- Single Layer Zone-Transfer Algorithms: Fixed-area transfer exchanges drones through pre-planned transfer zones that must be empty before movement, preventing collisions during zone swaps.The strategy supports left-to-right and right-to-left transfers and waits when the relevant transfer zone is occupied.
- Multi-Layer Zone-Transfer Algorithms: Collision-feasibility zones and coordinated movement rules support collision-resistant drone operation across single-layer, two-layer, and hybrid configurations.The hybrid strategy divides movement space into multiple layers and can apply different zone-transfer methods to different operational areas.
- Single Layer Zone-Transfer Algorithms: Zigzag and parallel strategies move drones across an n × n zone matrix using indexed paths, movement intervals, and zone-occupancy checks.The zigzag strategy maps matrix indices to drone locations, while the parallel strategy advances drone groups across rows at successive intervals.
- Multi-Layer Zone-Transfer Algorithms: Multi-layer strategies assign different drone activities to separate air layers and use layer changes or single-layer transfer routes for collision avoidance.The two-layer design separates zone transfer from COVID-19 operations, while hybrid movement applies different transfer strategies across multiple infrastructure-based layers.
- Social-Distancing Algorithms: Social-distancing algorithms calculate distances between consecutive or randomly distributed people, issue alerts when spacing is inadequate, and notify the control room.Distance measurement can use onboard sensors, geographic coordinates, image processing, or ground sample distance, depending on availability and setting.
- Distance Measurement: The distance-measurement procedures combine sensor observations, latitude-longitude calculations, camera-based measurements, and edge or image-processing inputs to identify people and spacing.The supplied algorithms include acoustic, laser, radio, infrared, mono/stereo sensors, and camera-image processing when available.
6. End For Control_Room_Notification( )
The paper describes a drone-network procedure that measures drone utilization, triggers COVID-19 operations based on thresholds, and evaluates distances among people or zones before notifying the control room.
- Drone and Control-Room Processing: Each drone’s utilization measurement determines whether it should return or begin a COVID-19 process based on upper and lower thresholds.The procedure calls a drone back when utilization reaches the upper threshold and starts COVID-19 processing below the lower threshold.
- Drone and Control-Room Processing: The algorithm iterates across drones and their associated entities, applying distance measurement and control-room notification procedures.Its stated goal is to measure distances between people in queues or randomly distributed populations and notify the control room when necessary.
- Person Detection: Sensor rotation and camera-image inspection provide alternative ways to count or detect people when direct distance sensors are available or unavailable.The procedure checks acoustic, laser, radio, infrared, and mono/stereo sensors before processing objects in camera images.
5. Performance Evaluation
The evaluation combines real-time COVID-19 drone operations with indoor and outdoor simulations covering sanitization, monitoring, thermal imaging, social distancing, medicine delivery, and network performance. Results quantify coverage, throughput, utilization, and collision-avoidance behavior across different drone deployments.
- Real-Time Evaluation: Real-time testing covered more than 15 densely populated areas, and a 2 km radius could be sanitized within 10 minutes.The system also performed thermal scanning and surface-area scanning in difficult-to-reach areas.
- Simulation Setup: The simulation modeled pedestrian movement and social distancing across areas varying from 250m2 to 1000m2.AnyLogic and JaamSim supported multimedia, agent-based, discrete-event, and system-dynamics modeling.
- Sanitization Performance: 1200 kilometers required 18900, 9390, 3680, and 2293 minutes with 3, 10, 20, and 30 drones, respectively.Drone recharging and sanitizer filling time were additional.
- Indoor Monitoring: 93% approximate simulation accuracy was reported for camera-based COVID-19 operations, with darker red thermal regions indicating more urgent sanitization.The thermal image represented density-based areas requiring sanitization.
- Social Distancing: 3389, 13398, 16298, and 19697 persons could be checked in 55 minutes with 3, 10, 20, and 30 drones, respectively.With 20 parallel medicine-supply units and a 120-second maximum supply time, 1612, 10073, 13129, and 16166 persons could be served alongside distancing checks.
- Network Performance: Average throughput ranged from 35 Mbps to 70 Mbps with approximately 22 drones and from 35 Mbps to 80 Mbps with approximately 17 drones using parallel movements.The throughput calculation used Bit Error Ratio and total packet transmission time.
- Collision Avoidance: The proposed fixed-area, zig-zag, and parallel single-layer algorithms generally executed faster than Nageli et al.'s video-graphing approach.The comparison concerned drone movement and collision avoidance under varying simulation times.
6. Conclusions and Future Directions
The paper presents a multilayer UAV healthcare system that combines sensor, movement, and thermal-image data for COVID-19 monitoring and decision-making. Implementations and simulations support broad operational coverage, while the authors identify idealized drone movement and hardware compatibility as limitations and propose expanded delivery and scanning capabilities.
- Conclusions: The proposed system combines wearable sensors, movement sensors, and thermal-image processing with edge, fog, and cloud computing for analysis and decisions.Edge computing manages collision-resistant strategies, while fog and cloud computing build commuter and patient profiles.
- Conclusions: 1200 kilometers could be covered in 2293 to 18900 minutes with 3 to 30 drones, and indoor thermal-image patient identification was found very effective.The approach was demonstrated through implementation and simulation for indoor and outdoor activities.
- Limitations: The Case-2 to Case-6 simulations assume ideal drone movement, so environmental conditions may change usage statistics in real scenarios.Indoor imaging and sanitization also assume compatible drones, whereas compact-drone design is required for comparable real operations.
- Future Directions: Future directions include large-scale medicine delivery with collision-resistant strategies and complete resident-record matching against scanned populations.The authors also discuss drone-enabled scanning and medicine services where medical infrastructure is limited.