Source-linked AI summary
An Amateur Drone Surveillance System Based on Cognitive Internet of Things
Guoru Ding, Qihui Wu, Linyuan Zhang, Yun Lin, Theodoros A. Tsiftsis, Yu-Dong Yao
TL;DR
Amateur drone deployment poses safety, security, and privacy challenges, while amateur drone surveillance remains largely unexplored. The paper surveys existing anti-drone techniques, proposes the Dragnet vision using cognitive IoT, and examines authorized-drone detection and classification through a case study.
Problem
Amateur drone deployment poses critical safety, security, and privacy challenges, making surveillance important despite limited prior exploration.
Method
The paper surveys anti-drone techniques, proposes Dragnet by tailoring cognitive IoT to amateur drone surveillance, and discusses its enabling techniques and ternary-hypothesis case study.
Results
The case study reports that detection performance improves with more samples and sensors, while higher correct detection increases false alarms.
Takeaways & Limitations
The paper develops a cognitive-IoT-based amateur drone surveillance system and illustrates detection and classification of authorized and unauthorized drones.
Abstract
from arXiv · showhide
Drones, also known as mini-unmanned aerial vehicles, have attracted increasing attention due to their boundless applications in communications, photography, agriculture, surveillance and numerous public services. However, the deployment of amateur drones poses various safety, security and privacy threats. To cope with these challenges, amateur drone surveillance becomes a very important but largely unexplored topic. In this article, we firstly present a brief survey to show the state-of-the-art studies on amateur drone surveillance. Then, we propose a vision, named Dragnet, by tailoring the recent emerging cognitive internet of things framework for amateur drone surveillance. Next, we discuss the key enabling techniques for Dragnet in details, accompanied with the technical challenges and open issues. Furthermore, we provide an exemplary case study on the detection and classification of authorized and unauthorized amateur drones, where, for example, an important event is being held and only authorized drones are allowed to fly over.
I. INTRODUCTION
Amateur drones offer broad civilian applications but create serious safety, security, and privacy risks, making surveillance important yet largely unexplored. The paper surveys existing approaches, proposes Dragnet, and illustrates authorized-drone detection and classification.
- Amateur drones support applications in communications, surveillance, agriculture, photography, and public services.
- Drone deployment threatens air traffic, public assets, military secrets, personal privacy, and physical safety.
- Amateur drone surveillance is important but remains largely unexplored.
- The paper surveys anti-drone studies, proposes the Dragnet cognitive-IoT vision, and discusses enabling techniques, challenges, and open issues.
- An exemplary case study addresses detection and classification of authorized and unauthorized amateur drones using multi-hypothesis testing.
II. AMATEUR DRONE SURVEILLANCE: STATE-OF-THE-ART
Existing amateur-drone surveillance research includes warning, spoofing, jamming, and mitigation techniques, but coverage remains narrow and systems are often scenario-specific and isolated. The section frames the need for broader, networked surveillance capabilities.
- Related work is generally grouped into warning, spoofing, jamming, and mitigation techniques.
- Warning systems use cameras, sensors, and radars to detect drone presence, while spoofing sends false signals to induce landing.
- Jamming disrupts drone control or navigation, whereas mitigation may destroy or capture drones using devices such as lasers or arrest nets.
- The paper presents only a brief summary of selected anti-drone systems rather than a comprehensive survey.
- Existing technologies often target specific drones or scenarios using relatively few methods, limiting application and generalization.
- Many anti-drone systems are designed in isolation and lack networked information-processing capability.
III. DRAGNET: COGNITIVE INTERNET OF THINGS-ENABLED AMATEUR DRONE SURVEILLANCE
The paper argues that connected surveillance devices alone are insufficient for amateur-drone surveillance. Dragnet extends connectivity with learning, reasoning, and understanding across physical and social contexts.
- Most related work connects surveillance devices to share observations and perform information fusion.
- Dragnet argues that surveillance devices should learn, think, and understand physical and social worlds rather than merely connect.
- Dragnet is a cognitive-Internet-of-Things-enabled amateur drone surveillance vision intended to provide high-level intelligence.
A. Cognitive Internet of Things
Cognitive IoT enhances conventional IoT with intelligence, enabling interconnected things to perceive context, learn from physical and social environments, and adapt decisions with limited human intervention.
- IoT has emphasized connectivity, communication, computing, and control, but comprehensive cognitive capability remains necessary to fulfill its potential.
- Cognitive IoT interconnects physical and virtual objects as agents that interact through a context-aware perception-action cycle.
- Its agents learn from physical environments and social networks, store semantic or knowledge representations, and adapt to changes through resource-efficient decisions.
- Cognitive IoT bridges physical and social worlds to form an intelligent physical-cyber-social system.
- The paradigm supports smart resource allocation, automatic network operation, and intelligent service provision.
B. The Vision: Dragnet
Dragnet is a cognitive Internet of Things vision for coordinated amateur-drone surveillance, combining machine and human crowd intelligence across fog-to-cloud infrastructure. It organizes surveillance into sensing, analytics, semantic knowledge discovery, and decision-making for detecting, understanding, and controlling drones.
- Dragnet tailors cognitive Internet of Things advances into a coordinated system for amateur drone surveillance.
- A joint fog-to-cloud framework uses local devices or people to sense environments and warn about drones, while cloud centers integrate data and make global decisions.Cloud actions may include jamming, capturing, or destroying amateur drones.
- The vision combines machine and human crowd intelligence to provide hybrid surveillance diversity and full-time, full-space coverage.Machines offer relatively high accuracy but are limited in number, while crowds contribute distributed surveillance data and coverage.
- Dragnet sequentially performs sensing, data analytics, semantic derivation and knowledge discovery, and intelligent decision-making.Analytics supports intruder detection, localization, and tracking; semantic processing derives meaning and discovers patterns or rules for intelligent operation.
- Sensing draws on cameras, sensors, radars, drones, and crowd participants, while decision-making determines drone presence and control actions.
IV. KEY ENABLING TECHNOLOGIES FOR DRAGNET
Dragnet’s key enabling technologies are detection, localization, tracking, and control. The paper surveys these techniques together with their technical challenges and open issues.
- The principal enabling technologies for Dragnet are detection, localization, tracking, and control.
- The paper presents an overview of each enabling technology alongside its technical challenges and open issues.
A. Detection
Drone detection can use acoustic, video, thermal, radar, and RF approaches, but each has limitations in range, weather sensitivity, target discrimination, emissions dependence, or health concerns. Combining heterogeneous sensors may improve performance, yet fusion is difficult because their data distributions differ.
- Audio, video, thermal, radar, and RF sensing offer distinct approaches to amateur-drone detection, each with specific merits and limitations.
- Acoustic sensors detect nearby drone presence, while video and thermal methods provide relatively short-range detection and are affected by weather.
- Radars are recognized as dominant drone detectors but struggle to distinguish small drones from birds and other clutter.
- Radar detection requires continuous transmission of RF radiation, which may raise health concerns near workplaces and residences.RF-based detection using emissions from the drone is effective but depends on the target transmitting RF signals.
- Combining detection approaches can provide diversity gain, but heterogeneous-sensor fusion is challenging because sensors may follow disparate probability distributions.Simple product or multivariate Gaussian models can yield suboptimal solutions; copula theory offers an alternative for coupling joint and marginal distributions.
2) Quickest Detection:
Quickest detection addresses the need to detect sudden amateur-drone intrusions rapidly despite accuracy–delay constraints. The surveillance system then requires passive three-dimensional localization without cooperation from the drone.
- Quickest Detection: Accurate amateur-drone detection is time-consuming, while sudden appearances and potential public risks make rapid detection important.Quickest detection means detecting an intruder drone in real time as quickly as possible, framed through optimal stopping theory.
- Localization: Localization is formulated as parameter estimation that supports subsequent tracking and control operations.The paper distinguishes this setting from GPS-, Bluetooth-, WiFi-, and cellular-network-based localization.
- Localization: Amateur-drone localization must estimate three-dimensional position, including latitude, longitude, and elevation.Most state-of-the-art localization studies discussed here focus on two-dimensional position estimation.
- Localization: Passive localization is needed because an intruding amateur drone may not actively share its position or signal features.The surveillance system should not require the drone to carry responding or positioning equipment.
C. Tracking
Tracking builds on localization by using location sequences and prior information to refine estimates. The paper highlights context-aware trajectory filtering, behavior prediction, multimodal sensing, and hybrid human–machine control.
- Tracking: Localization supplies the input and basis for tracking, while tracking can provide prior information that corrects localization errors.The two processes are described as inseparably connected.
- Tracking: Context-aware trajectory filtering and historical-data mining are identified as vital techniques for tracking amateur drones.The first addresses moving or flying trajectories, while the second predicts drone behavior from massive historical activity data.
- Tracking: Radar, video, and radio-frequency signal processing can support localization and tracking of high-flying amateur drones.The cited approaches provide different sensing capabilities for surveillance.
- Control: Drone control may use directional jamming, safe catching, or high-power destruction, depending on the application case.The paper notes that systems may choose one approach or combine several, while most current configurations rely on human operation.
- Control and Enabling Technologies: Hybrid human and machine intelligence is presented as an interesting challenge for highly efficient drone control.Additional Dragnet technologies include energy harvesting, full duplex, and cognitive radio networks for lifetime and spectrum objectives.
- Other Key Enabling Technologies: Full duplex and cognitive radio networks can improve spectrum efficiency and utilization when spectrum is scarce or unavailable.Energy harvesting and transfer is described as a way to prolong network lifetime when battery replacement or recharging is costly and inconvenient.
V. CASE STUDY: DRONE DETECTION AND CLASSIFICATION
The case study formulates detection and classification as a ternary hypothesis test distinguishing no drone, an authorized drone, and an unauthorized drone. Its simulations show tradeoffs between detection, false alarms, delay, and sensing resources.
- Problem Formulation: The ternary hypotheses represent no drone, an authorized drone, and an unauthorized amateur drone.The task is to determine whether a drone exists and classify it when present.
- Problem Formulation: The optimization maximizes correct detection of an unauthorized drone while constraining false alarms for no-drone and authorized-drone conditions.The two false-alarm probabilities correspond to misdetections when no drone or an authorized drone is actually present.
- Detection Scheme: The unauthorized-drone detection probability increases as false-alarm constraints α and β increase, and decreases when those constraints are tightened.This tradeoff is shown in the Fig. 4 simulation results.
- Detection Scheme: The proposed generalized likelihood ratio test addresses the unknown characteristics of unauthorized drones by treating them as a composite hypothesis.A genie-aided scheme with prior signal-strength distributions is included for performance comparison.
- Simulation Results: Increasing the number of samples improves detection performance but increases detection delay.The reported comparison applies to both evaluated schemes.
- Simulation Results: Increasing sensors and samples raises global detection probability, but incurs detection delay and multiple-sensor coordination costs.The marginal gain becomes minor when either the sensor count or sample count is sufficiently large.
VI. CONCLUSION
The article develops a cognitive-IoT-based amateur drone surveillance system called Dragnet. It surveys anti-drone techniques, discusses enabling technologies and open issues, and provides an exemplary illegal-drone detection simulation.
- Conclusion: The article develops an amateur drone surveillance system based on cognitive IoT, named Dragnet.Dragnet tailors the emerging cognitive IoT framework to amateur drone surveillance.
- Conclusion: The paper surveys state-of-the-art anti-drone techniques before presenting the Dragnet vision.It then discusses key enabling techniques together with technical challenges and open issues.
- Conclusion: An exemplary simulation demonstrates detection of illegal amateur drones.The conclusion describes this as an exemplary simulation rather than a comprehensive evaluation.
- Conclusion: The authors characterize amateur drone surveillance as a fruitful research direction and call for further research interest.They state that the article has touched only one tip of the iceberg.