A Review of Physical Layer Security in Aerial–Terrestrial Integrated Internet of Things: Emerging Techniques, Potential Applications, and Future Trends

Abstract

The aerial–terrestrial integrated Internet of Things (ATI-IoT) utilizes both aerial platforms (e.g., drones and high-altitude platform stations) and terrestrial networks to establish comprehensive and seamless connectivity across diverse geographical regions. The integration offers significant advantages, including expanded coverage in remote and underserved areas, enhanced reliability of data transmission, and support for various applications such as emergency communications, vehicular ad hoc networks, and intelligent agriculture. However, due to the inherent openness of wireless channels, ATI-IoT faces potential network threats and attacks, and its security issues cannot be ignored. In this regard, incorporating physical layer security techniques into ATI-IoT is essential to ensure data integrity and confidentiality. Motivated by the aforementioned factors, this review presents the latest advancements in ATI-IoT that facilitate physical layer security. Specifically, we elucidate the endogenous safety and security of wireless communications, upon which we illustrate the current status of aerial–terrestrial integrated architectures along with the functions of their components. Subsequently, various emerging techniques (e.g., intelligent reflective surfaces-assisted networks, device-to-device communications, covert communications, and cooperative transmissions) for ATI-IoT enabling physical layer security are demonstrated and categorized based on their technical principles. Furthermore, given that aerial platforms offer flexible deployment and high re-positioning capabilities, comprehensive discussions on practical applications of ATI-IoT are provided. Finally, several significant unresolved issues pertaining to technical challenges as well as security and sustainability concerns in ATI-IoT enabling physical layer security are outlined.

1. Introduction

1.1. Background and Motivations

Aerial–terrestrial integrated Internet of Things (ATI-IoT) represents a complex network structure that aims to achieve comprehensive data perception, transmission, and processing by integrating aerial platforms with terrestrial IoT systems [1,2,3]. This integrated technology primarily utilizes the broad view and rapid deployment capabilities of aerial platforms such as high-altitude platform stations (HAPSs), low-altitude platform stations (LAPSs), and drones, combined with the dense network of sensors and actuators on the ground, forming a multi-layered and efficient network system [4,5,6]. As shown in Figure 1, in this system, aerial platforms not only provide extensive sensing and communication coverage but also respond quickly in critical moments to supplement the shortcomings of ground facilities [7,8,9,10]. Meanwhile, terrestrial IoT devices are responsible for collecting fine-grained environmental data, achieving precise monitoring and control over localized areas [11].

With the development of 5G and B5G technologies, the potential of ATI-IoT has been tremendously unleashed [12]. The high speed, large bandwidth, and low latency characteristics of 5G networks provide powerful data processing and transmission capabilities for terrestrial IoT devices, enabling real-time data analysis. Additionally, the dense deployment of 5G networks enhances the connection stability and service range of terrestrial IoT devices [13,14,15]. In the future, the overall performance of the B5G network is expected to be further enhanced by increasing the frequency and bandwidth. In the B5G era, the ATI-IoT has a wide range of application prospects. For instance, aerial platforms (e.g., HAPSs and drones) can be used for temporary coverage and combined with cellular networks for traffic management, public safety, and traffic flow monitoring [16,17,18,19].

However, as shown in Figure 2, due to the characteristics of wireless communication networks, aerial–terrestrial integrated information transmission faces numerous security threats [20,21,22,23,24,25]. Therefore, in ATI-IoT, we can adopt the STRIDE threat model to analyze potential security threats. The STRIDE model covers six important types of security threats: spoofing, tampering, repudiation, information disclosure, denial of service, and elevation of privilege. Specifically, spoofing refers to attackers impersonating legitimate users or devices. In ATI-IoT, the identity verification of aerial platforms and ground devices can be vulnerable to attacks. By spoofing, attackers can gain unauthorized access to communication data, thus compromising the security of the ATI-IoT system. Next, tampering refers to attackers modifying transmitted data. In ATI-IoT, attackers can intercept and tamper with this data, leading to incorrect data transmission. Finally, denial of service attacks are common network threats that aim to exhaust the resources of aerial platforms by sending a large volume of meaningless requests, ultimately causing the service to become unavailable.

• Information Eavesdropping: Due to the broadcast nature of ATI-IoT, any user within communication coverage can receive information, so aerial–terrestrial integrated communications are at risk of eavesdropping [20]. Currently, ATI-IoT mainly uses the upper-layer key encryption technique to protect information security. However, the security of the key encryption technique is based on the limited computing capacity of eavesdropping users [21]. Therefore, the application area of key encryption techniques is limited. In addition, complex encryption algorithms and key distribution mechanisms also limit the use of key encryption techniques in future miniaturized and dense B5G networks [22].
• Information Interference: A large number of communication devices and data transmission requirements make various wireless signals superimposed on each other [23]. However, limited spectrum and space resources make it difficult to ensure the uninterrupted transmission of information in the current region. Therefore, with the use of interference within the network and interference between networks, malicious interference users can transmit interference signals to interfere with or even prevent legitimate users from correctly receiving information, resulting in the interruption of information transmission [24].
• Unauthorized Access: Due to the adoption of unified communication protocols and standards, ATI-IoT is easy to access and configure. However, malicious users can pretend to be legitimate users according to the characteristics of the current network and signals, illegally access the network, and collect users’ private information, resulting in economic losses [25].

In the face of the above security threats, ensuring the security, reliability, integrity of information and the legitimate use of resources has become the key issues of ATI-IoT. In this situation, physical layer security techniques can be used in ATI-IoT to improve the security of the B5G networks [26,27,28]. The key idea behind physical layer security is to adopt the intrinsic randomness of the aerial–terrestrial integrated transmission channel to ensure security in the physical layer [29]. Meanwhile, the main design goal of physical layer security techniques is to enhance the performance difference between the link of the legitimate receiver and that of the eavesdropper by using well-designed transmission schemes [30].

1.2. Related Works

In recent years, the interest in ATI-IoT has surged, and the currently deployed ground communication infrastructure is inadequate to satisfy the demands of emerging applications and user cases [20,21,27]. Meanwhile, physical layer security is crucial for ATI-IoT systems, ensuring robust protection against unauthorized access and interference [24]. Particularly, by using the physical layer security techniques, we can address vulnerabilities in the physical transmission medium, thereby enhancing the reliability of ATI-IoT systems [31].

Table 1. List of studies on solutions for wireless communication security and aerial–terrestrial integrated networks.

Reference	Focus	Content Type	Physical Layer Security	Aerial–Terrestrial Integrated Communications	Emerging Techniques
[32,33,34]	These works propose the comprehensive and expandable frameworks for classifying PLS techniques, categorizing them into signal-to-interference-plus-noise ratio-based and complexity-based approaches, and reviewing their applications in various communication systems.	Theoretical and analytical	These works exploit wireless channel characteristics, such as noise, fading, and interference, to design transmission schemes that enhance the security of legitimate users while preventing eavesdropping.	×	×
[35]	These works exploit wireless channel characteristics, such as noise, fading, and interference, to design transmission schemes that enhance the security of legitimate users while preventing eavesdropping.	Analytical	The paper explores the application of PLS coding to enhance data confidentiality using the intrinsic randomness of the transmission channel.	×	×
[36,37,38]	These works provide a comprehensive review of covert communication as a promising solution for securing wireless networks, emphasizing its principle, applications, and the challenges it faces in practical implementation.	Theoretical	These works focus on utilizing randomness to conceal transmitted signals in covert communication networks, ensuring security by avoiding detection by the warden.	×	These works discuss the application of covert communication in various network topologies, along with the emerging techniques aimed at enhancing its security and practicality in future networks.
[39,40,41]	These works present comprehensive surveys of the air-to-ground propagation channel models, focusing on both large and small-scale fading models, their limitations, and the future research directions for UAV secure communication.	Theoretical and analytical	These works emphasize the need for accurate air-to-ground propagation models for secure and efficient UAV communication.	Accurate air-to-ground propagation models can potentially enhance the security of ATI-IoT by ensuring robust and reliable UAV communication.	×
[42,43,44]	These works provide comprehensive surveys of computational intelligence-based algorithms designed to enhance the networking and collaboration capabilities of ATI-IoT.	Analytical	×	To improve the performance of ATI-IoT, task allocation, path planning, and communication efficiency can be optimized.	These works highlight emerging computational intelligence-based technologies such as heuristic behavior search, policy learning, and hybrid algorithms that aim to address the challenges in ATI-IoT.

summarizes the contributions of surveys in the field of wireless communication security and aerial–terrestrial integrated networks, the aerial–terrestrial integrated architecture is shown in Figure 3. The aerial–terrestrial integrated architecture consists of various network entities, including the HAPSs, LAPSs, drones, and IoT devices [13,24]. Specifically, the HAPSs provide wide coverage and serve as reliable communication hubs in remote or underserved areas. The LAPSs are often deployed in urban or suburban environments to enhance the reliability of data transmission. Drones serve as mobile nodes that can quickly adapt to dynamic environments, facilitating flexible deployment for specific applications like emergency response or disaster relief. IoT devices are terminals in the ATI-IoT architecture that collect and transmit data for various applications. Each of these network entities may be at risk of unauthorized access and data breaches.

Specifically, the surveys in [32,33,34] examined the research and application of physical layer security techniques beyond 5G (B5G), covering the integration of innovative transmission techniques with physical layer security, and addressing emerging security threats in B5G. The survey in [35] analyzed the new security challenges faced by B5G communications and reviewed physical layer security techniques that could address B5G security issues, including physical layer identity authentication and confidential transmission. The survey in [36] introduced physical layer security transmission techniques in device-to-device (D2D) cellular networks, focusing on resource allocation, power control, and interference coordination. Moreover, the survey in [37] presented the working mechanism and application prospects of intelligent reflecting surfaces (IRSs). Then, this survey summarized existing research on IRS-assisted physical layer security from the perspectives of information-theoretic security and covert communication, elucidating the model characteristics, optimization objectives, and key methods of various research hotspots. In [38], the authors systematically delineated the disparities and correlations between wireless covert communication and its associated concepts, with a specific focus on research pertaining to wireless covert communication grounded in hypothesis testing theory and information theory. Furthermore, the fundamental research models and specific model classifications were introduced.

The integration of air and ground communications networks represents a future trend in wireless communications development. In [39,40,41], the authors introduced the channel models for aerial–terrestrial integrated communications, and summarized techniques commonly used in physical layer security, such as the eavesdropper coding, beamforming, artificial noise, relay-assisted interference, and key encryption. They also highlighted the challenges faced by physical layer security in aerial–terrestrial integrated networks. Additionally, in order to satisfy the application requirements of intelligent information services, the authors in [42,43,44] explored strategies for restructuring aerial–terrestrial integrated networks within dynamic spatial resource constraints by using the significant advantages of spatial information networks in terms of coverage and situational awareness.

Although the aforementioned surveys have significantly advanced the application of wireless network security techniques in ATI-IoT systems, there are still several unresolved issues that require further exploration. First, the surveys in [32,33,34,35,36,37,38] primarily focus on the fundamental principles and related applications of existing physical layer security techniques. However, these overviews seldom address the specific operations within aerial–terrestrial integrated communications, resulting in a gap in solutions that are supported by aerial platforms. Second, the surveys in [39,40,41] concentrate on aerial–terrestrial integrated communication scenarios and provide further explanations on the performance improvements achieved by combining physical layer security techniques with aerial–terrestrial integrated communications. Unfortunately, these summaries primarily cover some traditional physical layer security methods, such as eavesdropper coding, beamforming, and artificial noise, while emerging techniques like IRS, D2D communications, and covert communications are not sufficiently discussed. Finally, the authors in [42,43,44] explore the potential applications of ATI-IoT, but they rarely focus on the critical aspect of wireless communication security. Motivated by the above discussion, this work aims to fill the gaps left in the existing surveys.

1.3. Contributions

This review comprehensively investigates the integration of physical layer security in ATI-IoT, distinguishing itself from previous studies by addressing the specific needs and challenges of aerial–terrestrial communications. The contributions of this work can be summarized as follows.

• Overview of Physical Layer Security and ATI-IoT Architectures: This review provides a detailed exploration of the fundamental principles of physical layer security in ATI-IoT, focusing on the mechanisms of key generation and agreement. Unlike prior work that primarily discusses general physical layer security techniques, this paper emphasizes how these techniques are applied within the aerial–terrestrial integrated architectures. In addition, an in-depth discussion of their components and functions is offered.
• Emerging Techniques for Enhancing Physical Layer Security: This work highlights several innovative technologies enabling physical layer security in ATI-IoT systems, including IRS-assisted networks, D2D communications, covert communications, and cooperative transmissions. Compared with previous research that focused on traditional methods, this paper explores the latest developed technologies and analyzes their applications in ATI-IoT.
• Potential Applications: This review provides valuable insights into the practical applications of ATI-IoT with physical layer security, particularly in areas such as emergency communications, vehicular networks, and intelligent agriculture. By considering the flexible deployment capabilities of aerial platforms, this paper provides novel perspectives on how ATI-IoT systems can be utilized in real-world scenarios to improve the security.
• Open Issues and Future Directions: Finally, this work discusses key unresolved challenges related to technical, security, and sustainability aspects in ATI-IoT systems. We outline potential research directions and suggest practical solutions for overcoming these challenges, ensuring the continuous advancement and adoption of physical layer security techniques in future ATI-IoT applications.

1.4. Structure

The structure of this review is as follows. First, Section 2 gives the technical principles of physical layer security. In addition, this section presents the state-of-the-art in architecture designs for ATI-IoT and provides details on endogenous safety and security. Then, Section 3 includes the emerging techniques for ATI-IoT enabling physical layer security, while Section 4 focuses on the potential applications. Subsequently, the future trends are discussed in Section 5. Finally, conclusions are given in Section 6. Overall, the structure of this review is shown in Figure 4.

2. Preliminaries of Physical Layer Security

2.1. Wireless Communication Security

The rapid development of wireless communication network technology has brought great convenience but also faces serious security challenges [45,46,47]. In traditional wireless networks, security issues mainly manifest in the following aspects. First, the openness of wireless channels makes it easy for attackers to obtain personal privacy information [48]. Due to the widespread transmission of wireless signals, attackers can eavesdrop on these signals to steal sensitive user information. Moreover, traditional secrecy technologies, mostly based on key management systems, have gradually shown their inefficiency and limited applicability with the increase in communication demands and the complexity of key management [49]. Second, limited spectrum and spatial resources have become targets for attackers. They maliciously interfere with the information transmission of users by generating massive requests or interference signals, reducing the quality of network services, and increasing network congestion and delays [50].

With the promotion of 5G technology, the variety of network-connected devices has increased, distributing personal privacy data across various devices and complicating data protection [51,52]. Therefore, in order to ensure information security, the endogenous security architecture model has been proposed [53,54,55], as illustrated in Figure 5. Although the high speed, low latency, and large-capacity features of 5G networks provide convenience to users, they also bring more security challenges [56]. For instance, edge computing reduces data transmission distances and leakage risks by processing data near the data source [57,58,59,60]. However, network edge nodes are often vulnerable to attacks due to the lack of secure communication protocols and encryption measures. In the future, the development of B5G technology is expected to further enhance the security and intelligence levels of communication networks. Physical layer security techniques, such as the use of IRS, are key to improving wireless network security [61,62,63,64]. IRS can optimize signal coverage and spectrum resource utilization, enhancing the confidentiality and integrity of data transmissions [65]. In addition, using covert communication and cooperative transmission schemes, private information can be securely distributed among users.

For physical layer security techniques, the following performance metrics are commonly used to measure the security [66,67,68].

• Secrecy Capacity: Secrecy capacity refers to the maximum rate at which secret information can be transmitted over a communication channel without being intercepted. It is calculated by the difference between the channel capacities of the legitimate user and the eavesdropper.
• Secrecy Outage Probability: Secrecy outage probability is the probability that the secrecy capacity falls below a given threshold. It represents the possibility of unsafe communication due to changes in channel conditions.
• Secure Throughput: Secure throughput is the average rate of secure data transmission over a network. It accounts for both the legitimate user’s transmission rate and the eavesdropper’s interference.
• Key Generation Rate: Key generation rate is the rate at which secret keys are generated for secure communication. It is often used in key establishment protocols, where a secret key is derived from the physical layer’s channel properties.

In addition to evaluation metrics, security-related standards are also critical for ATI-IoT. Specifically, the IEEE 802.15.4z standard, which specifies secure ultra-wideband (UWB) communication, plays a significant role in enhancing security for low-power, short-range communication devices [69]. The IEEE 802.15.4z standard introduces features such as time-of-flight measurements for accurate ranging and positioning. However, it has limitations in terms of scalability and long-range communication, which will restrict its use in ATI-IoT. Moreover, emerging 3GPP standards for non-terrestrial networks (NTNs) aim to improve connectivity in remote or underserved areas [70]. These standards offer enhancements in latency, throughput, and reliability for ATI-IoT applications that require global coverage. However, the complexity of integration and the potential for higher operational costs present challenges in deploying NTN for large-scale ATI-IoT systems.

2.2. Shannon’s Theory on Information-Theoretic Security

Claude Shannon’s theory of secrecy systems, first published in 1949, is a cornerstone in understanding secure communication over noisy channels [22,71,72]. Shannon’s work primarily focuses on the theoretical underpinnings of encryption and the ability to securely transmit information. Specifically, Shannon defined a secrecy system as a method to transmit encrypted information from a sender to a receiver, ensuring that a potential interceptor (eavesdropper) cannot understand the message [73,74]. The core idea is to transform a clear text message M into a cipher text C using an encryption algorithm, which depends on a secret key K. The encrypted message is transmitted over a channel where it might be intercepted, but without the key, the original message is unintelligible to the interceptor. The fundamental equation introduced by Shannon is

$$\begin{array}{c}\hfill C=E_K\left(M\right),\end{array}$$

(1)

where E represents the encryption function, K is the secret key, and M is the original message. Figure 6 gives a diagram of Shannon’s theory.

Meanwhile, Shannon introduced the concept of perfect secrecy, which occurs when the cipher text does not reveal any information about the plain text, mathematically expressed as

$$\begin{array}{c}\hfill P\left(\left.M\right|C\right)=P\left(M\right).\end{array}$$

(2)

This means that the probability of guessing the message M given the cipher text C is the same as the probability of guessing the message without seeing the cipher text. For perfect secrecy, the key needs to be at least as long as the message and should be used only once (one-time pad) [23,28,75]. In addition, Shannon discussed entropy as a measure of uncertainty or randomness in the system. The higher entropy of the key increases the security of the system. He also noted that redundancy in the language (e.g., the predictability of English) reduces the amount of security that can be achieved, as it provides clues to the interceptor. Shannon’s work highlighted that the security of a system is bounded by the entropy of the key used in relation to the entropy of the message [76,77,78]. If the entropy of the key is lower than the entropy of the message, perfect secrecy cannot be achieved. While Shannon’s theory provides a framework for understanding and achieving secure communication, it also shows the limits and challenges. The reason is that for long message or real-time communication systems, if Shannon’s theory is used to ensure information security, the key length and randomness cannot be satisfied [79].

Shannon’s theory on information-theoretic security provides a solid foundation for how to achieve secure transmission of communication systems through physical layer security techniques. For instance, beamforming can improve the security of ATI-IoT [80]. The core principle of beamforming is to direct the signal to a predetermined receiver by adjusting the transmission mode of the signal. Meanwhile, this technique can minimize the leakage of signals to eavesdroppers. In this case, the signal energy can be concentrated in a specific direction [81,82,83]. Artificial noise is another important technique in physical layer security, which further obscures communication by adding noise to the transmitted signal [84]. The principle of artificial noise is to protect the confidentiality of the data by degrading the quality of the signal received by the eavesdropper while having little effect on the legitimate recipient [85]. The combined use of beamforming and artificial noise techniques in the ATI-IoT can further enhance security.

2.3. Key Generation and Agreement

2.3.1. Key-Based Security and Keyless Security

In 1978, Csiszar further explored the issues of secure transmission under broadcast and Gaussian channel conditions. In 1993, Maurer and Csiszar researched the fundamental models and key capacity problems of physical layer security key generation [86]. Csiszar and Körner generalized Wyner’s work, proposing the secure capacity and the secrecy outage probability of a discrete memoryless wiretap channel: the probability that the instantaneous secure rate falls below a specific threshold through signal processing techniques that constrain the eavesdropper’s error rate or the signal-to-noise-plus-interference ratio (SINR) to a specific level. As discussed above, research on physical layer security based on information theory branched into two directions: key-based security (Shannon and Maurer) and keyless security (Wyner) [87,88,89].

Key-based security, as developed by Shannon and Maurer, revolves around the idea of using shared secret keys for encrypting messages [90]. In this model, security is ensured by the inability of an eavesdropper to access the secret key. This approach relies heavily on the initial sharing and continuous protection of the key. It requires robust methods for key distribution and management, which can be a challenge in dynamic or large-scale ATI-IoT networks. The strength of this security method directly depends on the secrecy and randomness of the used key [91].

On the other hand, keyless security, introduced by Wyner and expanded by his successors, leverages the physical characteristics of the communication channels to ensure security [92,93,94]. This method does not rely on a shared secret key but instead uses the inherent unreliability and noise properties of the communication medium to hinder potential eavesdroppers. It essentially turns the imperfections of the physical layer into a tool for securing data, making it particularly useful in environments where key management is impractical or impossible.

2.3.2. Key Agreement Mechanism

Ensuring the confidentiality and integrity of messages is the most fundamental requirement of secure communication. By using the randomness, time variability, and uniqueness of wireless channels, wireless channel information can be quantified into wireless keys with quasi-quantum security characteristics [95]. To obtain a symmetric key usable for encrypting wireless sessions, it is necessary for both parties in wireless communication to access reciprocal channel information. For time division duplexing (TDD), because the uplink and downlink transmissions alternate within a short gap, it is easier to obtain symmetric channel information using the reciprocity of the TDD channel [96]. However, in frequency division duplexing (FDD) mode, it is challenging for communication parties to obtain symmetric wireless channel information.

In the FDD channel environment, extracting physical layer keys requires reciprocity compensation and correction of the FDD channel. Typically, large-scale fading parameters of the channel (such as path loss and shadow fading) exhibit reciprocity [97,98,99,100]. By contrast, small-scale fading parameters (such as multipath and Doppler) often show partial asymmetry, especially in the amplitude and phase parameters of uplink and downlink multipaths. Differences in multipath amplitudes can be mitigated through normalization, while differences in multipath phases can be resolved through parameter estimation and compensation [101].

After the generation of wireless channel keys, the performance of physical layer key agreement algorithms is typically described using parameters such as key generation rate, key mismatch rate, and randomness [77,102]. The key generation rate, defined as the number of key bits generated per channel sample, describes the efficiency of key generation. In addition, the key mismatch rate refers to the probability of mismatch between keys generated by legitimate communication nodes, and the higher the randomness of the key sequence, the greater the security of confidential transmission [79,92]. To achieve high key agreement performance, advanced techniques (such as quantizing wireless channel keys and large-scale antenna arrays) are often combined.

3. Emerging Techniques in ATI-IoT to Guarantee Information Security

3.1. Physical Layer Security Enhancement Exploiting IRS

The IRS consists of a large number of small, passive, electronically controlled reflective elements. Each element can independently adjust the phase of the incident signal, thereby controlling the direction and strength of the reflected signal [103,104,105]. Typically, these components are made of electromagnetic materials and optimized for signal reflection based on real-time environmental and network conditions. The IRS is managed by an intelligent controller. The controller can interact with the base station or the central processing unit [106]. As shown in Figure 7, deploying IRS can significantly improve the overall security of ATI-IoT by providing an additional communication link for the transmission of confidential information, allowing it to bypass eavesdroppers and reach legitimate receivers. Therefore, for ATI-IoT, IRS can be used as a key enabling technique to improve the security of the physical layer [107].

For example, in complex urban environments, the line-of-sight (LoS) links between aerial platforms and IoT devices are frequently obstructed, which significantly degrades channel quality. To maximize the secure transmission rate, literature [108,109,110] optimized the phase shift coefficients of IRS. Study in [111] explored the minimization of system transmission power under constraints of a certain secure transmission rate, through joint optimization of transmitter power allocation and IRS reflective phase shifts. The literature in [112] proposed a joint optimization of base station active beamforming and IRS-assisted beamforming, utilizing artificial noise at the physical layer to maximize the system’s secure rate. Moreover, the IRS was deployed on drones as a movable relay to assist communication, with joint optimization of user scheduling, phase shifts, and drone trajectories to maximize the minimum average secure transmission rate among multiple users [113]. In the literature [114,115], drones equipped with IRS were considered for enhancing the communication capability between ground base stations and vehicular users. Then, an iterative algorithm was developed to jointly optimize IRS phase shifts, ground vehicle scheduling, drone trajectories, and power allocation, with the goal of maximizing the minimum throughput for vehicular users. Furthermore, literature in [116] investigated an IRS-assisted multi-drone communication system enabled by non-orthogonal multiple access (NOMA) communications. This study conducted joint optimization of drone deployment, decoding order of the NOMA technique, and phase shifts to maximize the secure transmission rate of ATI-IoT systems. Overall, key considerations are outlined in

Table 2. List of papers on IRS-assisted security communications.

Reference	Optimization Objective	Method	IRS Type
[108,109,110]	Secure transmission rate maximization	Optimization of phase shift coefficients	Fixed IRS
[111]	Transmission power minimization	Optimization of power allocation and phase shift, and consideration of secure transmission rate	Fixed IRS
[112]	Secure transmission rate maximization	Optimization of beamforming, and consideration of artificial noise	Fixed aerial IRS
[113]	Minimum average secure transmission rate maximization	Optimization of power allocation, phase shifts, and drone trajectories	Mobile aerial IRS
[114,115]	Secure transmission rate maximization	Optimization of user scheduling, phase shifts, and drone trajectories	Mobile aerial IRS
[116]	Secure transmission rate maximization	Optimization of drone deployment, decoding order, and phase shift	Fixed aerial IRS

.

3.2. Physical Layer Security in D2D-Aided Cellular Networks

When eavesdroppers are present in ATI-IoT, the combination of D2D communication and physical layer security techniques can significantly enhance network security. This is because D2D communication allows devices to connect directly, reducing reliance on central nodes that could be targets of attacks [117,118,119]. As shown in Figure 8, in multiplexing modes, the spectrum resources of a cellular user are shared with multiple D2D pairs, and a D2D pair can also reuse the spectrum resources of several cellular users. Despite the mutual interference caused by the sharing of spectrum resources among multiple IoT devices, the implementation of appropriate spectrum reuse and power control strategies has effectively mitigated co-channel interference [120,121]. This has resulted in an increased gap between legitimate and eavesdropping channels, thereby potentially facilitating secure communications.

The authors in [122] focused on drone-assisted cellular networks with the goal of maximizing the energy efficiency of secure communications by jointly optimizing user scheduling and drone three-dimensional flight trajectories. The works in [123,124,125] addressed scenarios in ATI-IoT environments where active attacks were conducted by full-duplex eavesdroppers. Meanwhile, LAPSs were capable of tracking and interfering with eavesdropping channels. Inspired by the above, these studies aimed to maximize the secrecy capacity and proposed trajectory optimization algorithms based on Q-learning. In [126,127,128], the authors introduced covert communication strategies that integrated mode selection and cooperative interference, allowing each user to adaptively switch between half-duplex and full-duplex modes. In addition, idle D2D users collaborated to send interference signals to disrupt eavesdroppers, thus enhancing the secure transmission performance. Overall, key considerations are outlined in

Table 3. List of papers on D2D-aided ATI-IoT.

Reference	Optimization Objective	Method	Operating Mode
[120,121]	Co-channel interference minimization	Optimization of spectrum reuse and power control strategies	Half-duplex mode
[122]	Energy efficiency maximization of secure communications	Optimization of user scheduling and drone three-dimensional flight trajectories	Full-duplex mode
[123,124,125]	Secure transmission rate maximization	Optimization of resource allocation, and consideration of active attacks	Full-duplex mode
[126,127,128]	Secure transmission rate maximization	Optimization of mode selection, and consideration of cooperative interference and covert communication	Half-duplex and full-duplex modes

.

3.3. Covert Communications in ATI-IoT

As illustrated in Figure 9, ATI-IoT systems boast advantages in remote information transmission, overcoming obstacles related to geographical conditions, operational environments, and transmission distances. However, the presence of various forms of interference, interception, and detection in the aerial–terrestrial integrated environment continues to pose significant challenges to ensuring secure, timely, and effective communications [129,130]. Thus, in complex and open network settings, combining ATI-IoT with covert communication strategies can enhance the security of information transmission.

As described in [131], methods for implementing covert communication generally fall into two categories: methods related to the physical layer and methods related to the application layer. The former typically relies on techniques such as short-frame transmission, spread spectrum, and multi-beam forming, while the latter typically depends on the application of encryption algorithms. However, if the system complexity increases further, neither of these methods can be well applied. Therefore, researchers were dedicated to developing a more efficient aerial–terrestrial integrated covert communication scheme to meet the new demands of the B5G era. The literature in [132] developed a multi-carrier direct sequence spread spectrum system for achieving covert transmission in aerial–terrestrial integrated communications, with a key emphasis on the pre-emptive detection of electromagnetic signals in the environment. Then, the covert signals were transmitted within the same band as the sensed signals. In this situation, the useful data were hidden within the environmental electromagnetic noise, thereby enhancing the concealment of information transmission in ATI-IoT systems. In [133], the authors proposed a polar coding algorithm based on non-pilot-assisted parameter estimation and compensation. Compared with the traditional carrier synchronization algorithms, the carrier synchronization performance of covert communication systems was effectively improved.

Traditional research in covert communication primarily focused on channel models involving additive white Gaussian noise. As exploration in this field deepened, attention gradually shifted toward multiple-input multiple-output (MIMO) channels [134] and multi-user channels [135]. In traditional schemes, a single orthogonal resource block could only serve one user, which limited the potential for further enhancements in spectrum efficiency and channel capacity within ATI-IoT systems. In terrestrial 5G mobile communications, the NOMA technique was adopted. By grouping multiple users and allocating power properly, the same resource block was reused in a non-orthogonal manner. Meanwhile, the signals were separated at the receiver, thereby improving communication capacity and bandwidth utilization [136,137,138]. Building on this, the extension of this scheme to ATI-IoT could alleviate the problem of the shortage of frequency band resources. Moreover, combining the NOMA technique with covert communications has become one of the current research focuses. The authors in [139] explored multi-user access methods for the uplink in aerial–terrestrial integrated communications and compared the differences in covert performance among three mainstream NOMA schemes.

Furthermore, drones are widely utilized to enhance physical layer security against eavesdropping attacks. For example, the authors in [140] improved the secrecy performance by optimizing the flight trajectory of drones. The literature in [141] investigated the downlink security communication of drone networks, prohibiting their flight over no-fly zones. Then, they optimized their trajectories and transmission power within a given flight period to maximize the average secrecy rate. Additionally, the authors in [142] jointly optimized the position and transmission power of drones, enhancing the security of drone-enabled relay networks. As covert communication became increasingly prevalent in ATI-IoT applications, it could be further extended to space–air–ground-integrated networks. Therefore, the study of covert communication in ATI-IoT systems is still an area to be further studied.

3.4. Secure Transmission in Cooperative Relaying Systems

The cooperative relay security transmission scheme adopts one or more trusted nodes to relay secure information in the network [143,144,145]. Figure 10 displays its system model. Owing to channel fading and physical obstructions, direct communication between the HAPS and IoT devices is not feasible. Relay nodes (e.g., drones and LAPSs) employ the decode-and-forward relay protocol to assist in the secure transmission between the HAPS and IoT devices [146]. In addition, the eavesdropper attempts to obtain security information about the network. The cooperative relay system uses a time-division mechanism to transmit secure information, dividing each time slot into two phases corresponding to the broadcasting and relaying stages [147]. In the first transmission phase, the HAPS sends secure information to the relay nodes, which is also susceptible to interception by eavesdroppers. In the second transmission phase, relay nodes first decode the legitimate information then re-encode and forward it to the IoT devices [148]. Employing a maximum ratio combination mechanism, the eavesdroppers can receive secure information during both time slots.

Based on Shannon’s theory, information can be securely transmitted only if the quality of the eavesdropping link is inferior to that of the legitimate link. Therefore, using cooperative jamming techniques can further enhance the security performance of ATI-IoT systems. The cooperative jamming security transmission scheme employs one or more cooperative jamming nodes that transmit interference signals to disrupt eavesdroppers from intercepting secure information, thus guaranteeing the security of the data [149,150,151]. In other words, to implement cooperative jamming, we need to add several cooperative jamming nodes (such as drones and IoT devices) to the system model shown in Figure 10. These jamming nodes generate artificial noise to interfere with the eavesdroppers’ reception of information, thereby protecting the secure information.

3.5. Comparative Analysis

To sum up,

Table 4. Comparative analysis of emerging physical layer security techniques for ATI-IoT.

Technique	Description	Advantages	Disadvantages
IRS-based approach	This technique leverages IRSs to enhance physical layer security by dynamically adjusting the signal propagation environment.	The IRS enhances security by mitigating interference and eavesdropping risks and improves both the channel capacity and reliability of wireless communication.	IRS deployment requires advanced hardware and significant infrastructure. In addition, the signal processing and optimization involved in IRS configurations are complex.
D2D communication	This approach integrates D2D communication within ATI-IoT to enhance security by allowing direct communication between devices without involving base stations.	D2D communication reduces transmission distance, decreasing interception opportunities, and typically operates with lower power consumption.	There is a risk of unauthorized communication between devices, which poses a security threat. In addition, this approach will suffer from interference caused by other D2D devices.
Covert communication	This method focuses on transmitting information in a manner that makes it difficult for unauthorized parties to detect the communication, thereby ensuring its covert nature in the ATI-IoT.	This technique ensures high-level security by hiding the communication from potential eavesdroppers, and enhances privacy for users and devices.	Covert communications often lead to a reduction in transmission efficiency due to the need to disguise the signal. In addition, there is a potential for detection if covert methods are not implemented effectively.
Cooperative relaying systems	This method employs cooperative relaying to forward signals from a source to a receiver, with relays performing additional processing to enhance transmission security.	The presence of multiple relay nodes helps to obscure eavesdropping and improve security. Meanwhile, cooperative relaying increases communication coverage and reliability.	The use of relays introduces additional system complexity and can increase power consumption. In addition, the coordination and synchronization required among multiple relays can pose operational challenges.

provides a comparative analysis of four emerging physical layer security techniques for ATI-IoT. Specifically, the IRS-based approach improves security by mitigating interference and enhancing channel reliability but requires advanced hardware. D2D communication enhances security by reducing transmission distance but faces risks of interference. Covert communications ensure privacy and security but can result in reduced transmission efficiency. Finally, cooperative relaying systems improve communication reliability and security, though they introduce complexity and power consumption concerns. Overall, each technique has its advantages and disadvantages, which should be considered depending on the specific requirements of the ATI-IoT applications. In addition,

Table 5. Summary of security techniques, performance metrics, and beneficial scenarios.

Technique	Performance Metrics	Beneficial Scenarios
IRS-based approach	Secrecy capacity, signal-to-noise ratio, and secure throughput.	In large-scale antenna systems and dynamic channel conditions, IRS can be used to achieve confidential information transmission.
D2D communication	Secrecy outage probability, secure throughput, and bit error rate.	Scenarios in which D2D communication can enhance direct interaction while ensuring secure data transmission.
Covert communication	Secrecy capacity, bit error rate, and key generation rate.	Covert communication can be applied in scenarios with low detectability, such as military or privacy-sensitive applications.
Cooperative relaying systems	Secure throughput, secrecy capacity, and transmission delay.	Cooperative relaying systems can be deployed in scenarios where relay nodes are needed to enhance communication reliability and security, especially in areas with poor channel conditions.

briefly summarizes the emerging physical layer security techniques, their typical performance metrics, and the application advantages in specific scenarios.

4. Application Scenarios of Physical Layer Security Techniques in ATI-IoT

The emerging physical layer security techniques discussed above are highly applicable to a variety of real-world scenarios. In this section, we elaborate on how physical layer security techniques can be applied to practical applications, including emergency communication, vehicular ad hoc networks, intelligent agriculture, and maritime communication networks.

4.1. Emergency Communication

Emergency communication systems are designed for control centers to provide urgent command during sudden incidents, and are used for major disaster relief, security at large events, etc. Due to the unpredictability of the time and location of incidents, as well as the uncertainty of the communication environment, these systems must possess excellent mobility, interference resistance, and the ability to function in harsh conditions [152,153,154]. Aerial–terrestrial integrated communication, an important modern communication technique, is widely used globally [155]. It offers high mobility, flexibility, and robust operability. Compared to other communication methods, aerial–terrestrial integrated communication provides broader coverage and faster communication rates. In scenarios of significant natural disasters or severe damage to ground lines, this form of communication plays a unique and irreplaceable role [156]. Expanding on the above discussion, the ATI-IoT establishes a link between ground IoT devices and aerial platforms, significantly enhancing communication capabilities and facilitating seamless connections, based on which a reliable emergency communication framework can be constructed.

As depicted in Figure 11, ATI-IoT enables rapid deployment of communication links in disaster areas where connections are disrupted or of poor quality. This integration employs the HAPSs to achieve wide-ranging emergency coverage. Additionally, due to drones’ capability to operate unrestricted by terrain and their flexible deployment, they can quickly establish communication links with ground terminals in disaster areas [157]. By coordinating drones with ground emergency communication vehicles and fully utilizing their transmission resources, communication quality can be ensured. Furthermore, emergency communication systems should also possess secure communication features to guarantee safe and reliable communications in emergency scenarios related to disaster relief. This is because the ATI-IoT operates within complex electromagnetic environments, where IoT devices are connected through air–ground links [158]. Meanwhile, the aerial and terrestrial networks are highly susceptible to malicious eavesdropping by unauthorized users, thus requiring consideration of security performance during deployment.

4.2. Vehicular Ad-Hoc Networks

As a typical application scenario of IoT, intelligent traffic management, traffic information service, and smart vehicle control can be realized by using vehicular ad hoc networks (VANETs). In VANETs, we can use vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) transmissions to satisfy the basic communication needs [159]. However, in special scenarios such as traffic hotspot areas, the demand for various types of information by vehicle users surges dramatically, and relying solely on roadside units cannot meet these huge communication demands. Additionally, during natural disasters, when communication infrastructure is damaged, vehicular communication requirements cannot be met. In the face of these challenges, VANETs based on ATI-IoT are expected to be a key solution [64,160,161].

As illustrated in Figure 12, VANETs based on ATI-IoT introduce multi-dimensional resources from the air into existing intelligent transportation systems, thereby further ensuring vehicle-to-everything (V2X) communications. Specifically, drones can employ a “store-carry-forward” mechanism to enhance communication efficiency and reliability [162]. First, the drone can collect data from one location and store it temporarily. Then, the drone carries the data as it moves, and eventually forwards the data to another location or node within VANETs. This approach is particularly useful in scenarios where direct communication between ground vehicles or between vehicles and fixed infrastructure is hindered by distance, obstacles, or network congestion. By leveraging their mobility and flexibility, drones can bridge communication gaps, extend network coverage, and improve data transmission rates of VANETs.

Current research on aerial platform-assisted VANETs primarily focuses on routing protocols, trajectory optimization, content caching, and computational offloading. However, with the increasing openness of cellular vehicle networking architecture, it is easier for malicious eavesdropping nodes to infiltrate the network and steal confidential information [163]. The leakage of sensitive information transmitted within VANETs will lead to traffic accidents and substantial financial losses. Therefore, it is very important to study the communication security in VANETs based on ATI-IoT.

4.3. Intelligent Agriculture

The implementation of ATI-IoT systems supports the realization of intelligent agriculture [164]. When there is no ground communication infrastructure or when transmitting signals between two agricultural devices under poor channel conditions, drones with rapid deployment capabilities can provide network access. Serving as relay nodes, drones can connect isolated agricultural devices with the aid of macro base stations, effectively overcoming transmission barriers caused by long distances or blocked communication links.

As shown in Figure 13, in agricultural IoT, drones can be effectively utilized as cluster heads to enhance network connectivity through clustering routing protocols [165]. These protocols organize the agricultural network into clusters, with each drone acting as a central node. This arrangement allows the drones to manage communications within their respective clusters, facilitating efficient data transmission between agricultural devices. Nevertheless, when transmitting confidential information, agricultural networks assisted by aerial platforms face severe security challenges [166]. To enhance the security of agricultural networks, the aerial platform can select the appropriate relaying location for covert information transmission, reducing the risk of being illegally monitored.

4.4. Expansion to Maritime Communication Networks

The integration of aerial and terrestrial networks into maritime systems utilizes the unique strengths of both cellular networks and spatial information networks, establishing a robust foundation for maritime information transmissions. The deployment of aerial platforms offers distinct advantages over satellite networks for maritime communication by providing higher reliability, faster data transmission rates, and lower operational costs [167]. As illustrated in Figure 14, aerial platforms facilitate seamless connectivity by leveraging existing cellular networks, shore-based stations, and vessels to form a heterogeneous network. This integration enables the delivery of diverse and high-demand user services directly on the ocean edge network, reducing reliance on the core network and significantly enhancing service efficiency. However, given that air–sea–ground networks pose security risks, it is essential to implement physical layer security techniques to greatly enhance the security [168].

5. Challenges and Future Trends for ATI-IoT

Integrating physical layer security techniques and aerial platforms into IoT can improve coverage, channel capacity, reliability, security, and deployment flexibility. As research on this area has only recently started, there are several open issues and interesting research directions worth exploring.

5.1. Technical, Security and Sustainability Challenges

Currently, ATI-IoT faces technical, security and sustainability challenges as follows.

• Technical Challenges: The main technical challenges in ATI-IoT systems are related to signal coverage, data processing capabilities, and multi-platform collaboration. First, in ATI-IoT, we need to ensure and sufficient computational ability to support real-time data analysis and decision-making. Third, the challenge of multi-platform collaboration concerns the compatibility between different types of devices, such as HAPSs, LAPSs, drones, and ground terminal systems, which must be seamlessly integrated to achieve efficient data collection and resource sharing. Addressing these technical challenges requires continuous technological innovation and standardization efforts to facilitate better integration among devices and platforms.
• Security Challenges: In terms of security, ATI-IoT faces key challenges including data security, cybersecurity, and physical security of devices. First, to protect data transmitted between devices and platforms from unauthorized interception, further research on encryption algorithms and security protocols is needed. Second, to prevent external attackers from damaging systems through ATI-IoT, protection is needed at the software level as well as the continuous monitoring and management of network traffic. Third, the physical security of the devices is crucial. In this situation, enhancing the physical robustness and tamper-resistance of ATI-IoT access terminals is essential.
• Sustainability Challenges: Sustainability poses another significant challenge for the implement of ATI-IoT, including environmental and economic impacts. On the one hand, IoT devices and operations may disturb natural ecosystems, such as frequent drone usage potentially affecting wildlife habits, and the manufacturing and disposal of devices may create a certain environmental burden. On the other hand, while ATI-IoT technologies can improve efficiency, the high initial investment and maintenance costs may limit its use in low-income areas.

5.2. Possible Solutions

In the future, for ATI-IoT to enable physical layer security, several aspects need further study. The first is how to integrate the mentioned emerging techniques with artificial intelligence (AI). For example, we can automatically adjust the phase shift of IRS through AI to achieve dynamic optimization of the signal path. This scheme can not only intelligently adjust the direction and intensity of signal propagation according to environmental changes, but also predict potential threats and eavesdropping behavior based on machine learning models, so as to adjust the communication strategy in real time to ensure security during the transmission process. Moreover, D2D communication, covert communication, and cooperative transmission techniques will also be further developed by incorporating AI. For example, in D2D communications, AI can be used to analyze user behavior and communication patterns, intelligently deciding when and where to enable D2D communications to optimize network load and enhance spectrum efficiency. Similarly, in terms of covert communication, the use of AI models can analyze the communication environment in real time and automatically adjust the transmission strategy. Furthermore, cooperative transmission can strengthen the information sharing and processing ability between nodes through AI, such as using deep learning to optimize the encoding and decoding process of signals and improve the overall anti-interference and security performance of multi-node systems. The research combined with these emerging techniques will promote the development of ATI-IoT in a more intelligent, secure, and adaptive direction, effectively coping with complex and changeable security threats, and opening a new path for technological innovation in the B5G era.

6. Conclusions

ATI-IoT significantly enhances the connectivity of B5G networks by enabling seamless data exchange between aerial and terrestrial devices. However, ATI-IoT faces security issues because the signals it transmits can be accessed by anyone within the broadcast range. Facing this challenge, designing efficient secure transmission schemes for ATI-IoT that exploit the propagation properties of radio channels at the physical layer has recently attracted wide research interest. Therefore, this review presents the role of physical layer security in ATI-IoT. First, as a foundation, we outline Shannon’s pioneering work on information-theoretic security and introduce key generation and agreement mechanisms. Next, we describe the aerial–terrestrial integrated architecture, and endogenous safety and security in ATI-IoT. Then, we review emerging techniques (e.g., IRS, D2D communication, covert communication, and cooperative transmission) that enable physical layer security in ATI-IoT, while also presenting practical applications. Finally, we discuss several important open issues related to efficiently integrating and exploiting physical layer security techniques for ATI-IoT in the B5G era, aiming to attract high research interest in this important domain.