Review of Physical Layer Security in Integrated Satellite–Terrestrial Networks

Abstract

With the success and commercialization of 5G, 3GPP has started working toward the sixth generation of communication systems. While 5G explored the concept of non-terrestrial networks like satellites and unmanned aerial vehicles working alongside terrestrial networks, 6G is expected to take this integration a step further, aiming to achieve a more coherent network where satellites and terrestrial infrastructure work together seamlessly. However, the complexity and uniqueness of such networks create numerous attack surfaces that make them vulnerable to cyberattacks. The solution to such cyberattacks can be addressed by encryption and other upper-layer authentication methods. However, with the move to higher-frequency bands, such encryption techniques are difficult to scale for low-latency networks. In addition, the recent progress in quantum computing will make networks more vulnerable. To address such challenges, physical layer security (PLS) is proposed as a secure and quantum-resistant way to implement security by taking advantage of the physics of the channel and transceiver. This article reviews the latest trends and progress in PLS in integrated satellite–terrestrial networks (ISTNs) from a signal processing perspective. This work provides a comprehensive survey of the state-of-the-art research conducted, challenges, and future directions in the PLS of ISTNs.

1. Introduction

In the last few years, there has been exponential growth in mobile devices, with the commercialization of 5G significantly impacting traditional terrestrial and non-terrestrial networks [1,2,3]. The implementation of 5G has led to significantly higher data transfer rates compared to previous generations like 4G and has allowed terrestrial networks to handle more data traffic, enabling faster downloads, uploads, and streaming for users. The technology can now handle a larger number of users. In the development of 5G, 3GPP has played a fundamental role by providing (a) standardization, such as technical specifications for mobile networks and standardized protocols; (b) technical development, such as NR technology, network slicing, air interface, and core network functions; and (c) continuous improvement by releasing new versions with additional features and improvements.

As a dynamic collaboration among telecommunication standards organizations [4,5], 3GPP’s core mission revolves around the continual evolution of mobile communication protocols. The expansive scope of 3GPP’s work includes the foundations of transformative technologies such as LTE and 5G [6]. From defining the intricacies of the air interface and RAN to shaping the architecture of the core network and services, 3GPP exerts its influence across the entire spectrum of mobile communications. Today, 3GPP standards serve as the cornerstone for cellular providers and equipment manufacturers globally. As the preeminent cellular standard organization, 3GPP occupies a pivotal role in charting the course of mobile communication’s future [7]. The role of 3GPP is essential in the context of PLS in ISTNs:

• Detailed channel models that accurately represent real-world propagation conditions are developed by 3GPP. These models are crucial for designing effective PLS algorithms. It defines modulation schemes, coding techniques, and other physical layer parameters that directly influence PLS in ISTNs. In addition, 3GPP addresses interference mitigation techniques, which are essential for PLS, as interference can degrade the performance of secure communication systems.
• Technical standards for mobile communication systems are set by 3GPP. These standards directly influence the design and implementation of physical layer technologies. By understanding 3GPP’s role in standardization, we can better appreciate the constraints and opportunities for PLS development within the context of ISTNs.
• Furthermore, 3GPP is at the forefront of technological advancements in mobile communications. Analyzing their role in technical development provides insights into emerging trends, challenges, and potential solutions that directly impact PLS research and development.

2. Background

Congestion at lower frequencies will require satellites to be designed at mm-wave bands. Such frequencies offer vast bandwidth, which is crucial for supporting high-data-rate services in satellite communications [8,9]. Understanding propagation characteristics in the mm-wave band is essential for efficient satellite link design and beamforming. MIMO technology can significantly improve spectral efficiency and capacity in satellite systems, enabling higher data rates and supporting more users [10]. MIMO techniques can be used for advanced beamforming, allowing for precise antenna pattern control and improved link quality. They also provide diversity gains, which can mitigate the effects of fading and interference, especially in satellite channels [11].

In Release 15, 3GPP laid the foundation for 5G networks, with the ITU defining some of the KPIs: (a) a more than 10 Gbps peak data rate for eMBB, (b) a connection density of 1 million/${\mathrm{km}}^2$ for MMTC, and (c) URLLC with a latency of less than 1 ms [12]. Extended LTE support was provided to aerial vehicles to support cellular connectivity [13,14]. The release also laid the foundation for URLLC to satisfy the technical requirements of new applications, such as virtual reality, industrial automation, and intelligent transportation [15].

Release 16 supported industrial IoT and mission-critical communication [16,17]. The surge in the popularity of IoT led to a massive increase in the number of IoT devices, which could extend beyond 75 billion by 2025. Such expansion led to enhanced URLLC in Release 16 [18]. Some of the notable enhancements compared to Release 15 are MIMO, improved beamforming, dynamic spectrum sharing, dual connectivity, carrier aggregation, and user equipment power saving [19]. The standardization process also encompassed V2X communication through sidelink communication, enabling direct communication between two user devices without the need for the involvement of a base station [20].

In Release 17, 3GPP provided robust support for NTNs, encompassing satellites, high-altitude platforms, and unmanned aerial vehicles. With NTN capabilities, 3GPP protocols for broadband (NR protocol) and massive IoT (NB-IoT and eMTC) can seamlessly extend to satellite communications. The release laid a strong foundation for establishing communication between satellites, smart phones, and other user equipment. The integration of TNs with NTNs will provide faster and more reliable eMBB to smartphones and IoT [21]. The added NTN capabilities will address the surging need for connectivity in remote and underserved regions where conventional terrestrial networks may be absent, such as deserts, oceans, and mountainous regions [22]. The satellite system consists of GEO, MEO, and LEO based at different altitudes.

Release 18, which supports multi-RAT systems, was introduced to provide satellite and terrestrial RANs with their own dedicated radio resource management [23]. A 5G network slice instance is composed of a set of virtual network function (VNF) instances to form the end-to-end virtual network for the slice to operate independently [24]. In a network slice instance, both terrestrial and non-terrestrial RANs can be created by the network operator [25]. Satellites can be configured to operate in either the transparent mode or the regenerative mode, and the choice depends on the specific requirements of the communication system. In the transparent mode, they act as a simple relay, amplifying and forwarding the received signal without modifying or processing the data. In the regenerative mode, the satellite can demodulate the received signal, extract the data, and then remodulate and transmit it [6]. It will also allow for functionalities like error correction, thus improving signal quality. Three use cases were defined in 18: (a) channel state feedback information, (b) beam management, and (c) positioning [26].

In Release 19, 3GPP continued with the advancement of 5G and introduced improvements in spectral efficiency and coverage across the sub-7 GHz and mm-wave spectrum [27]. It aimed to boost capacity in massive MIMO radio by enabling the sharing of the same time and frequency resources. A cost-effective transceiver would enable a distributed MIMO system. Additionally, 5G-Advanced would, for the first time, have one standard for TNs and satellites. With satellites, it focused on increasing downlink coverage. AI has now become imperative to improving the functionality and automation of communication networks. In 19, for both NR-NTNs and IoT-NTNs, it provided support for regenerative load and increased uplink capacity. Ambient IoT and extended boundless XR were considered [28]. With Apple and Samsung launching SOS services via iPhone and smartphones, respectively, the importance of satellites will keep growing in the near future [29,30].

To identify its work on developing 6G specifications, 3GPP unveiled a new logo [31]. The evolution to 6G presents many challenges in terms of reliability, availability, and responsiveness that are difficult to address alone by terrestrial cellular networks, especially in the case of natural disasters and terrorist attacks [32]. Satellites can provide (a) wider coverage, (b) backhaul links, and (c) edge computing. Moreover, 6G is expected to play a more holistic and human-centric role by providing services like human bond communication and tactile and holographic communication. Additionally, it will improve experiences over 5G, such as conventional mobile communication, indoor positioning, and mMTC for vertical industries [33].

As NTNs and TNs will be integrated to form a seamless, coherent network in 6G as shown in Figure 1, security challenges will be very complex in nature. The complexity of such a network opens many attack surfaces, for which a multi-layered approach would be preferable; i.e., the authentication scheme should be designed across the layer. Due to its unique characteristics, an NTN faces challenges due to its extended coverage, propagation environment, dynamic topology, and resource constraints [34,35]. The vast coverage area of satellite networks increases the attack surface, making it difficult to monitor and protect all potential vulnerabilities. Satellite channels are susceptible to jamming, spoofing, and eavesdropping due to the open nature of the space environment. The constantly changing satellite constellation and user locations pose challenges for secure key management and authentication. Satellite platforms have limited power, computational resources, and storage capacity, which impact the feasibility of complex security algorithms. Strong authentication mechanisms are essential to verify the identity of users, devices, and network entities. This can be achieved through multi-factor authentication, digital certificates, and secure key management protocols. The unique identification and authentication of satellites are crucial for preventing spoofing and unauthorized access. Secure user authentication methods are required to protect user data and prevent unauthorized access to network resources.

PLS can play a significant role in such a scenario [36]. PLS leverages the physical characteristics of a communication channel to ensure secure data transmission. Unlike traditional cryptographic methods that operate at higher layers of the network stack, PLS operates directly at the physical layer, exploiting the inherent properties of the communication medium. PLS exploits channel impairments such as noise, fading, and interference as a means to secure communication. Such properties make PLS an effective mechanism to address vulnerabilities introduced by the advent of quantum computing.

While quantum computing poses a significant threat to traditional cryptographic algorithms based on computationally hard problems, it has a limited impact on PLS. PLS techniques primarily rely on the inherent characteristics of the physical communication channel rather than complex mathematical computations. Therefore, quantum algorithms like Shor’s algorithm, which excels at factoring large numbers and solving discrete logarithms, are not directly applicable to breaking PLS schemes. This makes PLS a promising approach for ensuring secure communication in a post-quantum era, complementing traditional cryptographic methods.

Secure information is embedded into the transmitted signal in a way that is imperceptible to an eavesdropper but can be recovered by the legitimate receiver. Physical layer coding can be employed to protect the embedded information from being extracted by unauthorized parties. PLS techniques need to be adapted to the unique characteristics of satellite channels, such as high path loss, Doppler shift, and shadowing. PLS could be combined with upper-layer security mechanisms to provide a comprehensive security solution. The security functions of PLS can be summarized as follows:

• Confidentiality: PLS safeguards information from unauthorized access through various methods. It can establish a secure key exchange by leveraging channel characteristics as a shared secret. Moreover, sensitive data can be embedded subtly within the transmitted signal, rendering it imperceptible to eavesdroppers. Additionally, PLS complements traditional encryption by providing an extra layer of protection rooted in the physical layer’s properties.
• Authentication: PLS contributes to verifying the authenticity of communication participants. By utilizing unique channel characteristics, physical layer fingerprinting can authenticate the transmitter. This technology also aids in confirming the identity of devices through their distinct physical layer parameters. Furthermore, PLS enhances message integrity by detecting and mitigating tampering attempts at the physical layer.
• Integrity: To preserve data integrity, PLS incorporates error correction codes to rectify transmission errors. It also functions as an anomaly detection tool, identifying irregularities that may signal malicious activities. Additionally, PLS contributes to data origin authentication by incorporating physical layer information into the verification process.

By harnessing the physical properties of the communication channel, PLS offers an approach that complements traditional cryptographic methods, enhancing overall security. PLS, enabled by information-theoretic models, entails lower computational complexity compared to cryptographic techniques. It lowers the burden of additional code space required and saves the energy required to perform complex mathematical computations for cryptographic techniques [37]. PLS utilizes the randomness of the wireless channel, such as fading and noise, to secure information [38]. Leveraging the inherent characteristics of the physical layer can significantly reduce the energy overhead for security, crucial for resource-constrained systems where additional computational burdens are impractical. By exploiting the physical properties of the channel, PLS can lead to a significant reduction in the required energy overhead for security, which is critical for resource-constrained devices where additional computational burden may be needed. It becomes an efficient way to ensure security in novel cases of 6G, such as mMTC, URLLC, and cyber-physical systems. Newly emerging services like IoT, tactile internet, remote surgery, and mMTC are required to have lower latency and are power-limited applications. Encryption methods would be less suitable to scale for such applications [39]. PLS can complement upper-layer authentication methods to provide robust security to ISTNs [40].

In summary, PLS can be useful in the following scenarios:

• Latency requirement: For low-latency requirements, delays incurred by communication and computation overhead should be minimized.
• Massive IoT deployment will need the low complexity that can be provided by PLS, or a hybrid approach can be taken, where a cross-layer low-weight mechanism can ensure security.
• Post-quantum era: With the advent of quantum computing, encryption algorithms would be more vulnerable to cyberattacks.
• Edge intelligence: As satellites can serve as edge computing devices, AI edge intelligence based on PLS can play a larger role.
• Physical layer key generation: This utilizes the randomness inherent in wireless channels to establish a shared secret key between communicating devices, e.g., Alice and Bob.
• RF-fingerprinting-based authentication: This technique creates a unique fingerprint based on the RF characteristics of the device’s communication signal. Several factors, such as signal strength, fading patterns, and noise, can contribute to this fingerprint. The fingerprint is then compared to a pre-stored reference fingerprint for authentication.
• Localization-based authentication: This method relies on the physical location of a device to verify its identity. The location is determined by using techniques like GPS, WiFi fingerprinting, or cellular network positioning. If the reported location matches that expected for a legitimate device, authentication is successful. It should be noted that 6G aims for centimeter-level accuracy in positioning.

PLS techniques can be useful in various scenarios in 6G and can also be applied to complement upper-layer authentication methods. PLS has some inherent challenges in ISTNs that arise due to (a) correlated channels due to the negligible distance between legitimate users and an eavesdropper and that from the satellite to legitimate users, (b) co-channel interference due to the spectrum being shared by NTNs and TNs, (c) the broadcast nature of satellites, which creates a multi-user multi-eavesdropper scenario, (d) reliability concerns caused by channel estimation errors and hardware impairments [41,42]. The correlation between the main and wiretap channels also presents challenges for CSI-based schemes.

This paper is organized as follows. Section 2 discusses the various performance parameters. Section 3 focuses on some of the existing surveys of ISTNs. In Section 4, we review the PLS techniques from the signal processing viewpoint by dividing them into resource allocation, beamforming, cooperative communication, and physical characteristics. In Section 5, we present future directions. Section 6 concludes the paper.

3. Existing Surveys

Recently, many excellent survey, review, and tutorial papers have been published regarding PLS. In [40], the authors showed how PLS can be utilized to achieve security goals such as confidentiality, integrity, authentication, and availability. It provides a background on cryptography and network security. Further, it explores hardware-based and statistical methods for user authentication, with a particular focus on techniques that leverage localization-based authentication and physical unclonable functions (PUFs).

In [41], various challenges in the PLS of ISTNs is discussed in detail. The study shows a solution to the correlated-channel problem by using a dual-beam, dual-frequency transmission scheme and a near-ground, co-frequency, co-time, cooperative interference relay network. For the co-channel interference problem, the article provides solutions such as adaptive and cooperative beamforming, artificial noise, and multi-satellite scheduling. For the multi-user multi-eavesdropper scenario, they suggest relay and user selection.

In [43], a survey of space–air–ground–sea integrated networks integrating satellite, aerial, terrestrial, and maritime networks was conducted. The article discusses the characteristics and potential challenges. A detailed discussion on security threats, attacks, and countermeasures is provided. Possible solutions, such as anti-jamming, secure routing, secured key management, handover schemes, and intrusion detection, are described. Future challenges, e.g., cross-layer attacks and countermeasures in such networks, are presented.

In [44], the article examines current security challenges in satellite–terrestrial communication systems, proposes potential solutions, and suggests research directions to address existing security gaps. It provides an overview of the security landscape in ISTNs, highlighting key security challenges from three perspectives: satellite-to-satellite communications, satellite-to-ground station communications, and satellite-to-ground user equipment (UE) communications. It discusses potential solutions to these identified security challenges and identifies existing research gaps. Additionally, it offers insights into new technologies that could enhance security in these integrated networks.

In [45], the study presents a comprehensive survey of satellite communications with a focus on PLS, providing background information and discussing challenges in the emerging integrated network architecture. It reviews popular satellite channel models and secrecy performance metrics, and categorizes state-of-the-art PLS research across different architectures: land mobile satellite communication networks, hybrid satellite–terrestrial relay networks, and satellite–terrestrial integrated networks. The article also identifies open research problems for future exploration.

In [46], the authors provide a comprehensive survey of the convergence of satellite and terrestrial networks, identifying motivations and requirements, summarizing related architectures, classifying research taxonomy, and presenting performance evaluations. It also reviews the state of the art in standardization, projects, and key application areas of satellite–terrestrial networks, concluding with open issues and future directions.

In [47], the authors discuss the potential of ISTNs to extend network coverage and reduce reliance on terrestrial infrastructure, which is key for 6G communication. However, challenges remain in areas such as architecture, air interface protocols, mobility management, and experimental validation. The article outlines four stages of ISTN architecture development and explores key technical enhancements. It also details an experiment using LEO satellites to link two 5G networks, demonstrating low latency and high data rates, showing the ISTN’s potential for industrial applications. Future trends include programmable satellites and AI-driven networking.

In this paper, the authors discuss the challenges of spectrum utilization in the transition from 5G to 6G and highlight integrated sensing and communication (ISAC) as a key 6G technology [48]. ISAC aims to improve energy and spectrum efficiency by combining radar and communication signals for environmental awareness and scene interconnection. The integration of ISAC with ISTNs has become a significant research direction for achieving seamless 6G coverage. The paper reviews key ISAC technologies, describes the development of ISTNs, and explores key topics and future applications for 6G ISAC systems in ISTNs.

However, to our best knowledge, a signal processing perspective for understanding the problem and its solution has not yet been presented in any review paper. Our goal is to gain a better understanding of how PLS might improve the security of future communication networks, such as 6G, by tackling vulnerabilities that are brought about by merging terrestrial and satellite systems. This is an important subject generally, but it takes on further significance when considering 6G communication systems, as new attack surfaces are introduced by the seamless integration of NTNs and TNs. This research fulfills a particular need in PLS from a signal processing standpoint, and the intricacy of ISTNs causes several weaknesses. For this reason, this subject is novel and will play an important role in the development of future forms of secure communication. This research takes a more targeted approach than previous works by examining, from a signal processing standpoint, the most recent state-of-the-art PLS methods developed for ISTNs. Benefiting from the classification of PLS methods into four main areas—physical properties, cooperative communication, resource allocation, and beamforming—this study provides value. These categories provide a holistic view of how to optimize the physical features of satellite–terrestrial communication for security.

4. Review

Our work focuses on the PLS of ISTNs from a signal processing perspective. Our contributions can be summarized as follows:

• We classify PLS methods into four main categories: resource allocation, beamforming, cooperative communication, and physical characteristics. For each category, we provide a comprehensive review of state-of-the-art research conducted in recent years.
• In the area of resource allocation, we discuss various strategies designed to optimize the use of network resources while ensuring secure communication. These strategies often involve complex optimization techniques to efficiently manage the allocation of bandwidth, power, and other resources, aiming to maximize secrecy capacity and enhance overall network security.
• Beamforming is another crucial aspect of our review. We examine advanced beamforming techniques that focus signals in specific directions to improve signal quality and reduce the risk of eavesdropping. By leveraging optimization algorithms, these techniques enhance the secrecy capacity and energy efficiency of the network.
• Cooperative communication involves multiple network nodes working together to achieve secure communication. We highlight recent advancements in this field, including methods where nodes cooperate to create a more robust and secure communication environment. These methods often use optimization techniques to balance cooperation benefits with energy efficiency and security requirements.
• Finally, we explore physical characteristics that inherently enhance the security of ISTNs. This includes the exploitation of unique propagation properties of wireless signals, such as path loss, fading, and channel properties, to improve security. Techniques in this category often involve optimizing these physical properties to achieve better secrecy capacity and normalized secrecy capacity.

By providing a detailed review of the latest advancements in resource allocation, beamforming, cooperative communication, and physical characteristics, we aim to offer valuable insights into how to enhance the PLS of ISTNs. First, we take a look at various performance metrics that are commonly used in providing PLS for a network.

4.1. Performance Parameters

In this section, we discuss key performance indicators that are required to assess the security provided to a communication system using a PLS scheme. Such parameters include the following:

• Secrecy Capacity ($C_s$): This metric measures the maximum rate at which confidential information can be transmitted reliably over a communication channel while keeping it secret from unauthorized users. It takes into account both the legitimate channel and the eavesdropper’s channel. It is defined as [49]

  $$C_s={log}_2(1+{\mathrm{SNR}}_{\mathrm{Bob}})-{log}_2(1+{\mathrm{SNR}}_{\mathrm{Eve}})$$

  (1)

  where ${\mathrm{SNR}}_{\mathrm{Bob}}$ and ${\mathrm{SNR}}_{\mathrm{Eve}}$ denote the SNRs of Bob and Eve, respectively.
• Secrecy Outage Probability ($P_s$): The secrecy outage probability represents the probability that the secrecy capacity falls below a certain threshold. It indicates the likelihood of failing to maintain secure communication due to channel conditions. This metric is expressed as [50]

  $$P_s=P(C_s<R_0)$$

  (2)

  where $R_0$ is the threshold or target secrecy rate.
• Secrecy Diversity Gain ($G_s$): Secrecy diversity gain, also known as secrecy diversity order, measures the improvement in secrecy capacity achieved by employing multiple antennas or multiple transmit/receive paths. It quantifies the robustness of the system to eavesdropping attacks and is defined as [51]

  $$G_s=\underset{{\mathrm{SNR}}_{\mathrm{Bob}}\to \infty }{lim}{\displaystyle \frac{-log{P}_s}{log{\mathrm{SNR}}_{\mathrm{Bob}}}}$$

  (3)
• Secrecy Rate ($R_s$): The secrecy rate represents the achievable rate of secure communication over a given channel. It considers the trade-off between the transmission rate and the level of secrecy [52]. The achievable secrecy rate is defined as

  $$R_s=max\{R_m-R_e,0\}$$

  (4)

  where $R_m$ denotes the data rate of the legitimate channel, and $R_e$ denotes the data rate of the eavesdropper’s channel.
• Ergodic Secrecy Rate: This metric calculates the average secrecy rate over all possible channel realizations. It provides a more comprehensive view of the system’s security performance under varying channel conditions [53].
• Outage Secrecy Capacity: The outage secrecy capacity represents the maximum achievable secrecy capacity subject to a certain outage probability constraint. It accounts for the randomness in channel conditions and ensures a certain level of security even under adverse conditions [54]. It is defined as

  $$P_{out}\left(C_{out}\left(ϵ\right)\right)=Pr\{C_s<C_{out}\left(ϵ\right)\}=ϵ$$

  (5)

  where $ϵ$ is the tolerable secrecy outage probability.
• Mutual Information Secrecy: MIS quantifies the information-theoretic secrecy provided by the system. It measures the difference between the mutual information of the legitimate receiver and that of the eavesdropper. The metric is defined for a discrete memoryless-channel as follows [55]:

  $$I_s=I_b-I_e$$

  (6)

  where $I_b$ and $I_e$ are the mutual information of Bob’s and Eve’s channels, respectively.
• Secure Energy Efficiency: This metric defines the secret bits transmitted per unit energy consumption and is defined as [56]

  $$E_B={\displaystyle \frac{R_S\Delta T}{\Delta E}}={\displaystyle \frac{R_S}{P}}$$

  (7)

  where P is the total power consumption, and $\Delta E$ is the amount of energy consumed over the duration $\Delta T$.
• Probability of Nonzero Secrecy Capacity ($P_{NZC}$): This metric indicates the probability that the secrecy capacity is nonzero, meaning that secure communication is possible. It reflects the reliability of achieving secrecy in the system. The metric is defined as

  $$P(C_s>0)={\int }_0^{\infty }{\int }_0^{{\mathrm{SNR}}_{\mathrm{Bob}}}p({\mathrm{SNR}}_{\mathrm{Bob}},{\mathrm{SNR}}_{\mathrm{Eve}})d{\mathrm{SNR}}_{\mathrm{Eve}}d{\mathrm{SNR}}_{\mathrm{Bob}}$$

  (8)

  where $p({\mathrm{SNR}}_{\mathrm{Bob}},{\mathrm{SNR}}_{\mathrm{Eve}})$ denotes the joint pdf of the SNRs of Bob and Eve.
• Equivocation Rate: The equivocation rate quantifies the uncertainty an eavesdropper has about the actual message sent by Alice to Bob after it has been transmitted through a physical channel.
• Normalized Secrecy Capacity ($\overline{C}$): This parameter takes into account the relative strength of Eve’s signal compared to Bob’s signal and is defined as follows [57]:

  $$\overline{C}={\displaystyle \frac{{log}_2(1+{\mathrm{SNR}}_{\mathrm{Bob}})-{log}_2(1+{\mathrm{SNR}}_{\mathrm{Eve}})}{{log}_2(1+{\mathrm{SNR}}_{\mathrm{Bob}})}}$$

  (9)
• Physical Layer Authentication Rate: While secrecy focuses on keeping information confidential, authentication ensures the legitimacy of the communicating parties. This metric assesses the effectiveness of PLS techniques in verifying the identity of the intended receiver, taking into account true positives (TP), true negatives (TN), false positives (FP), and false negatives (FN). It is defined as [58]

  $$\mathrm{AR}={\displaystyle \frac{\mathrm{TP}+\gamma \times \mathrm{TN}}{(\mathrm{TP}+\mathrm{FN})+\gamma \times (\mathrm{TN}+\mathrm{FP})}}$$

  (10)

  where $\gamma$ is defined as

  $$\gamma ={\displaystyle \frac{\mathrm{TP}+\mathrm{FN}}{\mathrm{TN}+\mathrm{FP}}}$$

  (11)

4.2. Resource Allocation

Resource allocation can play a crucial role in both enhancing security and reducing power consumption in ISTNs [59]:

• Dynamic Resource Allocation: ISTNs are inherently heterogeneous, with terrestrial links offering high bandwidth and satellite links providing wider coverage. By dynamically allocating resources based on the type of data and security requirements, network operators can prioritize security for sensitive information [60]. For instance, critical data can be routed through terrestrial links with stronger encryption, while less sensitive data can utilize satellite links with lighter security measures.
• Cognitive Radio Techniques: Cognitive radios can intelligently sense the network environment and adapt resource allocation to optimize security. They can identify and exploit weaknesses in potential eavesdropper channels, thereby enhancing the overall security posture of the ISTN [41,61].
• Energy-Efficient Resource Allocation: Resource allocation strategies can be designed to prioritize energy-efficient components of the network. For example, by offloading traffic to terrestrial links when possible and utilizing satellites only when necessary, network operators can reduce the reliance on power-hungry satellite communication [62].
• Resource Sharing and Power Control: The efficient sharing of resources between terrestrial and satellite networks can lead to significant power savings. Additionally, power control techniques can be employed to adjust transmission power based on traffic demands and channel conditions, further reducing energy consumption [63].

While each method offers potential benefits, it also comes with specific drawbacks. The choice of approach to adopt will depend on the specific context and the importance of different factors:

• Dynamic Resource Allocation: Implementing dynamic resource allocation requires sophisticated algorithms and real-time monitoring of network conditions. Frequent resource reallocation might introduce additional latency, especially for time-sensitive applications.
• Cognitive Radio Techniques: Developing and deploying cognitive radio systems can be technically challenging and expensive. Ensuring that cognitive radios do not interfere with other licensed users is a critical consideration.
• Energy-Efficient Resource Allocation: Prioritizing energy efficiency might sometimes lead to lower network performance, especially in areas with limited terrestrial infrastructure. Determining the optimal balance between energy efficiency and performance requires careful consideration of economic factors.
• Resource Sharing and Power Control: Coordinating resource sharing and power control between terrestrial and satellite networks can be complex. Ensuring that shared resources do not interfere with each other is a critical challenge.

All of the proposed resource allocation techniques have the potential to significantly improve the security, performance, and energy efficiency of ISTNs. However, the optimal choice of techniques will depend on specific network requirements, technological capabilities, and economic constraints. A combination of these techniques, tailored to the unique characteristics of the ISTN, is likely to yield the best results.

In Figure 2, we show a power-adaptive scheme for a LEO satellite. The parameters are considered using Telesat data [64]. As the elevation angle of LEO changes due to its movement, the power adapts based on the channel condition and elevation angle and the required SNR to be maintained. By adapting power to different elevation angles, the LEO satellite can maintain a very low SNR at the ground station, making it difficult for an eavesdropper. As the SNR remains low, the receiver will receive the message with a high bit-error rate. The simulation parameters are shown in

Table 1. Simulation parameters for a LEO satellite for the example shown in Figure 2.

Definition	Value
Transmitter antenna diameter	3.5 m
Receiver antenna diameter	1 m
Satellite altitude	1000 km
Atmospheric height	100 km
Antenna efficiency	0.9
Frequency	28.5 GHz
Bandwidth	2.1 GHz
Receiver noise temperature	500 K

. The gaseous attenuation has been calculated as per ITU recommendations [65]. The atmospheric channel is taken as the standard atmosphere model [66]. The SNR is taken to be 5, 10, and 15 dB in this case; however, even lower SNRs can be considered, but then we have to assign more bits to error-correcting codes to receive messages with high fidelity.

In [67], a support vector machine-based algorithm is presented for the detection of spectrum occupancy, and a convolutional neural network is proposed for spectrum prediction. The proposed system utilizes an intelligent resource management strategy that incorporates spectrum sensing, prediction, and allocation. This approach prioritizes user needs and optimizes spectrum allocation, resulting in lower error rates and improved spectrum efficiency. In [68], a dual-beam dual-frequency scheme is used to extend the difference between the main and wiretap channels. By limiting the satellite power and the user’s QoS, the scheme is solved using convex semidefinite programming, which was shown to improve the secrecy capacity even when the eavesdropper is close to the legitimate user. In [69], PLS in satellite communication was studied under conditions of similar legitimate and eavesdropper channels. Using relay selection and power allocation, the secrecy outage probability is optimized. The relay selection is made based on the optimal choice of relays that the satellite has access to. Artificial noise is sent toward the eavesdropper to reduce the quality of the signal.

In [70], a power-efficient beamforming scheme was studied for a cognitive satellite–terrestrial network. An optimal beamforming scheme was proposed to minimize the transmit power of the terrestrial base station while maintaining the secrecy rate of the satellite user and the communication rate of the terrestrial user. In [71], stochastic geometry was used to study the secrecy performance of a satellite–terrestrial network with random locations for one legitimate user and one eavesdropper. A closed-form expression of secrecy outage probability was derived, and maximum ratio combining achieved better results. In [72], the authors focused on designing PLS for next-generation LEO satellite links. The scenario involves an eavesdropper at a certain distance from the legitimate ground station (Bob) communicating with a LEO satellite (Alice). By exploiting the unique characteristics of the THz atmospheric channel, the study derived the optimal power spectral density (PSD) for the transmitted signal to maximize the normalized secrecy channel capacity. Using the higher losses from the atmospheric channel and the optimal PSD, the PLS of the satellite link was improved.

In [73], a method to enhance cybersecurity for LEO satellite links operating at sub-THz frequency bands is presented. The short wavelength at these frequencies allows for highly directional antenna beams with small ground spots, making eavesdropping challenging. Further security is achieved by tuning the satellite transmitter power close to the noise floor and communicating within specific angles of arrival. This approach results in a lower SNR at the ground station receiver, increasing the likelihood of eavesdropper (Eve) outages. To improve Bob’s SNR, an optimization algorithm is derived, which adapts the inter-element spacing of phased array elements to shape the gain pattern and minimize the required transmitter power for a given SNR.

By judiciously distributing resources such as bandwidth, power, and computational capabilities, network operators can enhance security, reduce power consumption, and improve the overall network performance. Dynamic resource allocation strategies, which adapt to changing network conditions and user demands, are particularly effective in ensuring the efficient utilization of resources. Furthermore, cognitive radio techniques can enable intelligent resource management by dynamically sensing the network environment and adapting resource allocation to optimize performance and security.

In addition to enhancing network performance, effective resource allocation can also contribute to reducing the environmental impact of ISTNs. By optimizing power consumption and minimizing the reliance on energy-intensive components, network operators can reduce their carbon footprint and promote sustainable operations. Furthermore, resource sharing and power control mechanisms can help conserve resources and reduce overall energy consumption.

By implementing these strategies, communication networks can achieve a delicate balance between security and power efficiency in ISTNs. This is an active area of research, and future advancements in resource allocation algorithms are expected to play a key role in optimizing ISTN performance.

While ITSNs hold promise for seamless global connectivity, achieving efficient resource allocation in these networks presents several challenges. The heterogeneous nature of satellite and terrestrial networks, with diverse propagation environments and link capacities, complicates the optimization process. Dynamic changes in network topologies and traffic patterns further increase the complexity. Additionally, imperfect channel state information and interference between different network components can degrade performance.

To address these challenges, innovative approaches are necessary. Advanced optimization techniques, such as convex optimization and dynamic programming, can help find optimal solutions. Intelligent resource allocation algorithms that can adapt to dynamic network conditions are essential. Effective interference management techniques, including frequency planning and power control, can mitigate interference and improve performance. Accurate channel estimation is crucial for making informed resource allocation decisions. Collaborative resource allocation between satellite and terrestrial networks can further enhance efficiency and reliability.

By overcoming these challenges and leveraging advanced techniques, ISTNs can deliver high-quality, reliable, and ubiquitous connectivity. This will enable a wide range of applications, from remote sensing and disaster response to global telecommunications and IoT services.

4.3. Beamforming

Beamforming is a signal processing technique used in both terrestrial and satellite communications to focus the transmission or reception of radio waves in a specific direction [74]. Beamforming allows for the highly directional transmission of signals, focusing energy toward specific intended receivers. This can reduce the likelihood of signals being intercepted by unauthorized parties. By concentrating energy in desired directions, beamforming can help mitigate interference from other sources, both within and outside the network [75]. This can make it more difficult for attackers to eavesdrop on or disrupt communications. In conjunction with other physical layer security techniques, beamforming can enhance the overall security of the communication channel. For example, it can be used to create artificial noise or to exploit channel fading to make it difficult for attackers to decode transmitted signals [76]. Beamforming can be used to direct signals toward specific UE and minimize interference with other users or eavesdroppers as shown in Figure 3 [77]. This can be particularly beneficial in densely populated areas where multiple users share the same spectrum. Beamforming on satellites can focus signals on designated areas on Earth, reducing signal leakage and improving security. Additionally, it can enhance uplink transmissions from ground stations to the satellite, making them more difficult to intercept. While PLS leverages the physical properties of the channel itself for security, beamforming can strengthen this security posture in several ways [68]:

• Reduced Signal Leakage: By directing signals toward intended receivers, beamforming minimizes the signal power broadcast in unintended directions. This makes it more challenging for potential eavesdroppers to intercept data, even if they are not actively trying to jam the communication [78].
• Improved SNR: Beamforming concentrates the transmitted signal toward the receiver, leading to a stronger signal and a better SNR. This can make it more difficult for attackers to inject noise or manipulate the signal to compromise data integrity [79].

In Figure 4, we show the beamforming scheme to secure a satellite link. Assuming that the eavesdropper is located slightly far away from Bob, in our case, we have taken the difference in elevation angle to be $0.5^{\circ }$. The main beam points toward Bob, while the gain in the direction of Eve will be taken as the gain away from the boresight with a difference of $0.5^{\circ }$. The simulation parameters are shown in

Table 2. References showing the use of resource allocation to improve the PLS of the communication network.

Reference	Optimization Variable	Constraint
[67]	Spectrum allocation
[68]	Secrecy capacity	Limited power and quality of service
[69]	Secrecy outage probability	Total transmission power of satellite and relay
[70]	Transmit power	Secrecy rate and communication rate
[71]	Secrecy outage probability	Secrecy rate
[72]	Power spectral density
[73]	Signal-to-noise ratio

. The array antenna at the satellite is taken to be a uniform linear array with half-wavelength spacing. The antennas at Eve and Bob are parabolic reflectors with a diameter of 1 m. The transmitted power of the satellite is taken to be 1 W. We observe that as the number of elements increases, the beamwidth of the array antenna decreases, and therefore, the gain in the direction of Eve will be lower. We show the performance in terms of normalized secrecy capacity as the number of elements in the array is varied. It is observed that as the number of elements increases, the secrecy capacity improves, thus improving the PLS of the communication link. Without beamforming, the difference in capacity would be zero under uniform illumination by the satellite.

Static beamforming, where the direction remains fixed, is susceptible to attackers who can learn the beam pattern and position themselves to intercept signals [80]. Dynamic beamforming, where the direction changes frequently, offers better security but requires more complex implementation. Even with dynamic beamforming, there is a possibility of misalignment due to imperfect channel knowledge or rapid user movement. This can create unintended signal leakage and weaken security [81,82].

In [83], convolutional neural network-based beamforming is proposed to handle complex data in a multi-user multi-input single-output (MISO) system. For the optimization of the SNR, a novel unsupervised learning approach is utilized. In [84], system security and spectral efficiency are improved using intelligent reflecting surfaces. Assuming imperfect CSI, the system weighted achievable sum rate was maximized, subject to various constraints, such as the minimum achievable rate constraint of each user and the maximum transmit power of the satellite and base station. In [85], the authors explore secure communication in a cognitive satellite–terrestrial network with a software-defined architecture. They propose beamforming schemes that use terrestrial network interference to enhance satellite network security, leveraging shared millimeter-wave frequencies. The goal is to minimize the total transmit power while meeting quality-of-service and secrecy rate requirements. Two efficient BF schemes are introduced: one for a single eavesdropper using a second-order cone transformation and penalty function, and another for multiple eavesdroppers using a two-layer iterative scheme.

In [86], the authors studied the PLS of a hybrid satellite–terrestrial relay network with multiple eavesdroppers. The secrecy performance using amplify-and-forward and decode-and-forward protocols for both non-colluding and colluding eavesdroppers was analyzed. The study derived secrecy outage probability (SOP) expressions for different fading models, provided user-relay selection criteria, and validated the analysis through simulations, highlighting the impact of various parameters on secrecy performance. In [87], the minimization of power consumed by a multibeam satellite was achieved with a constraint on the secrecy outage probability by proposing joint beamforming and power allocation via alternating optimization to find better solutions iteratively.

In [88], the authors explore 6G wireless networks, emphasizing global high-speed connectivity, especially in under-connected areas. ISTNs are proposed to address challenges like long delays, large-scale device coverage, and real-time services. To tackle these, the authors introduce an optimal multibeam design using machine learning, focusing on user clustering, energy-efficient resource allocation, and distributed computing. Case studies cover disaster relief and satellite–terrestrial IoT networks. Future research directions include digital twins and quantum-inspired optimization to enhance multibeam design in 6G ISTNs.

In [89], the authors address resource allocation in ISTNs, focusing on improving spectrum efficiency and fairness by using beam-hopping technology. The study examines how satellite and terrestrial systems can share the same frequency band, creating dynamic protection zones to avoid interference. Two resource allocation problems are formulated: maximizing the weighted sum of capacity and maximizing the minimum capacity-to-demand ratio. The problems are solved using mixed-integer linear programming, with optimal and suboptimal greedy algorithms proposed. Simulations show that the proposed algorithms enhance spectrum efficiency and ensure fairness and demonstrate that the greedy algorithms perform nearly as well as the optimal ones with lower complexity. In

Table 3. References using beamforming to improve the PLS of the communication network.

Reference	Optimization Variable	Constraints
[83]	Signal-to-noise ratio	Maximum transmit power and secrecy rate
[84]	Achievable sum rate	Intercept probability and transmit power
[85]	Total transmit power	QoS requirement and secrecy rate
[86]	Secrecy outage probability
[87]	Power consumption	Secrecy outage probability
[88]	Optimal multibeam design	Energy-efficient allocation
[89]	Weighted sum of capacity and minimum capacity-to-demand ratio

, we summarize the works that have used beamforming to secure the communication network.

Overall, beamforming, a signal processing technique, has emerged as a valuable tool for enhancing the security of both terrestrial and satellite communication systems. By focusing the transmission or reception of radio waves in a specific direction, beamforming can reduce the likelihood of signals being intercepted by unauthorized parties. This is achieved by concentrating energy toward the intended receivers, minimizing signal leakage and mitigating interference from other sources. In conjunction with other physical layer security techniques, beamforming can create a more robust and secure communication channel.

While beamforming offers significant security benefits, it is not without its challenges. Static beamforming, where the direction remains fixed, can be exploited by attackers who learn the beam pattern and position themselves accordingly. Dynamic beamforming, which changes direction frequently, offers better security but requires more complex implementation and can be susceptible to misalignment due to imperfect channel knowledge or rapid user movement. To address these limitations, researchers have explored various techniques, including machine learning-based beamforming and the use of intelligent reflecting surfaces. These approaches aim to improve the efficiency and adaptability of beamforming, making it a more effective tool for securing communication systems in dynamic and challenging environments.

4.4. Cooperative Communication

Cooperative communication offers a promising approach to enhancing PLS in ISTNs for 6G networks [47]. While PLS leverages the inherent randomness of the wireless channel for security, it faces certain limitations in complex ISTN environments:

• Uncertainty in Channel Characteristics: The diverse nature of terrestrial and satellite links within ISTNs can lead to significant variations in channel characteristics like fading and noise. This makes it challenging to exploit these variations for optimal security using traditional PLS techniques [90].
• Limited Range of Terrestrial Networks: Terrestrial base stations may not always have a direct line of sight to users, especially in remote areas. This can weaken the signal strength and limit the effectiveness of PLS [91].

By exploiting the combined channel variations across different paths in the network, cooperative communication creates a more complex and unpredictable environment for potential eavesdroppers [92]. Cooperative communication strategies can address these limitations by enabling collaboration among different network elements to establish a more robust and secure communication channel. The potential applications are as follows:

• Relaying: Nodes (relays) strategically positioned within the network can receive signals from the source and retransmit them toward the destination. This can help overcome signal weaknesses from long distances or obstructions in terrestrial links, improving the overall channel quality and strengthening PLS mechanisms. Relay nodes can extend the reach of terrestrial networks, enabling secure communication even in remote areas where relying solely on terrestrial links might be challenging for PLS [93,94,95]. Such a scenario is depicted in Figure 5.
• Joint Signal Processing: Network elements can collaborate to jointly process and analyze the received signals. By combining information from multiple sources, they can gain a more accurate understanding of the channel characteristics and exploit them more effectively for security purposes. Relaying and joint processing techniques can enhance signal strength and mitigate channel fading, leading to a more reliable and secure communication channel [94,96].
• Diversity Techniques: Cooperative communication allows for the creation of spatial diversity by utilizing multiple transmission paths. This makes it harder for attackers to exploit specific channel weaknesses and enhances the overall security of the communication link.

In [97], the authors utilized relay nodes to help forward information from the satellite to the weak user, who then uses maximum ratio combining to boost signal strength. Two relay protocols, i.e., decode-and-forward (DF) and amplify-and-forward (AF), were investigated. They derived closed-form and asymptotic expressions for the outage probability and analyzed the diversity order. The results show that DF outperforms AF in outage probability, though with a higher computational load. In [98], the authors explore using LEO satellites as jammers to enhance PLS. An iterative scheme is proposed to maximize system secrecy energy efficiency (SEE) by optimizing transmit power allocation and user association. Extensive experiments demonstrate that the optimization scheme significantly improves SEE and determines optimal power allocation and user association strategies.

In [99], the authors study secure communication in a satellite–terrestrial network overlaid on a legacy cellular network using mm-wave frequencies. The aim was to maximize the secrecy rate of an eavesdropped satellite service by designing a cooperative secure transmission beamforming scheme involving adaptive beamforming, artificial noise, and cooperative beamforming by base stations (BSs). A non-cooperative beamforming scheme was also considered. The secrecy rate maximization problem was addressed with an approximation and iteration-based genetic algorithm. Simulations show that multiple antenna arrays and cooperative beamforming significantly improve secure communication and validate the algorithm’s efficiency. In [100], the authors study the security and reliability of a satellite–terrestrial network, where a satellite sends confidential information to a ground user via multiple relays, with an eavesdropper trying to intercept. A friendly jammer enhances security. Using generalized Rician fading channels and imperfect CSI, the study derives closed-form expressions for outage and intercept probabilities with amplify-and-forward relaying. Monte Carlo simulations validate the results, examining the effects of channel estimation errors, transmit power, number of relays, and fading severity.

In [101], the authors propose an adaptive power–bandwidth cooperative scheme to combat eavesdropping on LEO satellite links communicating with ground stations when drones are present in the line of sight between the satellite and ground station. The algorithm dynamically adjusts resources, i.e., power and bandwidth, to maximize the normalized secrecy capacity while maintaining a reasonable SNR at the legitimate receiver, using a nearby terrestrial party for signal amplification. Simulations show a significant performance improvement of over 70% compared to non-adaptive methods across various elevation angles.

We summarize the works that have used cooperative communication to enhance the PLS of the ISTNs in

Table 4. References using cooperative communication to improve the PLS of the communication network.

Reference	Optimization Variable	Constraint
[97]	Outage probability
[98]	Secrecy energy efficiency	Transmit power and secrecy rate
[99]	Secrecy rate	Power and transmission quality
[100]	Outage probability

. Overall, cooperative communication offers a promising approach to enhancing PLS in ISTNs for 6G networks. While PLS leverages the inherent randomness of the wireless channel for security, it faces certain limitations in complex ISTN environments, such as uncertainty in channel characteristics and the limited range of terrestrial networks. Cooperative communication can address these limitations by enabling collaboration among different network elements, establishing a more robust and secure communication channel.

Through techniques like relaying, joint signal processing, and diversity techniques, cooperative communication can improve signal strength, mitigate channel fading, and create a more unpredictable environment for potential eavesdroppers. This can enhance the overall security of the communication link and make it more difficult for attackers to intercept or disrupt data transmission.

While cooperative communication offers significant potential, its implementation can add complexity to network management. Further research is needed to develop efficient resource allocation algorithms, ensure synchronization and coordination among network elements, and address challenges related to data exchange. Despite these challenges, cooperative communication presents a valuable tool for boosting PLS in ISTNs and creating a more secure and reliable communication environment for future 6G applications.

The disparity between satellite and terrestrial networks can lead to significant differences in signal strength, delay, and interference levels, making it difficult to coordinate cooperative transmissions effectively. Additionally, the dynamic nature of satellite orbits and terrestrial network topologies further complicates the design of cooperative protocols.

Another significant limitation is the high latency associated with satellite links. This can delay the exchange of information between cooperating nodes, impacting the effectiveness of cooperative schemes. Furthermore, the limited bandwidth of satellite links can constrain the amount of data that can be exchanged for cooperative purposes. Finally, the imperfect channel state information available at the nodes can hinder the optimal design of cooperative protocols.

4.5. Physical Characteristics

The physical characteristics of the communication channels themselves offer unique advantages for achieving PLS [102]. Such characteristics can be exploited to secure communication:

• Fast Fading: Terrestrial links experience rapid changes in signal strength and phase due to multipath propagation. This inherent randomness can be utilized for PLS by embedding secret information into the fading patterns. An eavesdropper without knowledge of these patterns would struggle to decipher the transmitted data [103,104].
• Slow Fading: Slower fading, often observed in satellite links due to longer propagation delays, can be used for key generation. By exploiting the time-varying nature of the fading channel, a shared secret key can be established between legitimate communicators without explicitly exchanging any information [105].
• Background Noise: The inherent noise present in all communication channels can be leveraged for security purposes. By adding carefully designed noise to the transmitted signal, it becomes difficult for eavesdroppers to distinguish the actual data from the background noise, enhancing the secrecy of the communication [106].
• Forward and Reverse Links: Terrestrial and satellite links often exhibit different channel characteristics in the forward (source to destination) and reverse (destination to source) directions. This asymmetry can be exploited to create a “secrecy code” embedded in the channel itself. Legitimate communicators with knowledge of this code can successfully decode the message, while eavesdroppers without this knowledge would be unable to do so.
• Terrestrial vs. Satellite Links: The inherent differences between terrestrial and satellite links can be a valuable asset for PLS. By strategically exploiting the unique fading and noise characteristics of each type of link, communication networks can create a more complex and unpredictable environment for potential attackers.

In [107], the authors propose physical layer authentication (PLA) for LEO satellites using Doppler frequency shift (DS) and received power (RP) characteristics. Hypothesis testing via thresholds or machine learning (ML) with a one-class classification support vector machine discriminates between legitimate and illegitimate satellites. Using real satellite data from the system tool kit, the results show that DS achieves higher authentication rates at low elevation angles, while RP is better at high angles. A machine learning algorithm provides a higher authentication rate than the threshold scheme with minimal outliers, but the threshold scheme excels with more outliers.

In [103], the authors propose an advanced signal processing algorithm for authenticating satellite signals received at ground stations. This algorithm leverages the spatio-temporal characteristics of signals as they traverse atmospheric paths, enabling differentiation between satellites. Simulation studies at a ground station in Beer Sheva, Israel, using the GEO satellites AMOS-17 and AsiaSat-7, demonstrate an authentication rate exceeding 95%. By enhancing PLS, the algorithm safeguards satellite and terrestrial communication systems, ensuring robust and secure global services. In [108], the proposed method capitalizes on the unique physical characteristics of electromagnetic signal propagation through stochastic atmospheric channels. This approach aims to safeguard ground stations from machine-in-the-middle (MITM) attacks launched by aerial platforms (APs). By exploiting atmospheric effects like thermal noise and tropospheric scintillation fading, the architecture detects rapid signal fluctuations at ground stations. Signals from legitimate satellites and spoofing APs exhibit distinct noise patterns due to their differing spatial paths through dynamic atmospheric conditions. Wavelet filtering is utilized to extract these variations, processing them through a Long Short-Term Memory network. This approach highlights the efficacy of leveraging physical layer attributes to improve cybersecurity in evolving communication networks.

In [104], the authors propose a novel signal processing algorithm to enhance the PLS of an ISTN against spoofing attacks. The data were collected from two GEO satellites: Nilesat $7^{\circ }$ W and Eutelsat ${33}^{\circ }$ E. The attack model assumes Nilesat to be the legitimate satellite and Eutelsat to be the adversary. The $E_S/N_0$ (dB) of the adversary is varied by adding a term $\delta$ so as to mimic the legitimate satellite. From the time series $E_S/N_0$ (dB) of the satellites, the rapidly varying signal component (RVSC) is extracted. The algorithm utilizes atmospheric signatures that cause rapid signal variations at ground stations, leveraging the spatial time–frequency characteristics associated with incoming signals from different satellites to distinguish their sources, such as the mean frequency, median frequency, bandwidth, standard deviation, kurtosis, and skewness [109]. The parameters derived from these signal variations are fed into a machine learning algorithm, such as a neural network, support vector machine, or Ensemble method, to authenticate legitimate satellites. The details of the ML models can be found in [104].

The results are shown in Figure 6, Figure 7 and Figure 8. These results show the MDR, FAR, and AR as the adversary tunes its $E_S/N_0$ (dB) by varying $\delta$. Also, the results obtained with direct samples of $E_S/N_0$ (dB) are shown for comparison. As the results show, the performance is consistent with the physical properties obtained with the RVSC of the satellite’s signal. The results demonstrate an authentication rate exceeding 97%, with low missed detection and false alarm rates. This method offers the potential for more secure ISTNs. Future research may explore the method’s performance with closely positioned GEO satellites, where increased channel correlation could impact performance.

In

Table 5. References using the physical characteristics of the channel to improve the PLS of the communication network.

Reference	Physical Characteristics
[107]	Doppler shift
[104]	Rapid signal variation
[103]	Spatio-temporal characteristics of signal
[108]	Tropospheric scintillation fading

, we provide summary of works done using physical characteristics to improve the PLS of ISTNs. In the context of ISTNs, modeling and estimating channel characteristics pose greater challenges. Accurately modeling the complex and dynamic nature of ISTN channels is crucial for effective PLS implementation. The complex interplay between terrestrial and satellite links, coupled with diverse propagation environments, introduces additional complexities. Factors such as satellite mobility, atmospheric conditions, and shadowing effects further complicate channel modeling. The accurate characterization of these channels is crucial for efficient resource allocation, interference management, and link performance optimization. To address these challenges, researchers are exploring hybrid channel models that combine terrestrial and satellite channel characteristics, leveraging advanced techniques like machine learning and deep learning for model calibration and prediction.

While noise can be a powerful tool in PLS, it can also introduce challenges. The goal in PLS is to exploit noise to make eavesdropping difficult, but excessive noise can degrade the signal quality to the point where it becomes unreliable. This can lead to ineffective authentication. Excessive noise can obscure the intended signal, making it difficult for the receiver to accurately decode the transmitted data. This can lead to authentication errors, as the receiver may not be able to verify the sender’s identity correctly. Noise can introduce errors that mimic legitimate authentication signals. This can lead to false positives, where unauthorized parties are mistakenly authenticated. Noise can also cause legitimate authentication signals to be missed or misinterpreted. This can lead to false negatives, where authorized parties are denied access.

The physical characteristics of communication channels offer unique advantages for achieving PLS in ISTNs. By leveraging factors such as fast fading, slow fading, background noise, forward and reverse link asymmetry, and the differences between terrestrial and satellite links, networks can create a more complex and unpredictable environment for potential attackers.

While these physical characteristics provide a promising foundation for PLS, their effective exploitation requires accurate channel modeling and reliable parameter estimation. The complex and dynamic nature of ISTN channels, coupled with diverse propagation environments, introduces challenges in accurately characterizing and modeling these channels. Addressing these challenges through hybrid channel models and advanced estimation techniques is crucial for maximizing the effectiveness of PLS in ISTNs.

It is important to note that while noise can be a powerful tool for PLS, excessive noise can introduce challenges such as degraded signal quality, authentication errors, and false positives/negatives. Therefore, the careful management of noise levels is essential for ensuring the effectiveness of PLS while maintaining reliable communication.

Extensive work is ongoing to develop advanced channel models that capture the unique characteristics of terrestrial and satellite links. Reliable estimation of channel parameters at the transmitter and receiver is essential for exploiting channel variations for security purposes, but it is difficult in this scenario. Techniques need to be robust and adaptable to handle dynamic channel conditions. The physical characteristics of ISTN channels offer a powerful foundation for achieving PLS. By leveraging fading, noise, and asymmetry, networks can create a secure communication environment that is inherently resistant to eavesdropping. However, accurate channel modeling and reliable estimation techniques are crucial for maximizing the effectiveness of PLS in these complex network architectures.

5. Future Directions

Research on physical layer cybersecurity for satellite communications is gaining importance as reliance on satellite systems continues to grow across various sectors, from defense and communication to navigation and Earth observation. The future directions is shown in Figure 9.

5.1. Resource Allocation

The use of AI-based resource allocation can be a smart choice in such complex networks for emerging security threats, user demands, and dynamic channel conditions. For adaptive resource allocation, reinforcement learning can be used to interact with the environment to learn the optimal policies. Federated learning can be utilized to develop a distributed resource allocation algorithm to protect user privacy.

In integrated networks, dynamic spectrum sharing can be investigated to ensure secure and efficient utilization of spectrum resources. Furthermore, cognitive radio techniques can be used for spectrum sensing and access for improved spectrum utilization and security. A UAV could serve as an aerial base station for spectrum sharing and security enhancement. Joint beamforming and resource allocation can be applied to maximize secrecy capacity while minimizing energy consumption and interference. Resource allocation techniques should be robust to channel estimation errors and other uncertainties. Cross-layer solutions, including cryptographic solutions, should be attempted along with PLS for enhanced protection. The allocation schemes should be privacy-preserving, support secure data offloading between terrestrial and satellite networks, and secure edge computing. Various trade-offs, such as QoS, security, and energy efficiency, should be taken into consideration.

5.2. Beamforming

As the environment is complex in an ISTN, a careful combination of analog and digital beamforming, i.e., hybrid beamforming, should be considered to balance complexity and performance. Developing deep learning-based beamforming that can dynamically learn about this complex environment should be explored. Minimizing power allocation with beamforming to maximize the secrecy capacity, energy efficiency, and user fairness should be given consideration. Based on channel conditions, user demands, and security requirements, we should investigate the adaptive schemes for robustness and improved system performance. Robust beamforming schemes should take into account channel estimation errors while minimizing interference.

Cooperative beamforming can be explored to enhance secrecy performance and mitigate eavesdropping threats. Physical layer jamming could be considered to degrade the eavesdropping channel. This will improve the security of legitimate communication. Information leakage can be minimized by privacy-preserving beamforming techniques. Secure data aggregation methods should be developed for ISTNs. A mechanism to improve trust among network nodes and manage user reputations should be investigated to enhance security and reliability. Furthermore, cross-layer solutions and trade-offs can be explored for improved system performance.

5.3. Cooperative Communication

Hybrid cooperative schemes combining amplify-and-forward (AF) and decode-and-forward (DF) relaying could be explored to optimize system performance. Scenarios with diverse channel conditions should be taken into account. Intelligent reflecting surface (IRS)-assisted cooperative communication is another potential way to improve the secrecy capacity while minimizing interference and improving energy efficiency. AI-based cooperative schemes can be used to optimize performance in real time by adapting to dynamic channel conditions and eavesdropping attacks.

A cooperative scheme with adaptive resource allocation is another potential way to improve secrecy capacity and energy efficiency. Such techniques can be made resilient to channel estimation errors, interference, and uncertainties. Secure cooperative jamming could be explored to degrade the eavesdropper’s channel and improve the security of the legitimate channel. Secure edge computing considering data privacy and confidentiality can be investigated for complex ISTNs. Furthermore, a trade-off between different requirements based on diverse and dynamic channel conditions can be considered.

5.4. Physical Characteristics

Satellite and terrestrial channel conditions such as Rician fading and Rayleigh fading can be exploited to improve security. By analyzing the Doppler shift and shadowing, an improved security mechanism can be created to improve the system performance. PLS techniques for inter-satellite link security can be considered for specific channel characteristics and potential threats. Terrestrial channel characteristics, such as urban and rural models, can be investigated to optimize the system performance. Large- and small-scale fading is another possible way to distinguish between channels.

Joint channel estimation and beamforming in diverse ISTN channels can be considered for improved security. Channel estimation based on deep learning would be helpful in such scenarios. Considering the diverse channel model, specific security metrics can be devised. Physical layer key generation based on different fading models can be considered for secure communication. By exploiting the channel characteristics, PLKG offers the potential for information-theoretic security, eliminating the need for traditional cryptographic algorithms that rely on computational hardness assumptions.

Advancements in signal processing techniques are crucial for enhancing the resilience of satellite communications to a range of physical layer attacks, including jamming, spoofing, and eavesdropping. Researchers are developing adaptive signal processing algorithms that can dynamically adjust to changing threat environments, employing machine learning to detect and mitigate attacks in real time [110]. Physical layer authentication methods are also being explored, which utilize the unique characteristics of the communication channel, such as the propagation delay, Doppler shift, and angle of arrival, to authenticate signals and detect potential intrusions. This approach adds an additional layer of security by ensuring that the physical properties of the signal match expected parameters, making it difficult for attackers to successfully impersonate or alter legitimate communications.

5.5. Emerging Techniques

Another key area of focus is the integration of Quantum Key Distribution (QKD), which represents a groundbreaking advancement in secure communication [111,112,113]. QKD leverages the principles of quantum mechanics, where quantum bits (qubits) are used to transmit encryption keys [114]. The unique property of quantum states is that they cannot be observed or measured without altering their state, which means any eavesdropping attempt would be immediately detectable by the communicating parties. This makes QKD theoretically immune to many of the vulnerabilities that affect classical encryption methods [115]. However, implementing QKD in satellite communications presents significant challenges, particularly over long distances, where maintaining quantum coherence is difficult due to atmospheric interference and the vast distances involved in space. Research is therefore focused on developing robust satellite-based QKD systems that can maintain quantum states over long distances, exploring methods such as free-space optics and satellite relays to extend the range and reliability of QKD [116].

Dynamic spectrum access and cognitive radios are another area of active research, with the goal of enabling satellite communication systems to dynamically adjust their spectrum usage in response to threats or interference [117]. Cognitive radios are designed to sense their electromagnetic environment and intelligently select the best frequency bands to use, thereby avoiding congestion and potential attacks [118]. This adaptability is particularly important in contested environments where spectrum access may be limited or under threat from adversaries.

Anti-jamming and anti-spoofing techniques are being further developed to protect satellite communications from deliberate disruptions. Spread spectrum techniques, such as frequency hopping and direct-sequence spread spectrum methods [119,120], are being refined to make it more difficult for adversaries to jam or intercept signals. These techniques, combined with advanced encryption methods, provide a robust defense against a wide range of attacks. Additionally, machine learning algorithms are being employed to detect and respond to jamming and spoofing attempts in real time, allowing for rapid adaptation to emerging threats [121,122].

Secure satellite network architectures are being explored to ensure that satellite communications are inherently secure at the physical layer. This includes the development of decentralized network designs that reduce the risk of single points of failure and the integration of blockchain technology to ensure the integrity and confidentiality of data transmitted across satellite networks. Blockchain can provide a transparent and tamper-resistant record of transactions and data exchanges, which is particularly valuable in ensuring the security of communication in a decentralized network [123,124].

Resilient satellite communication protocols are also being developed to maintain operation even when the system is under attack. These protocols incorporate features such as error correction, redundancy, and secure routing [125] to ensure that communications can continue even in the presence of physical layer attacks [126]. By incorporating these features into the design of communication protocols, researchers aim to create systems that can degrade gracefully and continue to function in a compromised environment.

The integration of AI and machine learning into satellite communication systems is increasingly seen as vital for enhancing security. AI and machine learning can be used to detect, classify, and mitigate physical layer attacks by analyzing patterns in communication signals and identifying anomalies that may indicate an attack. These technologies can also be used to optimize system performance and resource allocation in real time, further enhancing the resilience of satellite communications [127,128].

Cross-layer security approaches are being pursued to provide comprehensive end-to-end protection for satellite communication systems. This involves integrating physical layer security with higher-layer security protocols, creating a holistic security framework that addresses vulnerabilities across all layers of the communication stack. By considering security at every layer, researchers aim to develop systems that are resilient to a wide range of threats.

Standardization and policy development are also critical for ensuring the security of satellite communications on a global scale. International collaboration between academia, industry, and government agencies is necessary to create standardized security protocols, certification processes, and regulatory frameworks that protect global satellite systems. These efforts are particularly important as the number of satellites in orbit continues to grow, increasing the complexity and potential attack surfaces of satellite communication networks.

There is growing interest in understanding the impact of space environment effects, such as space weather and cosmic radiation, on the physical layer security of satellite communications. Space weather events, such as solar flares and geomagnetic storms, can disrupt satellite signals and potentially compromise security. Research is therefore focused on developing techniques to mitigate the effects of these environmental factors, ensuring that satellite communication systems remain secure and reliable even in challenging conditions.

The inclusion of signatures in NTN watermarking offers a promising avenue for enhancing data authenticity and traceability. While this technique is well established in GNSS, its application to NTNs warrants further investigation [129]. By embedding signatures within the message or spreading code, it becomes possible to verify the origin and integrity of transmitted data, mitigating the risks associated with unauthorized modifications or fraudulent activities.

The dissemination of signatures in NTN environments presents unique challenges due to the distributed nature of these networks. The TESLA protocol, designed for the secure dissemination of cryptographic keys, can be adapted to address this issue [130]. By carefully considering the signature type, placement, and dissemination mechanisms, researchers can develop effective and secure signature-based watermarking techniques that cater to the specific requirements of NTN applications.

Collectively, these research efforts aim to enhance the resilience and security of satellite communication systems in an increasingly connected and threatened world, addressing the unique challenges posed by the space environment and the evolving nature of cyber threats.

6. Conclusions

The convergence of terrestrial and non-terrestrial networks in ISTNs for 6G presents both exciting possibilities and significant security challenges. Traditional cryptographic methods face limitations in ISTNs due to potential scalability issues and the evolving threat of quantum computing. PLS emerges as a promising alternative, leveraging the inherent physical properties of communication channels for secure communication. This review explores the potential of PLS in ISTNs, examining how channel characteristics like resource allocation, beamforming, cooperative communication, and physical characteristics can be exploited to create a secure communication environment. Additionally, this paper discusses the challenges and future directions associated with implementing PLS in ISTNs, such as channel modeling and reliable estimation. As research on 6G and ISTNs continues to evolve, PLS holds immense potential to ensure secure and reliable communication in this complex network landscape. By harnessing the unique physical properties of the channels and employing advanced signal processing techniques, network designers can create a future where ISTNs offer not only ubiquitous connectivity but also robust security for a wide range of 6G applications.