Encyclopedia
Please note this is an old version of this entry, which may differ significantly from the current revision.
Subjects:
Computer Science, Cybernetics
The small-drone technology domain is the outcome of a breakthrough in technological advancement for drones. The Internet of Things (IoT) is used by drones to provide inter-location services for navigation. But, due to issues related to their architecture and design, drones are not immune to threats related to security and privacy. Establishing a secure and reliable network is essential to obtaining optimal performance from drones. While small drones offer promising avenues for growth in civil and defense industries, they are prone to attacks on safety, security, and privacy.
- aerial vehicles
- autonomous vehicles
- cyber-security
- Internet of Things
- machine learning
1. Introduction
Drones are commonly utilized for military and defense applications. They come in a range of sizes, from large, 200-foot war machines to small, inch-wide micro drones that fly through the air. Size is a critical factor in determining the appropriate use of a drone. Additionally, the flying range of a drone can vary significantly depending on the type, with some advanced military drones being capable of flying up to 17,000 miles without the need for ground control. Maximum flight time also varies based on factors such as altitude, surface area, and terrain. Drones can fly at varying heights, from just a few meters off the ground to as high as 65,000 feet [1].
2. Threats to Drone Security
Drone security measures comprise multiple layers and types, which are dependent on their usage, size, and control techniques. Typically, drones utilize an IEEE 802.11-based [13] Wi-Fi communication protocol [14] and include both ground stations and a Wi-Fi network. Due to the absence of encryption technologies in the drone, these gadgets are vulnerable to cyber-attacks and hijack [4]. Man-in-the-middle attacks, which can occur within a range of 2 km, are also a common method used to hijack drones [15]. In the military industry, the IoD has become increasingly popular, posing a challenge to privacy and security concerns during design [16]. To ensure data protection, privacy issues such as data accessibility, information leakage, encryption, and decryption techniques need to be addressed [17]. In recent years, researchers have identified four categories of security threats related to sensor- and protocol-based threats, jammers, and conceded components. Table 1 presents a literature review of these categories.
Table 1. Frequently occurring data privacy and cyber-security threats to smart drones.
| Attack | Cyber-Security Threats | Threats Found in | Countermeasures Introduced in |
|---|---|---|---|
| Protocol-based attacks | Security of communication link | [13,17,18,19] | [17] |
| Data confidentiality protection | [9] | ||
| Replay attack | [20,21] | [22] | |
| Privacy leakage | [9,19] | ||
| De-authentication attack | [4,13] | ||
| Sensor-based attacks | GPS-spoofing/-jamming attack | [12] | [23] |
| Motion sensor spoofing | [24] | [25] | |
| UAV-spoofing/-jamming attack | [12] | ||
| Compromised component | IoT security threats | [12] | |
| Control/data interception | [12,18] | ||
| Jammers | Denial of service | [4,12,13] | |
| Stop packet delivery | [16] | [16] |
Table 1 demonstrates the review of the literature that primarily focused on identifying cyber-security loopholes in drones, with limited discussions on potential solutions. One potential research avenue has been to utilize encryption algorithms to ensure safe and secure data transmission between the drone and its base station [16]. Small drones have obtained popularity due to their size and likely peril to the government and general public’s data privacy [26]. Researchers also established a risk challenge for drones [14,27,28,29,30]. For instance, Tian proposed an operative and smart validation model for the IoD assisted by edge, ensuring the drone networks’ data-related security [31]. Similarly, a system was presented by Hell to ensure the safety of drone data in a commercial industrial/factory area [2]. In 2019, a gas leakage-sensing drone idea was projected by the authors to ensure timely action to curb the fatal scenario [3]. Drones are mainly used for monitoring in the agriculture and security fields.
Over the last decade, drone-related security threats are the talk of the town in the research arena. The privacy issues associated with smart-city drone applications are discussed in [19], and Table 1 highlights other important issues. Drone network attacks, prospects, and limitations are also the interest areas of researchers in the cyber-security domain [32]. The business sector has similar challenges and applications, as presented by similar studies [5,33,34] using blockchain/crypto technology using 5G and drones based on the IoT for the safe transmission of data [34]. This system has limitations in manually identifying the intensity and nature of the threats. A secure and smartly effective drone system with the ability to investigate attacks and implement security measures for drone data integrity is the need of the hour. Some studies have attempted to solve device authentication problems by using key agreement [35] and key-enabling data [6] for secure drone data delivery. Commercial drones [6,35,36,37] are facing the general issue of the hijacking of drones, UAVs, and drones in the agricultural sector [22,38] aided by the IoT. Solutions to these general issues are proposed in [7] and [8]. Another concern relating drones and UAVs is GPS (global positioning system) tracking [39], which requires robust, authentic, and foolproof resolution. Drone interception and hijacking are also part of the studies carried out in this domain [23,24,25,40].
3. Implementation of Drone Security with Machine Learning
ML techniques are classified into semi-supervised, supervised, and unsupervised categories. Cloud computing [41], mobile networks [42], IoT systems [43], and sensor-based wireless networks [44] are the areas where ML models have been widely used by researchers to handle cyber-attacks. For example, self-learning models were combined with supervised learning models by Vedula et al. [45] to use two features to detect DDoS attacks. They combined LSTM Autoencoder and RF classifier. For the scenarios of all and sparse traffic, their window identification approach achieved accuracy rates of 94% and 93.5%, respectively.
ML techniques are classified into semi-supervised, supervised, and unsupervised categories. Cloud computing [41], mobile networks [42], IoT systems [43], and sensor-based wireless networks [44] are the areas where ML models have been widely used by researchers to handle cyber-attacks. For example, self-learning models were combined with supervised learning models by Vedula et al. [45] to use two features to detect DDoS attacks. They combined LSTM Autoencoder and RF classifier. For the scenarios of all and sparse traffic, their window identification approach achieved accuracy rates of 94% and 93.5%, respectively. Their proposed hybrid LSTM-RF model showed the best results, with a window size of 100.
No research on ML model usage in drone networks for cyber-attack recognition was found. However, another study suggested a probabilistic approach in a constrained cyber–physical system to control and detect actuation attacks [46]. Their research was primarily concerned with the PA2 attack, in which the attacker blocks communication between the actuators and the controller. Based on a hypothesis-testing methodology, a group of parallel detectors was suggested. The detection and control goals were written as two distinct stochastic objective functions using a probabilistic technique to cope with uncertainty. The authors also proposed a drone security access control system and have previously used ML for wireless networks (wi-net) security systems, as shown in Table 2.
Table 2. Machine learning for frequently occurring data privacy and cyber-security threats to smart drones.
| Attack | Security Technique | Machine Learning Solution |
|---|---|---|
| Jamming | Secure offloading | Q-learning [42,44], DQN [47] |
| Denial of service | Secure offloading | Neural networks [41], Multivariate correlation analysis [48], Q-learning [49] |
| Malware | Access control | Q/Dyna-Q/PDS [50], K-nearest neighbors [51], Random Forest [51] |
| Intrusion | Access control | Naive Bayes [43], Support vector machine [43], neural network [52], K-NN [53] |
| Spoofing | Authentication | SVM [54], DNN [55], Dyna-Q [56], Q-learning [56] |
| Traffic blockage | Authentication | Q-learning [57] |
The research suggests that there is a need for a comprehensive solution to address cyber-security threats and ensure the safety of drone data. While many studies have identified the challenges and issues related to drone security, few have proposed effective solutions to mitigate these risks [34]. The use of machine learning models has shown promise in dealing with cyber-attacks in various networks, but there is a lack of research on their application in drone networks. Additionally, the authentication system proposed in some studies may not be suitable for IoT-based drone networks. Therefore, this research gap needs to be addressed to make drones compliant for the industry and for commercial use while ensuring their security and privacy.
Existing studies related to drone security have certain limitations that need to be addressed. First, the architecture and design of small drones have not received sufficient attention, resulting in vulnerabilities that can be exploited by potential attackers. Additionally, the current data transformation and privacy mechanisms of small drones may not align with the specific requirements of the domain, leaving them susceptible to security breaches. Furthermore, while the Internet of Things (IoT) is utilized for inter-location services in drones, there is a lack of comprehensive research on establishing secure and reliable networks for optimal drone performance. Moreover, previous studies have not fully explored the integration of intelligent machine learning models into the design and structure of IoT-aided drones, which could enhance their adaptability and security. Overall, these limitations highlight the need for further research and development to overcome security challenges and ensure the resilience of drone systems in an interconnected world.
To ensure drone security, a smart vigilant system is a prerequisite to investigate the attacked data automatically and take corrective measures according to the scenario and the situation at hand without in-person interference. ML models have previously been deployed for mobile-based and wireless sensor-based networks for cyber-security, but they are yet to be applied to the security of drone-based vehicles. This study addresses the issue of access control authentication methods for drones with an ML-based solution.
This entry is adapted from the peer-reviewed paper 10.3390/s23167154
This entry is offline, you can click here to edit this entry!
