Routing Challenges and Enabling Technologies for 6G–Satellite Network Integration: Toward Seamless Global Connectivity

Abstract

The capabilities of 6G networks surpass those of existing networks, aiming to enable seamless connectivity between all entities and users at any given time. A critical aspect of achieving enhanced and ubiquitous mobile broadband, as promised by 6G networks, is merging satellite networks with land-based networks, which offers significant potential in terms of coverage area. Advanced routing techniques in next-generation network technologies, particularly when incorporating terrestrial and non-terrestrial networks, are essential for optimizing network efficiency and delivering promised services. However, the dynamic nature of the network, the heterogeneity and complexity of next-generation networks, and the relative distance and mobility of satellite networks all present challenges that traditional routing protocols struggle to address. This paper provides an in-depth analysis of 6G networks, addressing key enablers, technologies, commitments, satellite networks, and routing techniques in the context of 6G and satellite network integration. To ensure 6G fulfills its promises, the paper emphasizes necessary scenarios and investigates potential bottlenecks in routing techniques. Additionally, it explores satellite networks and identifies routing challenges within these systems. The paper highlights routing issues that may arise in the integration of 6G and satellite networks and offers a comprehensive examination of essential approaches, technologies, and visions required for future advancements in this area. 6G and satellite networks are associated with technical terms such as AI/ML, quantum computing, THz communication, beamforming, MIMO technology, ultra-wide band and multi-band antennas, hybrid channel models, and quantum encryption methods. These technologies will be utilized to enhance the performance, security, and sustainability of future networks.

1. Introduction

The field of network technologies is rapidly evolving to meet the growing demands for speed, reliability, and ubiquitous connectivity. Historically, a new generation of wireless communication systems has emerged approximately every decade, each aiming to address the limitations of its predecessor [1]. From the analog 1G systems of the 1980s to the digital GSM-based 2G networks of the 1990s, and onward to 3G, 4G LTE, and the currently deployed 5G, each generation has introduced substantial enhancements in data rates, latency, and user capacity [2,3,4,5]. The commercial launch of 5G in 2019 brought significant improvements, such as ultra-reliable low-latency communication (URLLC), enhanced mobile broadband (eMBB), and massive machine-type communication (mMTC) [1]. However, the escalating demands from emerging applications now highlight several critical limitations of 5G.

Applications such as immersive extended reality (XR), holographic communication, autonomous systems, and brain–computer interfaces require latency on the order of microseconds and data rates approaching terabits per second (Tbps), which are beyond the capabilities of 5G [6,7,8]. Furthermore, the exponential increase in the number of connected devices, particularly in Internet of Everything (IoE) scenarios, demands extremely high connection density, energy efficiency, and intelligent resource management. These limitations have accelerated the need for sixth-generation (6G) networks.

6G, expected to be realized around 2030, aims to support ultra-low latency (<0.1 ms), Tbps-level throughput, extreme reliability, and pervasive global connectivity [6,9,10]. Figure 1 summarizes latency requirements across different 5G and 6G application domains, illustrating the stringent demands placed on future network designs. This vision is underpinned by key enabling technologies such as terahertz (THz) communication, visible light communication (VLC), integrated terrestrial and non-terrestrial networks (NTNs), reconfigurable intelligent surfaces (RISs), and advanced artificial intelligence (AI)-driven network control [11,12,13,14,15,16,17,18]. For example, THz bands (0.1–10 THz) offer unprecedented bandwidth, but their high propagation loss and line-of-sight constraints demand new signal processing and routing strategies. Simultaneously, AI and machine learning (ML) are poised to transform network optimization and decision-making, providing adaptive and context-aware management across diverse, dynamic environments [19,20,21].

One of the most significant shifts in 6G is the seamless integration of terrestrial, aerial, maritime, and satellite segments—often referred to as the 3+ paradigm—to ensure truly global coverage. Satellite networks, particularly low Earth orbit (LEO) constellations, play a vital role in achieving ubiquitous broadband access, especially in underserved and remote areas [22,23,24,25]. However, this integration introduces new challenges in network architecture, mobility management, and, most notably, routing.

Routing in 6G and integrated satellite-terrestrial networks becomes increasingly complex due to the heterogeneous, dynamic, and large-scale nature of these environments. Traditional routing protocols, designed for static or slowly changing networks, are inadequate to ensure low latency and reliable communication in such settings. AI-enabled routing methods, including deep reinforcement learning (DRL), genetic algorithms (GAs), and neural networks, have been proposed to address these issues. These techniques offer adaptability, scalability, and real-time optimization capabilities necessary for 6G routing demands [26,27,28,29,30,31,32].

Despite the growing body of literature on 6G and routing, there remains a lack of comprehensive analysis that bridges the domains of AI-driven routing and integrated satellite-terrestrial networks. The contribution of this paper lies in its critical and structured review of the following:

• 6G architecture, enablers, and limitations;
• Satellite networks and their routing techniques;
• Challenges and solutions for routing in integrated networks;
• The role of AI algorithms in enhancing routing performance.

This survey not only synthesizes the state of the art but also identifies open research directions to guide future innovations in this emerging field. The subsequent sections of this paper are structured as illustrated in Figure 2. Section 2 discusses challenges and routing in 6G networks. Section 3 shows satellite networks, routing protocols in satellite networks, and routing protocols in LEO networks. Section 4 presents routing in integrated networks. Section 5 illustrates the concerns and implications of AI algorithms in routing. Section 6 focuses on future approaches to research challenges. Finally, Section 7 concludes this article.

2. Challenges and Routing in 6G Networks

In 6G networks, different routing scenarios are used to control and enhance data flow. These scenarios include unicast routing, broadcast routing, multicast routing, and anycast routing. Different forms of routing scenarios have unique traits, obstacles, and prerequisites. Unicast routing is the most common type of routing scenario in which a source delivers data to a specific destination. Unicast routing is used for applications such as video streaming, file transfer, and online gaming. The main obstacle in unicast routing is to identify the optimal path from sender to receiver, aiming to reduce transmission lag and dropped packets. However, broadcast routing is used to transmit data from one origin to every node in the network. Broadcast routing is used for applications such as critical warnings, software upgrades, and network announcements. The main challenge in broadcast routing is to make sure that information reaches its destination to all nodes in the network while minimizing transmission lag and dropped packets. Multicast routing is employed to send data from a single source to multiple destinations. Multicast routing is used for use cases like video conferencing, internet gaming, and live streaming. The main obstacle in multicast routing is to determine the best tree layout that links the source to all destinations while minimizing transmission lag and dropped packets. Anycast routing is used to transmit communication initiated by a single source towards the closest endpoint in a group of destinations. Anycast routing is used for applications such as content distribution networks and distributed databases. The main obstacle in anycast routing is to find the nearest endpoint that can provide the best quality of service (QoS) while minimizing transmission lag and dropped packets [33,34].

2.1. Challenges of Routing

High efficient routing is an important issue for the 6G and satellite networks. The existing routing protocols were designed many years ago. In the era in which they were designed, it was, naturally, not foreseen that the technology, the internet, and the number of users would experience such an explosion. Although traditional routing protocols are far from meeting the needs of 6G networks, they are also insufficient to meet the commitments of 6G. Existing routing protocols fail to meet the needs for the following reasons, and these can also be seen in summary form in Figure 3.

2.1.1. The Complex Structure of 6G

The aim is to develop new structures and approaches that will contribute to the complexity of the system in 6G and next-generation networks. Below, the most significant versions are demonstrated.

THZ levels

It is aimed that the data transfer capacity in 6G networks will be more than 20 Gbps per second. Spectrum and energy efficiency are projected to be 3 times and 10 times higher than the previous generation, respectively. In the 6G era, goals such as high data rates, spectra, and energy efficiency require the successful implementation of the THz radio communication spectrum [35,36]. Ultimate extreme connectivity will be possible thanks to the THz spectrum, which will play an active role in 6G. The higher the data rate, the wider the bandwidth in the THz spectrum. The 230 GHz spectrum is assigned to mobile services in the 100–450 GHz THz bandwidth range [37]. Thanks to this notable feature, it makes it possible to reach high data rates in certain distance ranges. The wide bandwidth and short wavelengths of the THz spectrum provide high resolution. This potential will take sensitive and high perception to a more advanced level in the future. The development of sensitivity and high detection will contribute to location determination and imaging systems. The 6G world has the potential to become “a new door” that opens to the physical, virtual, biological, and cyber worlds supported by “sensing-supported communication” [37]. It will also support real-time detection and digital twin technology to achieve a major breakthrough. Achieving the targeted high data rates makes the use of the THz spectrum band inevitable. In addition to its potential to provide high data rates, the THz spectrum band also increases the difficulties of communication. High frequencies between 0.1 THz and 10 THz are used in the THz spectrum. The use of these high frequencies leads to high path loss [38,39,40,41,42]. One of the biggest challenges for the THz band is the excess of the free space path loss. To compensate for this challenge, beamforming technology will be needed, which will be even easier and more effective with the support of sensitive and high detection [37]. Otherwise, the area that will benefit from the potential of THz technology will have to narrow. In addition, the THz band is a challenge in interacting with gases in the atmosphere. Since the 6G THz ranges are short, there is power loss between the antennas, the receiver, and the transmitter. This makes it difficult to produce a stable signal at ultra-wide band frequencies. In studies on this subject, a power amplifier is being developed to make it easier to produce a stable signal [43]. In addition to the propagation-related challenges, another critical barrier in the implementation of THz communication is the fabrication of antennas and components that can efficiently operate at such high frequencies [44]. As the THz band demands extremely small wavelengths, the design and manufacturing of compact, high-precision antennas become increasingly complex. Material limitations, thermal stability, and fabrication tolerances at the nanoscale significantly impact performance, reliability, and cost [44]. Moreover, the development of THz components, such as mixers, modulators, and detectors, is still in its infancy compared to microwave and millimeter-wave technologies. These hardware limitations remain major obstacles to the large-scale deployment of THz systems, necessitating continued advancements in nanotechnology, microfabrication techniques, and novel materials suitable for THz frequencies.

VLC

VLC is one of the potential technologies for use in order to achieve the targeted outputs. VLC uses the visible spectrum. The benefits of VLC technology are a license-free 400–800 THz spectrum range, very high spectrum reuse, ultra-high bandwidths, low cost, and zero electromagnetic interference [45,46,47,48].

Blockchain

It is estimated that there will be an incredible explosion in the number of users and devices linked to 6G networks. There will also be an explosion in the data that will be obtained from these connected users and devices. However, the increasing digitalization of life and the increase in the number of data collected every day lead to the fact that the issue of data security is on the agenda more and more. It is also thought that Blockchain technology will be useful in managing and securing these huge data sets [49]. The decentralized nature of Blockchain technology enhances its resilience against attacks in various digital environments. In this technology, the data is encrypted, anonymized, and saved as public. Keeping the recorded data in a distributed and encrypted manner constitutes a commitment to the system’s security. Blockchain technology can be employed to handle and coordinate vast quantities of data and extensive networking within the 6G network. Additionally, it can be utilized for spectrum sharing, allowing users to access the same spectrum, thereby resolving the challenge of substantial spectrum demands in 6G and guaranteeing dependable, affordable, intelligent, and efficient spectrum utilization [50,51,52,53].

Machine learning

The approaches of artificial intelligence, machine learning (ML), and deep reinforcement learning have managed to find a place in the center of 6G networks, as well as in the center of many industry verticals. The keystone of AI ML technology is data. They are technologies that learn from data, make inferences, and take action in this direction. It is thought that the data boom in 6G networks will also be an enormous resource for AI. Increasing coverage, integrating space–air–underwater areas into terrestrial networks, and increasing the number of users and devices connected to the network are elements that increase the complexity of the network. In combating these elements, AI is not a magic wand, but it provides a driving force. In the management of the network regarding the physical layer and higher layers [54] and in the management of resources, the need for solutions that are smarter is growing more and more. There are studies in which machine learning algorithms are used to facilitate resource management, traffic management, and routing management in networks [55]. How to compress, store, and transmit this increasing volume of data poses a new challenge. In addition, processing and storage costs must be minimized. Parameters such as the reliability of machine learning algorithms, the legal regulations for artificial intelligence, the reliability of personal data, and flexibility are important [56]. Studies continue to focus on various solutions to these parameters. In order to ensure data security, the federating learning approach is proposed. With federating learning, data is trained on end devices rather than in a public environment. By using this learning technique, the objective is to protect data security [57].

2.1.2. The Dynamic Structure of 6G

With the increasing population that will connect to 6G networks, it is envisaged that more and more user devices, autonomous driving vehicles, drones, satellite networks, unmanned aerial vehicles, and IoT devices will be connected to the network in a highly dynamic way. The aim is to design a dynamic and intelligent 6G topology, together with critical services such as uMBB, ULBC, and mULC, which are targeted to be reached in 6G networks. This will significantly increase the complexity of the network. Moreover, the dynamic nature of the network often increases routing and maintenance, and this will cause a weakening in routing performance. At this point, it is hoped that more intelligent and foresighted systems will help to overcome this complexity. AI and deep AI approaches have critical potential in this area and are the first approaches that come to mind.

2.1.3. High Packet Loss Regarding Traditional Routing

It is not possible for traditional routing techniques to respond to the dynamic topology of 6G networks, communication occurring in multiple environments at any time, and ultra-reliable and low-latency targets. Traditional routing techniques are limited when it comes to working effectively in dynamic environments. High road loss and congestion problems could cause major losses in the performance of next-generation communication networks. Existing routing approaches need to be improved for 6G integration, and new routing algorithms need to be studied. Efficient and sustainable routing structures are of great importance for 6G and satellite networks. Various studies have been carried out to reduce transmission delay, road loss, and congestion problems. In these studies, machine learning, deep learning, and deep reinforcement learning algorithms were used, together with new routing approaches for efficient traffic management [29,55,58].

2.1.4. Heterogeneous Structure of 6G

It is aimed that 6G and next-generation networks will meet the various service needs of many communication environments with different dynamics. Virtual reality, the integration of space, air, ground, and even underwater spaces, IoT, holographic technologies, smart cities, unmanned aerial vehicles, and unmanned ground vehicles need better and new approaches and heterogeneous systems to meet and optimize the increasing demands of users. The independence of communication from place and time is among the greatest expectations of users and systems. It requires a coalition of cellular and ad-hoc networks to serve this expectation in heterogeneous networks (HWNs). Ad-hoc networks are network connections that do not need a wireless access point or router for communication. In these networks, the devices act as a router and wireless access points [59]. HWNs provide communication through stations by using the structure of the cellular network or by directly using the structure of ad hoc Networks. HWNs, thus, represent an increased version of the capability of both cellular networks and ad-hoc networks. This ensures high-speed transmission of data [60,61].

2.1.5. Three-Dimensional Networking

The biggest vision of 6G networks is to provide timeless and spaceless communication. In order to provide communication independent of time and space, many areas of communication must be integrated. It is aimed that 6G networks will cover all of the following areas: space, air, land, and even underwater. Satellites, unmanned aerial vehicles (UAVs), drones, and high-altitude platform stations (HAPSs) will play a crucial role in 6G networks [62]. Owing to the integration of ground, air, space, and underwater areas and 6G networks, this should support communication in three-dimensional space. New methods are needed to combine these areas with different needs into a single platform.

2.1.6. Low Ability to Adapt to Dynamic Networks

As discussed in detail in the previous sections, 6G networks contain many complex systems, techniques, approaches, and goals. It is not possible for existing traditional methods and systems to efficiently serve such diverse networks. In order to produce optimal and efficient outputs, maximize network performance, achieve the predicted goals, and respond quickly to changes in the network, existing mobility management and routing techniques need to be developed to produce a quick reaction. It is impossible to achieve quick reactions using manual techniques. For this reason, strategies involving machine learning and deep learning come to the fore, as the existing routing protocols cannot meet the needs of 6G networks. Existing routing protocols should be redesigned intelligently in order to provide network optimization in 6G and adapt routing protocols to rapid changes. One of the main goals of satellite networks is the simplification of the routing process. The routing step is a key point of communication technology. Optimizing the routing process is a process that increases the end-to-end efficiency of communication. As mentioned earlier, existing routing techniques have difficulties in meeting the needs of 5G and beyond. Therefore, it is not possible for us to achieve full efficiency in satellite networks by using existing routing protocols that we cannot integrate into ground networks at full efficiency. The dynamics of satellite networks differ significantly from those of ground networks [63,64,65]. In addition, we will be talking about and even trying to implement the incorporation of 6G networks into satellite networks in the future. For this reason, there is already a need for more viable, efficient, and new architectures and approaches. Routing problems in satellite networks can be summarized by the following items:

• LEO and MEO satellite networks have a dynamic structure due to the fact that they are mobile relative to the ground;
• Their coverage area constantly changes due to their movement;
• Satellite networks have high latency, high power, and high bandwidth;
• Transmission losses are greater in satellite networks.

2.2. Routing in 6G Networks

Factors such as the great increase in the number of smart devices on a global scale, technological developments, the exponential increase in available information, and the expansion in the speed, capacity, and coverage area demanded by users constantly trigger new developments in mobile broadband technologies. The rising count of devices bonded to the internet, including machine-to-machine applications, Internet of Things, energy efficiency in communication devices and equipment, and increasing speed and capacity demand have brought 5G and beyond directly into the agenda [66]. The global mobile sector is focused on 5G technology. In addition, 5G technologies continue to evolve. The global research community has quickly started 6G studies. With the technologies that will come with 6G technology, more intelligent, efficient, environmentally friendly, and autonomous communication systems will be possible. At this point, network routing optimizations are critical to ensure the full efficiency of networks regarding 6G technologies.

2.2.1. State of the Art of 6G

The focus of the earliest studies on 6G has been around the question of whether 6G vision is really needed. One of the studies that focuses on this point is the study of David and Berndt, published in September 2018. They examined each generation of wireless technologies according to the advantages and disadvantages they have. They also drew conclusions about the 6G vision and what the user wants [67]. Nawaz et al. [68] conducted a study on the combination of quantum computing and machine learning with the latest technologies in the field of telecommunications. In [19], Rappaport et al. describe many of the technical difficulties and instances of 6G for cordless networks above terahertz (THz). The authors of [20] reviewed 6G in terms of time, frequency, and space resource utilization, especially its problems such as security. In [69], two key 5G technologies are SDN and NFV, which are projected to move to a new level with 6G. In [70,71,72,73,74,75,76,77,78], the authors aimed to shed light on the far-sightedness, necessities, key approaches, and architecture of 6G. In [45], Strinati et al. started with reference to Nikola Tesla’s prophecy in 1926; it was emphasized that this prophecy can be realized with the studies that will take place towards 2030 and that the end user will access a network surrounded by a “huge artificial brain”. It has been stated that 6G will take on a key task at this point and that 5G has great potential; it has transformed many sectors since its launch in 2019, created new sectors, and has even influenced society and revolutions. Is there any need to talk about 6G despite the breakthroughs that 5G has made compared to existing technologies? In the study, the focus is on technologies that will not appear in 5G but will be in 6G. It focuses on sub-THz, visible light communication (VLC), prevalent AI on the network side, three-dimensional (3D) inclusion consulting of tellurian networks, airy platforms, and satellite configuration technologies, which will be in the 6G architecture. The terrestrial and non-terrestrial network architecture of 6G is shown in Figure 4. Huang et al. [79] present a detailed survey approach, trying to unify all tellurian and non-tellurian networks, with a focus also on VLC and THz communication techniques, which are also prominent in 6G.

The 6G Flagship schedule was started by the Finnish University of Oulu in May 2018. The aim of this program is 5G acceptance and elaborating on 6G. As the most comprehensive 6G research schedule in the world, it focuses on developing future wireless technologies. To date, the program has produced more than 1500 peer-reviewed papers and 13 6G white papers. In March 2019, 6G Flagship initiated the world’s very earliest series of 6G events, the 6G Wireless Summit [80]. The Next Generation Mobile Networks (NGMN) Alliance is working intensively on 6G activities to respond to the needs of the ecosystem. In July 2018, the International Telecommunication Union created the Network 2030 center ensemble. This ensemble aims to research the proficiencies of networks for 2030 and beyond [81]. The European Commission has adopted the freshly launched Smart Networks and Services Joint Undertaking (SNS JU) toward 6G, which will be supported by a public fund of EUR 240 million in the first work program of 2021–2022. This initiative, which seeks to facilitate European players in their construction of research and development (R&D) capacities for 6G systems, will be supported by an amount of EUR 900 million to be covered by the private sector in the next seven years. At the Mobile World Congress held in Barcelona on 1 March 2022, speakers from the industry emphasized 6G technologies as the next step in transitioning from gigabit capacities to terabit capacities in performance and achieving sub-millisecond response times. It was stated that such technologies will offer strategic opportunities for the development of new markets. It was also emphasized that 6G will contribute to sustainable development goals by supporting energy efficiency, greening the economy, and digital participation. A public–private budget of EUR 2 billion euros was committed to carrying out 6G R&D activities.

Table 1. Summary of existing survey articles on 6G networks.

Refs.	Major Features and Characteristics
2018; ref. [67]	They examined each generation of wireless technologies according to the advantages and disadvantages they have.
2019; ref. [68]	They conducted a study on the combination of quantum computing and machine learning with the latest technologies in the field of telecommunications.
2019; ref. [19]	It describes many of the technical difficulties and occasions for 6G in cordless networks above terahertz (THz).
2019; ref. [20]	It examines 6G in terms of time, frequency, and space resource usage, especially security issues.
2019; ref. [69]	The potential integration of SDN and NFV technology with 6G is presented.
2019, 2020, 2021; refs. [70,71,72,73,74,75,76,77,78]	The authors aimed to shed light on the far-sightedness, vision, requirements, key approaches, technologies, and architecture of 6G.
2019; ref. [45]	It focuses on technologies that will not be in 5G but will be in 6G. The study focuses on sub-THz, visible light communication (VLC), prevalent AI in network side technologies that will be in the 6G architecture.
2019; ref. [79]	They present a detailed survey of the 6G approach, trying to unify all tellurian and non-tellurian networks; they also focus on VLC and THz communication techniques, which are also prominent in 6G.
2019, ref. [80]	The aim of the 6G Flagship program is to gain 5G acceptance and elaborate on 6G. It focuses on developing future wireless technologies.
2019; ref. [81]	This ensemble aims to research the proficiencies of networks for 2030 and beyond.

contains the studies given in this section.

Table 1 presents a compilation of significant survey and visionary articles that laid the foundation for early 6G research. To provide a deeper understanding, a comparative analysis is needed. The authors of [67] uniquely questioned the necessity of 6G and mapped generational progress, serving as an early reflection piece rather than a technical survey. On the other hand, the authors of [68] explored quantum computing and machine learning, bridging future computing paradigms with telecommunications. In [19], one of the first technical explorations of THz communication is offered, emphasizing the practical hurdles. In addition, the studies in [20,69] addressed the architectural and security challenges in 6G but lacked detailed implementation insights. The collective work in [70,71,72,73,74,75,76,77,78] provided a holistic vision of 6G requirements and architecture but remained conceptual. In [45], a philosophical and AI-centric perspective is provided, while [79] delivered a comprehensive unification of tellurian and non-tellurian domains, covering VLC and THz technologies with greater depth. The 6G Flagship program [80] stands out as the most extensive and ongoing effort, generating a large body of research and strategic whitepapers. Lastly, ref. [81] and the NGMN/ITU initiatives mark global governance perspectives but offer limited technical depth. In terms of pros, the early works are visionary and cover wide-ranging aspects such as frequency bands, intelligent surfaces, AI integration, and sustainability goals. The cons include redundancy across some studies, a lack of implementation scenarios, and the absence of unified modeling or standardized frameworks. Therefore, while these studies lay important groundwork, a need remains for more technically detailed, comparative, and implementation-oriented surveys to guide both academia and industry.

2.2.2. Review of 6G Routing

In academia, 6G is an especially hot topic that aims to supply users with rich quality of service (QoS) using ubiquitous mobile broadband (uMBB), Ultra-Reliable Low-Latency BroadBand Communication (ULBC), and massive Ultra-Reliable Low-Latency Communication (mULC) [82,83]. The optimization of the network is critical for 6G to achieve its intended end-to-end QoS at rapid speeds, ultra-low response times, and enhanced mobility. Guaranteeing end-to-end quality, dependability, and service depends mainly on routing algorithms and network congestion management. High efficient routing is an important issue for the 6G networks. The literature on routing protocols in 6G networks is given in

Table 2. Summary of existing survey articles on routing approaches in 6G networks (1—6G focused; 2—Satellite focused; 3—Integration of 6G and Satellite focused; 4—Taxonomy available).

Ref.	Year	Objective	Merits	Demerits	1	2	3	4
[55]	2021	- To present a survey on 6G networks to ensure end-to-end QoS and QoE	- The end-to-end communication process is examined, emphasizing network access and resilient routing management	- The answer to the question of what should be the general standards of the proposed models is incomplete	✓	✗	✗	✓
[84]	2021	- They proposed deep learning-based, smart stochastic routing to improve energy-efficient routing	- Faster, energy-efficient, low-processing latency, and reliable model	- The lack of diversity of deep learning algorithms used	✓	✗	✗	✓
[29]	2021	- Presents a DD-CCSR approach using Deleroi superposition and an onward-backward interface to reduce congestion and transmission delay, maximize bandwidth, and balance network load	- A better bandwidth utilization and packet arrival rate were achieved	- It should be tested on larger-scale networks. Additional loads that this method can bring to the network should be detailed	✓	✗	✗	✗
[58]	2022	- They proposed a novel routing strategy aimed at minimizing overall latency in a 6G wide area network	- The average end-to-end delay was found to be less than 1 ms	- It would be useful to evaluate the proposed approach under different network topologies, dimensions, and traffic conditions	✓	✗	✗	✗
[85]	2022	- They introduced a new prognostic QoS routing algorithm to increase QoS in beyond-5G networks - They developed an algorithm for latency-tolerant eMBB streams	- The proposed proactive model performed better than the reactive model	- The study can be enriched with different ML models and future network metrics	✓	✗	✗	✗
[86]	2022	- They propose P-HEUR, a new model that optimizes energy-saving user assignment, routing of backhaul traffic, and base station/backhaul link transitions	- P-HEUR is able to save energy, with lower execution times, higher feasibility, and lower unsatisfied user probability versus state-of-the-art models	- More information could be provided on the applicability of the proposed methods in real-world scenarios	✓	✗	✗	✗
[87]	2022	- The ONE approach addresses instantaneous user connections, the routing of traffic, and VNF allocation to enhance energy efficiency and user acceptance in mobile networks	- It achieved a maximum of 89% optimum energy efficiency while consuming up to 90% less computation time	- Beyond that, the development of the heuristic approaches used and comparative analyses in a more comprehensive manner	✓	✗	✗	✗
[58]	2022	- They proposed a new routing strategy aimed at minimizing end-to-end response time in a 6G wide area network	- The average end-to-end delay was found to be less than 1 ms	- It would be useful to evaluate the proposed approach under different network topologies, dimensions, and traffic conditions	✓	✗	✗	✗
[88]	2023	- They proposed a wireless intelligent router for resource allocation and re-routing	- It reduces average power consumption and delay	- Tested on a small network - Q-table scalability can be a problem as the network grows	✓	✗	✗	✗
[89]	2023	- A reliable and effective routing protocol is proposed for self-driving vehicles in 6G networks	- Better output regarding data transmission and reception, loss percentage, latency, and energy effectiveness	- It can be improved with real data and machine learning approaches	✓	✗	✗	✗

.

In 2021, Tang et al. worked on streamlined traffic handling on 6G networks to ensure end-to-end QoS and QoE. This study discussed the fact that the management processes in existing networks will not work efficiently for 6G networks through network access and routing to traffic management and streaming adaptation because of the intricate and ever-changing scenarios of 6G. They proposed that the use of machine learning, deep learning, and deep reinforcement learning algorithms in network access can pave the way for network routing, congestion control, and network flow control, optimizing a reduction in issues within the network. Proposals exist for both land-based and satellite network solutions. While attention was drawn to what should be the criteria for evaluating the performance of ML models, no answer was given to the question of what is the promising solution to address this problem [55].

In 2021, Abbas et al. worked on energy-efficient routing for resource-scarce massive-IoT networks. In 6G-enabled massive-IoT networks consisting of many devices, traditional stochastic routing algorithms use a lot of energy and time due to the computational load needed to find the best node for routing. They proposed deep learning-based, smart stochastic routing to improve this situation. Using deep neural networks and the Matlab simulation program as the methodology, they obtained a faster, energy-efficient, low-latency, and reliable model [84].

In 2021, Deebak et al. worked on dynamically driven congestion control in 6G networks. With the maturation of mobile networks, it is becoming increasingly difficult to ensure network efficiency and system performance in incredibly growing communication technologies. In this study, dynamically driven congestion control and segment redirection (DD-CCSR) using Deleroi superposed principles and onward-backward interfacing based on reducing congestion control to reduce transmission delay, maximize bandwidth utilization, and balance network loads are presented. With this method, the aim is to reduce data flow and over-crowding and boost network traffic observation and network trail rectification. As a result, better bandwidth utilization and packet arrival rates have been achieved. However, it should be tested on larger-scale networks. The additional loads that this method can bring to the network should be detailed [29].

In 2022, Das et al. proposed a new routing strategy aimed at minimizing overall latency in a 6G wide area network (WAN). In this proposition, they used a mixture of fiber optic and cordless architecture for the WAN, featuring a spine grid network established by high-speed optical fibers, complemented by local cordless network admission points. The initial and final routings are achieved via the wireless channel, and the intermediary points are achieved via the backbone optical network. In the theoretical and simulation results, the average throughout delay was found to be less than 1 millisecond (ms) [58].

In 2022, Akçapınar et al. introduced a new prognostic QoS routing algorithm to increase QoS in beyond-5G networks by using Auto Regressive Integrated Moving Average (ARIMA). The proactive model they recommended in their study, which focused on eMBB flows, outperformed the reactive model [85].

In 2022, Mesodiakaki et al. suggested a novel model called P-HEUR to address the integrated issue of optimizing energy-efficient user assignment, traffic routing in backhaul, and switching of base station/backhaul links. They aimed to solve problems such as high path loss, antenna gain, cost, and the unstable links of the small cells densification concept, which will be brought about by structures such as mmWave that will be needed to support high-capacity and low-latency future networks such as 5G and 6G. The proposed P-HEUR model is capable of achieving energy-efficient solutions with 83% optimality while requiring 200 k times lower execution times compared to the state-of-the-art models. P-HEUR demonstrated higher feasibility and lower unsatisfied user probability compared to the optimal case due to an increased number of active BSs and BH links accommodating rate uncertainty, with up to 57% and 64% improvements, respectively. More information could be provided on the applicability of the proposed methods in real-world scenarios. Additionally, future research could discuss how this method might be further improved [86].

In 2022, Mesodiakaki et al. proposed an approach named ONE, which was introduced to tackle the challenges of real-time user assignment, traffic routing, and Virtual Network Function (VNF) deployment, with the aim of improving the energy efficiency and user admission ratio of mobile networks. The suggested remedies are demonstrated to outclass the current best practices in energy efficiency, where ONE attains a maximum of 89% optimal energy efficiency while consuming as low as 90% less computational time, even in conditions with intense user traffic. Further development of the heuristic approaches used in the article and more comprehensive comparative analyses may provide more accurate and reliable results [87].

In 2022, Das et al. proposed a new routing strategy aimed at minimizing overall latency in a 6G wide area network (WAN). In this approach, they employ a hybrid fiber optic wireless framework for the WAN, featuring a grid network backbone composed of ultra-fast optical fibers and local wireless network access hubs. The initial and final routings are made via the wireless channel, and the intermediate points are made via the backbone optical network. In the theoretical and simulation results, the average end-to-end delay was found to be less than 1 ms. It would be useful to evaluate the proposed approach under different network topologies, dimensions, and traffic conditions to determine its effectiveness in different scenarios. While the proposed approach offers a new routing technique, it is unclear how it will be implemented in a real-world system. Therefore, future research may focus on the practical applicability of this approach and evaluate its performance in a real-world environment [58].

In 2023, Dogra et al. worked on a wireless intelligent router (WIR) in a network for efficient resource allocation and re-routing. Their purpose was to reduce both the power wasted and latency in the case of congestion. For this purpose, the reinforcement-based WIR, using the Q learning algorithm, monitors the network and enables users to re-route in case of congestion. It has been proven in simulation results that it reduces the average power consumption and latency. In addition, as the network grows, Q-table scalability can be a problem. The proposed model has been tested on a small network. The feasibility of a real-world version of the concept is not clear [88].

In 2023, Haseeb et al. put forward a steady and effective routing strategy for autonomous vehicles within 6G networks. Their aim was to provide a protocol that enables automatic routing decision-making and optimizes information delivery and transmission durations using IoT technology. For this purpose, they used computational intelligence in the model they presented. An assessment of the performance of the proposed method using the NS-2 simulation environment demonstrated superior results regarding data transmission, reception, loss rate, latency, and energy efficiency compared to the currently available approaches. The study can be improved with real data and machine learning approaches [89].

3. Routing in Satellite Networks

One of the greatest potentials of satellite networks is the advantage of being able to integrate ground and air networks. This is especially important in rural areas, areas with insufficient infrastructure, and natural disaster areas. Integrating this potential of satellite networks with developing mobile band technologies will produce exceptional outputs at the point of access anywhere at any time.

3.1. Satellite Networks

The satellite acts as a relay located in space, receiving signals sent from the Earth and sending back the signal it received. Satellite networks, on the other hand, communicate via satellite. Satellite networks consist of nodes that provide data transfer between two points on Earth and transmit data at high speeds, such as GigaHertz. The classification of satellite orbits is based on several parameters. One of these parameters is the orbit that the satellites follow. According to the orbit they follow, satellites use circular and elliptical orbits. In those that have a circular orbit, the midpoint of the globe is the same as the center point of the orbit they follow. In elliptical versions, the center point of the Earth is the same as one of the focal points of the ellipse. Another classification parameter is the classification according to the plane of the orbit that the satellites follow relative to the Earth. In this classification, it can be located in the equatorial plane, in the plane of the poles, or have a certain angle. The third classification metric sees satellites divided into three classes with respect to their interval and the globe: low Earth orbit (LEO), medium Earth orbit (MEO), and geostationary Earth orbit (GEO). LEO satellites have altitudes between 500–2000 kilometers (km) from the Earth. Their coverage area is approximately 8000 km and is narrow compared to other satellites. They have the lowest latency compared to other satellite networks. MEO satellites are located at an interval of 5000–15,000 km from the globe. They are widely used in GPS applications. GEO satellites are situated about 36,000 km from the globe, and these satellites move at the same speed as the Earth. For this reason, the point at which they are located on the Earth is fixed. Three GEO satellites can cover the entire globe. They are the satellites with the greatest delay due to the altitude at which they are located [90,91,92,93,94,95].

In the context of integrating satellite communication with 6G networks, it is crucial to compare LEO, MEO, and GEO satellites, particularly in terms of latency and instantaneous positioning [92]. LEO satellites, typically orbiting at altitudes of 500–2000 km, offer the lowest latency (around 20–40 ms) due to their proximity to Earth, making them suitable for real-time applications. However, their fast movement relative to the Earth’s surface requires a large constellation and frequent handovers to ensure continuous coverage. MEO satellites, located at altitudes of 2000–20,000 km, provide moderate latency (around 100–150 ms) and a balance between coverage area and mobility. GEO satellites, positioned at approximately 35,786 km, offer wide coverage and fixed positioning relative to the Earth, simplifying ground station tracking. However, they suffer from high latency (around 500–600 ms), which can be problematic for latency-sensitive applications. Each orbit class presents trade-offs between coverage, latency, and infrastructure complexity, and understanding these differences is key to optimizing satellite-augmented 6G networks.

3.2. Routing Protocols for Satellite Networks

Satellite networks are among the communication technologies whose prevalence is increasing due to advantages such as wide coverage area, providing access to areas where communication is limited, providing service in times of natural disasters, providing continuous service, and accessibility to areas with insufficient infrastructure. Nowadays, satellite technologies are widely used in many fields, such as meteorological services, navigation systems, internet access, Global Positioning System (GPS) applications, agricultural field traceability, and product quality determination. The use of satellite networks is increasing steadily. The Starlink satellite, which was placed in space by SpaceX (a private spaceflight company) to provide internet access, has already passed 2000 units. It is aimed to increase this number to 42,000 [96]. With the increasing use of satellite networks, it is also of great importance that they can be integrated with new, emerging technologies. Routing approaches play a critical role in order to ensure high velocity and low latency, which is one of the most basic features of the next epoch of communication technologies. Concerning the framework of a satellite network system, satellite network routing can be categorized into three components [97]:

• Boundary routing;
• Access routing;
• Inter-satellite routing.

3.2.1. Boundary Routing

The BGP-satellite (BGP-S) version is a recommended protocol for the smooth operation of IP connections in both networks during the consolidation of tellurian and satellite networks. It is a satellite variant of the traditional border gateway protocol. The BGP-S protocol was first introduced in [98], and the simulation results are shown in [99]. BGP-S is a unified routing protocol envisaged to work solely on a tellurian gateway in an autonomous system and transmits discovered routes using the BGP4 [100] protocol. Satellite networks are considered to be autonomous systems in terrestrial networks. However, autonomous systems in satellite networks have their own characteristics; for example, the delay is greater than that of autonomous systems in terrestrial networks [101].

3.2.2. Access Routing

Access routing is accountable for ensuring communication between the terrestrial gateway and LEO, MEO, and GEO satellites according to various parameters. The parameters that access routing, also known as User Data Link (UDL), consider time in providing communication: the lifetime, delay, and signal robustness of UDL [97,102,103].

3.2.3. Inter-Satellite Routing

When data packets are transmitted to the satellite network, the inter-satellite link (ISL) acts as an auxiliary tool in finding the appropriate trail from the origin point to the goal point for a robust communication inter-satellite setup.

3.3. Routing Protocols in LEOs

In satellite networks, which exhibit significantly different dynamics compared to ground networks, routing protocols designed for terrestrial systems often struggle to operate at peak efficiency. The techniques suggested for handling the distributions of routing load in satellite networks are outlined below.

3.3.1. Asynchronous Transfer Mode (ATM)

ATM is a network technique in which services such as audio, video, and data are transmitted in small, 53-byte stabled-size cells. ATM is an International Telecommunication Union-Telecommunication Standardization Sector (ITU-T) standard. It is a switching and multiplexing technique that blends packet switching and circuit switching for data transmission. It obtains the volume and stationary transmission delay avails of circuit switching. Packet switching, on the other hand, receives the benefits of elasticity and productivity for discontinuous traffic. ATM can be called the next generation of packet switching. ATM creates virtual circuits to blend these two techniques and was originally designed for terrestrial networks. Satellite networks use time division multiplexing (TDM) technology. TDM technology addresses time division, and each time division is assigned to one user. No other user can transmit data in the time zone assigned to this user. This means that if the user using it in that time period has no data to send, the time slot is occupied unnecessarily. Because ATM technology is asynchronous, time zones are used more flexibly. In other words, asynchronous ATM technology is more productive than synchronic techniques such as TDM. For this reason, it will be beneficial for use in satellite transmission systems. Since ATM networks are connection-oriented, two points must establish a connection before they can exchange data. For this, the connection setup package is sent. Thus, a virtual circuit is created [104,105].

The key advantage of ATM over TDM lies in its efficient use of bandwidth and support for dynamic traffic patterns. In TDM systems, time slots are pre-assigned to users regardless of whether they have data to transmit, leading to potential inefficiencies when some users remain idle [104,105]. ATM, in contrast, does not reserve fixed time slots for each user. Instead, it transmits data in small, fixed-size cells only when there is actual data to send. This asynchronous nature allows ATM to dynamically allocate bandwidth based on demand, minimizing idle time and maximizing channel utilization. Furthermore, ATM is more suitable for heterogeneous data types (e.g., voice, video, and data) due to its ability to handle variable bit-rate traffic with low latency and high quality of service (QoS). This flexibility and efficiency make ATM particularly advantageous for satellite networks, where bandwidth is a limited and costly resource.

3.3.2. A Finite State Automaton (FSA) Routing Algorithm

In this algorithm, the system period of the LEO satellite network is classified according to certain time periods. These classification ranges represent a state. Each situation of the FSA also matches these intervals. For this reason, the FSA algorithm is a connection-oriented routing algorithm. Each state is considered to have a fixed topology. Thanks to this structure, the best way is determined. The advantage of FSA is that it stimulates the satellite constellation as a static network within a stationary interval of time. Thanks to this, the problem of assigning connections in LEO satellite networks can be simplified [58,84,85,86,87,88].

3.3.3. Predictive Routing Protocol (PRP)

PRP is a recommended routing protocol for providing deterministic QoS guarantees, such as delay jitter in LEO systems. The fact that satellites have different characteristic structures than ground networks, such as limited onboard processing capabilities, makes it mandatory to use ground gateways in the routing process. Gateways perform the task of storing network information, such as the available bandwidth of connections between satellites, the location of users, and traffic patterns. The PRP protocol uses the predictable information of the satellites and the user location information to predict the traffic load that will be between inter-satellite connections in the future for a restricted interval of time. Thanks to this forecast, it determines multiple alternative road recommendations so that congestion can be avoided in the future. The purpose of this routing protocol is to maintain both QoS requirements and maximize the total number of users [106].

3.3.4. Explicit Load-Balancing Routing Protocol (ELB)

The rate of satellite use in space is increasing rapidly every day. We can predict that it will become one of the key legs of communication in the future. For this reason, traffic management in satellite networks needs to be directed intelligently from now on. This is where the proposed ELB protocol comes into play. The purpose of the ELB protocol is to obtain congestion information of neighboring satellites. In this way, intelligent communication can be achieved between satellites. A satellite that will soon increase congestion may ask its contiguous satellites for data reduction transmission rates. In reply to this request, the contiguous satellites take the action of searching for new, less congested routes and transmitting data over these routes. The purpose of this protocol is to reduce packet loss, prevent congestion, and achieve higher efficiency [107].

3.3.5. Dynamic Detection Routing Algorithm (DDRA)

DDRA is a routing protocol that relies upon an abstract topology snapshot. In the DDRA algorithm, the satellite interval is broken down into n time pieces. The satellite topology in the slices can be presented by one snapshot. Thanks to the DDRA algorithm, low time delay is achieved, and the ability to adjust to shifting network conditions is increased [108].

Standard satellite routing algorithms also encompass Traffic Prediction Distributed Routing Algorithm (TPDRA) [109], Control Route Transmission (CRT) [110], Compact explicit multi-path routing (CEMR) [111], Adaptive Load Balanced Routing (ALBR) [112], Dynamic routing algorithm (DRA) [113], Cross-Entropy Accelerated Ant Routing System (CEAARS) [114], etc. The literature on routing protocols in satellite networks is given in

Table 3. Summary of existing survey articles on routing approaches in satellite networks (1—6G focused; 2—Satellite focused; 3—Integration of 6G and Satellite focused; 4—Taxonomy available).

Ref.	Year	Objective	Merits	Demerits	1	2	3	4
[113]	2001	- Proposes a distributed datagram routing algorithm for LEO satellite networks that minimizes propagation delay, avoids overhead, and prevents congestion	- It has shown successful performance in both congestion avoidance and routing of packets in case of any failure	- Should a satellite or ISLs malfunction occur, the effectiveness of the recommended algorithm would experience a significant decline	✗	✓	✗	✗
[115]	2002	- A routing model is proposed that is similar to ATM switching and takes into account the memory requirements of the satellites	- Lighten the memory burden of satellites	- Sensitivity to communication between ground and satellite	✗	✓	✗	✗
[112]	2010	- A decentralized routing approach called agent-based load-balancing routing (ALBR) is introduced, designed specifically for LEO satellite networks	- The proposed algorithm has been proven to provide better load balancing	- It can be enhanced to increase compatibility with multilayer satellite networks	✗	✓	✗	✗
[109]	2011	- TPDRA, a dynamic, distributed, and adaptive routing algorithm, is proposed to address the poor adaptability of centralized, static, and non-adaptive routing algorithms in LEO satellite networks	- Improved performance regarding conduction delay and congestion when compared to ACO		✗	✓	✗	✗
[116]	2019	- An adaptive satellite communication routing algorithm based on SDN architecture is proposed	- Promising in finding the shortest path and optimizing that path in real time according to satellite movements	- Appropriate for limited-scale networks rather than extensive, dynamic networks	✗	✓	✗	✗
[117]	2019	- A new and memory-efficient routing approach, ’OPSPF’, is proposed	- The latency is effectively reduced, and there is no requirement to gather global information and incur large computational costs	- The algorithm is only suitable for a certain small network and cannot optimize the connection latency	✗	✓	✗	✗
[118]	2021	- Optimizing and improving traditional and centralized routing processes in LEO satellite networks - They proposed a DQN-based intelligent routing (DQN-IR) algorithm	- Better delay performance compared to traditional routing algorithms	- Poor scalability for real scenarios	✗	✓	✗	✗
[119]	2022	- They worked on load-balancing for LEO satellite networks - An ant colony optimization routing algorithm with window reduction (ACORA-WR) was proposed	- Better performance in terms of data delivery rate, average delay, throughput, and transmission overhead	- More information could be provided on the applicability of the proposed methods in real-world scenarios	✗	✓	✗	✗
[120]	2022	- The study of routing in large-scale low Earth orbit (LEO) satellite networks is provided	- A delay close to the minimum delay in real scenarios was obtained	- Only latency performance is considered - The article’s reliability is limited to a particular algorithm	✗	✓	✗	✗

.

In 2001, Ekici et al. worked on a routing algorithm for LEO satellite networks. Satellite networks have their own characteristics, and the interconnection pattern creates different shapes depending on their movements. Depending on the movements of satellite networks in different planes, their distances are also different, which causes the network topology to change. For these reasons, it is not enough to choose the best path from source to destination in satellite networks; it is necessary to maintain this path throughout the communication. The authors worked on more efficient and effective routing models for this purpose. They propose a distributed datagram routing algorithm that creates minimal propagation delay paths, does not cause overhead, and avoids congestion. The proposed system is distributed, with each data packet’s path being determined individually. The proposed algorithm has shown successful performance in both congestion avoidance and routing of packets in case of any failure. Should a satellite or ISL malfunction occur, the effectiveness of the recommended algorithm would experience a significant decline [113].

In 2002, Gounder et al. focused on creating a proficient resolution for the routing problem within a low Earth orbit (LEO) network. They also addressed the matter of situating the necessary Network Operations and Control Centers (NOCCs) to facilitate the satellite system. In satellite networks, both IP routing and ATM routing are not efficient due to large overheads, inefficient use of bandwidth, and the inability to handle a vast number of routing tables. Since the models arising from the integration of such models have similar problems, a routing model is proposed that is similar to ATM switching and factors in accounting for the memory requirements of the satellites. In the proposed model, the arrangement of the network is portrayed via a collection of snapshots of its topologies. The ground stations pre-calculate the snapshot order and update it using the unpredictable changes detected by the satellites. If the link between satellites and ground stations is severed, the proposed model cannot be executed [115].

In 2010, a routing algorithm for efficient load balancing over satellite networks was studied. Due to the uneven population distribution, some of the satellites of the satellite networks suffer from congestion, and some suffer from underutilization. This problem is becoming more important, especially in LEO satellite networks with low altitudes. Without an effective routing algorithm, the unequal allocation of network traffic can lead to considerable queue delays and massive packet loss in high-traffic areas. This article introduces a decentralized routing approach called agent-based load balancing routing (ALBR), designed specifically for LEO satellite networks. ALBR, which aims to guarantee lower packet loss, better efficiency, and end-to-end latency limits, uses two types of agents: stationary agents and mobile agents. The task of the station agent is to estimate the cost of the ISL (inter-satellite link) on the satellite and update the routing elements.

The task of the mobile agent is to discover the routing path and collect routing information. The proposed algorithm has been proven to provide better load-balancing. However, while it is an excellent option for single-layer satellite networks, it can be enhanced to increase compatibility with multilayer satellite networks [112].

In 2011, Zihe et al. proposed a new dynamic, distributed, and adaptive routing algorithm to overcome the problem of poor adaptability of centralized, static, and non-adaptive routing algorithms in LEO satellite networks to frequently changing situations and traffic loads. While the proposed TPDRA algorithm makes prediction decisions with the data collected from the ground networks, it also provides the routing decision using the mobile agent on the satellite network side. For traffic prediction, the traffic density of the Earth’s surface, which they divide into grids and classify according to population density, is estimated through a radial basis function neural network. On the side of the satellite networks, the final routing decision is made based on both the current and future status of the satellite. It was compared to ACO in the simulation results and showed better performance regarding conduction delay and congestion [109].

In 2019, Wang et al. centered on refining routing algorithms on LEO satellite networks. The fluctuating topology of LEO satellite networks causes difficulties in routing algorithm design. They recommended an adaptive satellite communication routing algorithm based on SDN architecture for communication with good communication quality. The proposed model is promising both in finding the shortest path and optimizing this path instantly, adjusting to the movement of satellites. Nonetheless, owing to the heightened intricacy of the algorithm, this approach is appropriate for limited-scale networks rather than extensive, dynamic networks [116].

In 2019, Pan et al. focused on memory-efficient routing approaches in LEO networks. During the movements of LEO satellite networks, the topology of the satellite network changes periodically. One of the challenges posed by this situation is how to construct a fault-tolerant and durable routing protocol. For this, a new and memory-conserving routing technique, ’OPSPF’, which takes the upside of the predictability of the satellite array, is proposed in the study. The OPSPF routing approach provides zero route convergence expense amid frequent topology modifications. It achieves a 57% reduction in communication costs amid infrequent, unpredictable topology shifts. It also achieves an 82% reduction in route convergence time during erratic topology changes. These advantages show that OPSPF is a more efficient and cost-effective routing protocol than conventional OSPF. Despite this, the algorithm is only effective for a certain small network with eight satellites, and the algorithm cannot optimize connection latency [117].

In 2021, Zuo et al. worked on routing processes in the LEO satellite network. Their goals were to optimize and improve traditional and centralized routing processes in LEO satellite networks with high node mobility, topology change, and connection congestion, which will hold significant importance in 6G networks. For this purpose, they proposed an intelligent decentralized routing algorithm based on deep reinforcement for LEO networks. The proposed algorithm uses the DQN table for decision-making. The algorithm receives data, such as the current channel bandwidth, the signal-to-noise ratio (SNR) of the channel, spacing, queue delay, and spatial location, that it collects from its neighbors as input. With the received data and its own state data, the DQN model gives the predicted output. In the simulation results of the model trained in the Keras environment, it was seen that the latency improved compared to traditional algorithms. However, it is necessary to expand the study with different deep learning networks in terms of feasibility for real scenarios. DQN tables have scalability issues as the network grows [118].

In 2022, Deng et al. worked on load-balancing for LEO satellite networks. In light of the uneven distribution of the world population and the continuous movement of the satellites, an imbalanced traffic load occurs on the satellites. The aim of this investigation is to produce a remedy by using the ant colony optimization algorithm, which has a distributed routing structure, to solve the imbalanced traffic problem in LEO networks. To improve the performance and reduce the complexity of an ant colony optimization routing algorithm with window reduction (ACORA-WR), they also used shrinking windows to restrict the scope that the algorithm will search. They showed the LEO satellite network as a graph, as in Figure 5, and represented it with $G=(V,E)$. V represents each satellite or node, and E represents the connections (ISLs). Here, $N_{i,j}$ depicts the logical address of the satellite, and i and j represent the orbit number and the satellite number in its orbit, respectively. They compared the efficiency of the introduced algorithm with LBRA-CP, SPR, and LCRA by modeling the iridium-like system in the NS2 simulation environment. In the simulation results, the proposed model performed well with regard to data delivery rate, average delay, throughput, and transmission overhead [119].

In 2022, a study of routing in large-scale low Earth orbit (LEO) satellite networks was conducted. The authors’ goal concerned how to provide minimum latency routing in time-varying topologies where the number and position of satellites are continuously evolving. For this, they focused on stochastic geometry-based approaches, which are more suitable and powerful for dynamic large-scale satellite constellations. This is the inaugural investigation of satellite routing utilizing stochastic geometry. They obtained output results close to the minimum delay in real scenarios. The assessment of this article’s reliability is limited to a particular algorithm and does not offer an analytical formulation for the probability of disruption, and only latency performance is considered [120].

4. Routing in Integrated Networks

In combined land-based and satellite networks, routing is essential for achieving streamlined and trustworthy communication between users and services. By choosing the best network pathways, routing algorithms help balance resource utilization and reduce latency while addressing the specific challenges posed by merging ground and satellite systems. The dynamic nature of LEO satellite networks, such as changing satellite positions and varying link conditions, adds complexity to the routing problem, making it essential to develop adaptive algorithms that can effectively respond to these fluctuations. As the desire for high-quality communication continues to grow, the development of innovative routing strategies for integrated networks becomes increasingly important to ensure optimal performance and user satisfaction. The literature on routing protocols in integrated networks is given in

Table 4. Summary of existing survey articles on routing approaches in integrated networks (1—6G focused; 2—Satellite focused; 3—Integration of 6G and Satellite focused; 4—Taxonomy available; 5—Routing).

Ref.	Year	Objective	Merits	Demerits	1	2	3	4	5
[121]	2018	- They focused on routing in DTN-Nanosatellite networks	- They suggested an innovative energy-aware routing algorithm derived from the Contact Graph Routing (CGR) named E-CGR - Better average data transmission time - The amount of transmitted data beams increases	- The evaluated variables can be expanded	✗	✓	✗	✗	✓
[122]	2021	- They focused on the study of point-to-point (P2P) connections for integrated satellite, high-altitude platform (HAP) networks, which is one of the key elements of the sixth generation (6G) wireless network vision	- It is the first of its kind to examine P2P connections for multilayer spatial networks from the perspective of 6G large-scale complex networks	- The advantages and disadvantages of different techniques and approaches can be extended	✓	✓	✓	✓	✗
[123]	2022	- It is the first study to examine the varying heights and minimum elevation angles of hot air balloons	- A model more suited to real-world conditions - An efficient approach to energy efficiency	- It can be extended with consideration of asymmetric time windows scenarios	✓	✓	✓	✗	✗
[124]	2022	- An effective network control and management framework has been proposed for the ultra-dense LEO satellite–ground integrated network	- Effective and efficient network management - It reduces the complexity of management	- It needs to be tested in real application situations	✓	✓	✓	✗	✗
[125]	2022	- They propose a machine learning algorithm that optimizes path selection by dynamically adapting the resources of satellite and land networks to environmental changes using network slicing	- It performs both routing and slice management - Doppler effect, atmospheric loss, and other factors are taken into account	- It can be improved by studying more satellite and base station scenarios, advanced machine learning algorithms, and different traffic profiles	✓	✓	✓	✗	✗
[126]	2024	- The goal is to improve internet access in remote areas by enabling the integration of LEO satellite communication and mobile edge computing (MEC). To tackle the issues posed by the high mobility of LEO satellites, a dynamic computation offloading and resource allocation framework (DCOOL) has been developed	- DCOOL performs better than other algorithms by reducing power consumption and latency, and it is particularly effective in low-latency scenarios	- Weaknesses of the study include the limitations of the simulations to a single satellite, the modeling of only specific frequency bands (C and Ka), and the use of a fixed noise level. Furthermore, the complexity of Lyapunov optimization may present challenges in real-world implementations	✗	✓	✗	✗	✓

.

In 2018, Marchese et al. focused on routing in DTN-Nanosatellite networks. Their aim was to alleviate the limitations of nano-satellites, including capacity for storage and energy supply, and to increase their lifetime and performance. At this point, they suggested an innovative energy-sensitive routing algorithm derived from Contact Graph Routing (CGR) named E-CGR. They used the Contact Graph Routing (CGR) algorithm, which is most compatible with the predictable conditions of satellite movements in DTN networks that store data when there is no suitable path. In this algorithm, when the ground networks receive the data to be transmitted, they store the data in their memory before transmitting it. When communication starts, the satellite network sends a bundle to the ground network, which indicates the current energy level, called the energy bundle. With this data, it is understood whether the energy of the satellite network guarantees the completion of the routing process. In their performance analysis in the Network Simulator 3 (NS3) environment, the results using E-CGR and standard CGR were compared based on two evaluation indicators: average time to delivery (ADT) and percent of packages delivered (PDB). The results obtained from the E-CGR algorithm show that the average data transmission time decreases, and the amount of data beams delivered to the destination increases compared to classical CGR. However, the variables that can affect the routing process can be expanded further [121].

In 2021, an extensive literature review was provided, focusing on the study of point-to-point (P2P) connections for integrated satellite high-altitude platform (HAP) networks, which are considered one of the key elements of the sixth generation (6G) wireless network vision. The aim of the article was to study point-to-point connections in multilayer spatial networks, which are expected to play an important role in 6G large-scale complex networks, and present their integration and upcoming research pathways. In this context, it is emphasized that the coverage, reliability, and scalability of 6G wireless networks can be increased by developing integrated spatial networks containing spatial nodes at different altitudes, such as satellites, high-altitude platforms (HAPs), and low-altitude platforms (LAPs). It details RF and FSO solutions used to link single or orbiting satellites and provides a link budget evaluation for RF- and FSO-based satellite-to-satellite connections. It was underlined that artificial intelligence, SDN technology, as seen in Figure 6, the use of intelligent reflecting surfaces, gateway selection, and innovative routing protocols will serve a crucial role in resource management and traffic routing optimization, which is critical due to the multilayered, heterogeneous structure of integrated spatial networks. However, the merits and demerits of various methods and approaches can be compared in more detail [122].

In 2021, there was a focus on the terrestrial-satellite network (TSN) model in 6G networks that utilize hot air balloons at varying altitudes and minimum angles of elevation as intermediaries between satellite and ground stations. The primary obstacle to this is enhancing the network’s energy performance by combining the management of caching, computing, and communication (3C) resources for both uplinks and inter-satellite relay latency and structured laser links per satellite in terrestrial-satellite networks (TSNs). Modeling a TSN system that includes hot air balloons floating at various altitudes and minimum angles of elevation, considering laser links and the traffic between satellites, was considered to be more suitable for real-world scenarios. First, to model hot air balloons at various heights and elevation angles, researchers proposed an algorithm that determines multiple sub-traffic-matrices (STMs) using the water-filling method for a traffic matrix, representing the amount of traffic transferred between satellites. They then forwarded the data traffic indicated by the STMs to the appropriate target satellites, utilizing a carefully crafted configuration matrix generation algorithm. Finally, considering the constraints, such as the number of lasers and traffic delay, they obtained the optimal parameters to enhance the system’s energy efficiency by using geometric programming and the Taylor series approach. The simulation results confirm the efficiency of the proposed approach. In addition, the scope of the study can be expanded by taking into account the scenarios of asymmetric time windows [123].

In 2022, Ma et al. focused on an effective network governance and administration architecture for the ultra-dense LEO satellite–ground integrated network. This aimed to provide a more efficient and effective network control and management method for the governance and administration of the network, which is quite complex due to the natural heterogeneous characteristic features, high dynamics, frequent changes in network situations over time, and the large number and density of satellites of ultra-dense LEO satellite–ground integrated networks. A hybrid multilayer network control and management framework was proposed for the ultra-dense LEO satellite–ground integrated network. This architecture includes the grouping and clustering methods of LEO nodes, and for each group, the MEO satellites and (CH) LEO satellites are designated as global and local controllers. In addition, different components, such as mobility management, network status control, resource management, and service management, are also included.

The suggested hybrid network governance and administration architecture provides an effective management approach to the ultra-dense LEO satellite–ground integrated network. In particular, the identification of global and local controllers using grouping and clustering methods is a suitable solution for managing the large-scale and complex structure of the network. However, since the proposed method has not been tested with real-time data and in real application situations, its performance in real application situations is not clearly known [124].

In 2022, researchers focused on network slicing for unified satellite and ground networks in 6G networks. Given that computer networks are inherently dynamic and that there will be a growing number of connected devices in 6G, the management of satellite networks and network segments requires dynamic solutions. In one study, the researchers also used network slicing, which allows a physical network to be divided into multiple logical networks (slices) to assign resources and customize each slice to meet different service requirements and user demands. The proposed intelligent framework combines the resources of satellite and ground networks using network slicing and optimizes path selection with a machine learning algorithm similar to ant colony optimization, which can dynamically adapt to environmental changes. Using a centralized and distributed control plane, it performs both routing and slice management according to user demands. The performance results show that the proposed network slicing framework provides more efficient resource utilization than other strategies by accepting more user requests and saving priority routes for services that need them. The work can be improved by studying more satellite and base station scenarios, advanced machine learning algorithms, and different traffic profiles [125].

In 2024, Kim et al. focused on the integration of LEO satellite communication and mobile edge computing (MEC) to improve internet access in remote areas. To mitigate the obstacles presented by the fast speed of LEO satellites, they present an agile approach to computation offloading and a resource distribution framework called DCOOL based on Lyapunov optimization, which aims to minimize power consumption and latency. As a method, the Lyapunov drift-plus-penalty technique is used to reformulate the stochastic optimization problem into an immediate optimization challenge. UE-GW communication in the C-band and GW-LEO communication in the Ka-band were modeled. Simulations were performed using a single satellite orbiting at an altitude of 550 km with an orbital velocity of 7.66 km/s. The level of background noise was set to −174 dBm/Hz. The simulations demonstrate that the DCOOL algorithm outperforms other algorithms (Selfish users, Offloading-DVFS, and Offloading-fixed frequency), providing lower power consumption and stable workload processing. DCOOL is particularly effective in the low latency regime [126].

5. Addressing Concerns and Implications of AI Algorithms in Network Routing

5.1. Comprehensive Analysis of AI Algorithms in Network Routing

The use of AI-based algorithms in network routing has gained notable focus in recent times, particularly in the advent of next-generation networks. AI algorithms have the ability to enhance network routing, improve network efficiency, and enhance the overall quality of service. Nonetheless, for any innovation, AI algorithms in network routing also have their advantages and disadvantages. The advantages of implementing AI in network routing are as follows:

• Improved network efficiency: AI algorithms can analyze network traffic patterns, optimize routing decisions, and reduce congestion, resulting in improved network efficiency and reduced latency;
• Enhanced scalability: AI algorithms can handle large volumes of data and scale to meet the demands of growing networks, making them ideal for next-generation networks;
• Real-time optimization: AI algorithms can optimize routing decisions in real time, adapting to evolving network conditions and ensuring optimal performance;
• Increased security: AI algorithms can identify and address security vulnerabilities in real time, improving network protection and limiting the risk of cyber-attacks.
• Reduced operational costs: AI algorithms can automate network management tasks, reducing operational costs and improving network reliability.

The disadvantages of AI algorithms in network routing are as follows:

• Complexity: AI algorithms can be intricate and challenging to implement, necessitating significant expertise and resources;
• Data quality: The effectiveness of AI algorithms hinges on quality data, which might not always be available, especially in networks with restricted visibility or incomplete data sets;
• Lack of transparency: AI algorithms can be difficult to interpret, complicating the understanding of how they make decisions and detect possible biases;
• Dependence on training data: AI algorithms are only as good as the data used to train them, and biased or incomplete training data can lead to suboptimal performance;
• Security risks: AI algorithms can introduce new security risks, including the possibility of AI-powered attacks or manipulation of AI-driven decision-making processes.

The barriers to effective AI-driven processes in network routing for decision-making are as follows:

• Lack of human oversight: AI algorithms may not always understand the context and nuances of network operations, leading to suboptimal decisions;
• Inability to handle unforeseen events: AI algorithms may struggle to respond to unforeseen events or abrupt changes in network behavior;
• Dependence on algorithmic design: AI algorithms are only as good as their design, and poorly designed algorithms can lead to suboptimal performance;
• Explainability and accountability: AI algorithms can be difficult to explain and hold accountable, making it challenging to identify and address potential biases or errors.

AI algorithms have the potential to revolutionize network routing, but it is essential to carefully consider their advantages and disadvantages, as well as their limitations for decision-making. By understanding these elements, professionals and researchers in networking can design and implement AI algorithms that optimize network performance, improve efficiency, and enhance the overall quality of service.

5.2. Ethical and Social Implications of AI Algorithms in Network Routing

The use of AI algorithms in network routing raises several ethical and social implications that need to be carefully considered. As AI algorithms become increasingly autonomous and pervasive in network management, it is essential to address the potential consequences of their deployment.

Privacy and Data Protection

• Data collection and storage: AI algorithms necessitate large volumes of data for learning and enhancement, raising concerns regarding data privacy and protection. Network operators must guarantee that data is gathered, stored, and utilized in ways that protect individual privacy and comply with data protection laws.
• Biased decision-making: AI algorithms can reinforce biases that exist in the data used for their training, leading to discriminatory decision-making. Network operators must ensure that AI algorithms are designed to avoid biases and promote fairness.

Fairness and Equity

• Fair resource allocation: AI algorithms may prioritize certain users or applications over others, leading to unfair resource allocation. Network operators must ensure that AI algorithms allocate resources fairly and equitably.
• Accessibility and inclusivity: AI algorithms may not be accessible or usable by all individuals, particularly those with disabilities. Network operators must ensure that AI algorithms are designed to be accessible and inclusive.

Accountability and Transparency

• Explainability: AI algorithms can be difficult to interpret, complicating the understanding of their decision-making process. Network operators need to ensure that these algorithms are designed to provide transparency and explainable decision-making processes.
• Accountability: AI algorithms can make decisions that have significant consequences, and it is essential to establish error-handling and bias-mitigation frameworks.

Workforce Reduction and Skills

• Workforce reduction: The automation of network management tasks may lead to job displacement, notably affecting individuals in entry-level positions. Network operators must ensure that they promote upskilling to help displaced workers adapt to evolving job markets.
• Skills gap: The increasing reliance on AI algorithms may create a skills gap, particularly in areas such as AI innovation and implementation. Network operators must prioritize the development of skills that enable effective AI-driven network management.

Policymaking and Regulation

• Regulatory Frameworks: Policymakers must establish regulatory frameworks that address the ethical and social implications of AI algorithms in network routing.
• Industry Standards: It is crucial for industry stakeholders to establish guidelines for AI algorithm creation and implementation in network routing to ensure fairness, transparency, and accountability.

6. Future Directions

As it is often repeated, routing is really a process located at a vital point of communication. The fact that routing techniques can be managed with full efficiency is a revolution in communication. However, it has always been a difficult process to manage. Although there are existing and used routing protocols, there is a need for new routing approaches that can both better adapt to new technologies and better integrate. With increasing demand, there is an explosion in communication technologies. The demands on the user, the width of the coverage area, accessibility from anywhere, accessibility at any time, and the integration of 3+ areas into each other show that the importance of both 6G networks and satellite networks will increase rapidly. For this reason, there is no suspicion that 6G networks and satellite networks will not become the main hot topic areas of research in the future. In order for the networks that will be majorly considered to meet the needs, the topics of routing in 6G and satellite networks will also be one of the hot research topics. It is essential that routing networks contain the following parameters as a priority:

• Adaptability of the dynamic alteration of network anatomy: Both 6G networks and satellite networks are networks with a dynamic network topology. For this reason, routing algorithms need to be developed to adapt to this dynamic structure in order to ensure stability.
• Invulnerability: Both 6G networks and satellite networks are complicated networks. These networks can be exposed to various attacks. It is possible that this will remain the case in the future. In the future, it is envisaged that many processes will be carried out based on satellite networks. In order for the routing protocols to be prepared for possible adverse situations, survivability must be taken into account.
• High efficiency: Efficiency is becoming increasingly important in all areas. In order to better manage resources, improvements should be made to make the system work more efficiently.
• Meeting multiple QoS needs. The key commitments of 6G networks are end-to-end low latency and data transmission at high speeds, as well as mobility. So, the future networks are very diverse in terms of QoS. For this reason, routing protocols should also be able to cover this diversity.
• Adaptability to changes: Routing protocols need to be more visionary to meet the future. As of the 2000s, space studies have gained rapid momentum. Routing networks also need to be managed effectively and flexibly in the face of developments.

The methods and approaches that can meet the needs of networks that will require smarter, more efficient, sustainable, wide coverage, low latency, fast, and flexible routing in the future are detailed below.

6.1. AI/ML-Enabled Networks

Considering generation-to-generation cellular systems, sixth-generation cellular systems have become complicated in many ways. It is becoming increasingly difficult to manage this complex structure effectively with the old approaches developed for older generations. There is a need for an in-depth introduction of artificial intelligence technology and solutions, which is one of the latest developing areas, into the sixth-generation cellular systems. The biggest difference that distinguishes 6G from its previous generations is that it has the feature of being a platform for AI. One of the developments that awaits humanity in the future is that both networks and AI will feed each other, and even AI will be presented as “AI as a Service (AIaaS)” [37]. The vision of AI is that it acts as the engine for automation. Artificial intelligence technologies have a wide range of lower legs, such as machine learning, deep learning, deep reinforcement learning, computational intelligence [127], distributed learning [128], and federating learning [129]. Artificial intelligence algorithms are vital for the success of goals such as low latency, low path loss, and fast data transmission in the integration of 6G and space networks. In the future, AI will be a technology that is natively in everyone’s pocket rather than a technology that is only in the cloud. 6G aims to make access to AI possible from anywhere. The main focus of new mobile communication technologies is “intelligence”. It is envisaged that the part of intelligence will also be provided with technologies such as AI ML based on data rather than code. However, the fact that artificial intelligence technologies are data-based creates a limit for these technologies. 6G and space networks are dynamic and complex networks. Therefore, the provision of data collected from these networks as input in artificial intelligence technologies may adversely affect the performance of the output. For this reason, collective implementation of traditional and artificial intelligence-based solutions can provide better results. The storage and processing of huge amounts of data that will be collected from billions of devices in the future will be critical. Quantum computing is promising at the point of processing data [78]. In addition, the use of centrally distributed technologies for this data is promising for security. More research should be carried out to use deep reinforcement learning approaches that produce results by trial and error with the experiences gained from their environment rather than data on the communication side. Concepts such as federating learning also give hope to minimizing privacy violations.

6.2. High Mobile Experience

It would not be wrong to say that along with 6G and space networks, a digital universe circle will be entered. With technologies such as augmented reality, virtual reality, and extended reality combined with 6G networks, the reality of holograms will grow significantly. In addition, the tactile internet [130] and the haptic internet will ensure that all of the five human senses have a digital counterpart. In daily life, it is expected to experience a mobile experience surrounded by driverless vehicles and IoE technologies. In order for these applications to be provided efficiently, the requirements of low delays below milliseconds, high-speed data transmission, and high reliability must be met. Developments in technologies such as THz communication and beamforming are very important for the interests of the coalition of XR, augmented reality (AR), virtual reality (VR), haptics, tactile internet technologies, and 6G and satellite networks. Privacy is one of the priority issues for a person who is so surrounded by the mobile world. In order to achieve this, solution techniques, such as transfer learning, distributed learning, and federating learning, that keep data locally have great potential.

6.3. Sustainable Networks

Many resources are limited in nature, such as the RF spectrum. It is vital that all resources are used efficiently and are sustainable globally. The sustainable use of resources in the digital world is also very important. In a world experiencing a global energy crisis, the use of energy-efficient networks is, of course, one of the inevitable priorities. Resource management, power consumption, and the optimization of all processes in telecommunication networks will help to make the sustainability goals realistic. 6G aims to increase energy efficiency by 100 times (bits per Joule) and keep the total energy consumption lower than the previous mobile generation. In addition, improvements should also be made in technologies on both the hardware and software sides in order to make energy efficiency goals a reality. Finding innovative solutions for reduced power amplifier (PA) optimization becomes a challenge, as spectra such as THz using higher frequencies will be used [37]. There are various studies to increase sustainability in the digital world [131,132,133]. The studies conducted show that AI has a positive effect on 134 of the sustainability goals, that is, 79% [134]. Thanks to the use of AI technologies, the use of energy resources at a more optimal level acts as a catalyst for the realization of sustainable development goals such as lower carbon cities. However, the increase in data providing resources in AI is increasing logarithmically every two–three months. This increasing data potential also brings new challenges to be overcome, such as compressing, storing, and processing. In order to develop more environmentally friendly and human-friendly technologies in the future, there is a need for deep studies on resources, energy, traffic, routing approaches, data compression, processing, and the development of new hardware in 6G and satellite networks [78].

6.4. Ubiquitous Global Coverage

One of the most exciting points is that 6G networks offer an integrated coverage area with space–air–underwater terrestrial networks. It is expected to enable communication from anywhere at any time with its wide coverage area. In this way, it will facilitate communication, especially in rural areas and disaster areas, where communication is difficult. Communication, regardless of place and time, also supports sustainable development goals by reducing many inequalities. It is better understood how important and essential digital technologies are, especially in cases of global epidemics and pandemics. Every individual’s access to communication networks from anywhere contributes to social equality. In addition, the digitalization of daily work is very promising in terms of sustainability by not causing a pause in the functioning of society in pandemic situations. Factors such as channel estimation and modeling, delay, and the Doppler effect at the point of seamless communication of multiple networks with each other are some of the challenges. The utilization of unmanned aerial vehicles as airborne base stations in integrated network management has been shown to increase coverage [135]. New approaches, techniques, and architectures are needed to optimize the integration of these areas.

6.5. New Antennas

One of the basic cornerstones of wireless systems is antenna and antenna technologies. In order for 6G technology to fulfill its promises, breakthroughs in antenna technologies will be needed. 6G networks will be technologies that will provide services at high frequencies and wide bandwidths. High-frequency bands and the mmWave wave frequencies are affected by gases in the atmosphere and undergo further attenuation and path loss [136]. To compensate for this situation, antennas must have high gain. High power supplies and highly sensitive receivers are needed. It is necessary to study antennas and antenna architectures for tolerating path losses and achieving high gains and wide coverage. MIMO technology was introduced, which was developed as a way to increase the distance of the THz spectrum in 6G networks [137,138,139,140].

6G networks have a broad range of uses and scenarios. According to all these usage scenarios, antenna needs will be different. The general antenna specifications needed in 6G networks are as follows:

• Cost-effective [38];
• Antenna types suitable for different 6G use cases, including omnidirectional antennas for broad-area coverage, highly directional antennas for focused, point-to-point communication, and high-gain antennas to compensate for signal losses at THz frequencies [38];
• The communication spectrum is the THz frequency band (0.1–10 THz);
• Ultra-wide band and multi-band antennas [38];
• For communication at 6G frequencies, a circular, polarized antenna is preferred, which receives signals in a horizontal and vertical plane [141].

However, developments in silicon technology, decreasing costs, and the development of materials in a more compact form also make it possible to obtain more integrated circuits in antenna designs [37]. In 6G, new antennas and new material technologies that will facilitate their deployment will continue to develop.

6.6. New Channel

Measuring radio propagation and channel modeling represent the foundation of communication systems. A very good understanding and knowledge of the channel makes it possible to carry out communication smoothly and softly. However, since the channel is not sheltered and unpredictable, it is difficult to provide channel optimization. This is one of the main focuses in the communication sector and one of the main topics that needs to be improved. New challenges are inevitable in 6G communication with new spectra, new technologies, user and data explosions, new fields, and new antenna integrations. Exceeding the frequency band beyond 100 GHz increases attenuation and path loss. To compensate for this loss, more advanced beamforming technologies will be needed [37]. New channel models will be needed for scenarios such as the integration of terrestrial and non-terrestrial networks, as well as the use of drones to act as mobile base stations. It may not be possible to respond to such a variety of scenarios with a single-channel model. It may be useful to investigate hybrid channel models according to content usage scenarios [37].

Recent advancements in non-terrestrial networks have been significantly shaped by large-scale initiatives such as Starlink, OneWeb, and Amazon’s Kuiper. These systems demonstrate the feasibility and growing role of LEO constellations in global broadband connectivity [37,142]. Although primarily designed for current-generation networks, their architecture, deployment models, and inter-satellite communication mechanisms provide valuable insight into the future realization of 6G satellite networks. Lessons learned from these commercial systems—such as phased-array antenna use, laser-based ISLs, and latency challenges—can inform the integration of terrestrial and space-based 6G infrastructures.

The deployment of 6G networks in satellite domains will require navigating complex international regulatory frameworks involving spectrum allocation, orbital slot coordination, and cross-border data governance. With the growing density of satellite constellations, issues such as orbital congestion, interference management, and space debris mitigation have become critical. Collaborative efforts among global regulatory bodies like the ITU, FCC, and national space agencies will be essential to establish policies that support secure, fair, and sustainable 6G satellite operations.

6.7. Security and Secrecy

Due to the coverage areas of 6G and satellite networks, the number of places and users they can reach, and the scale of their services covering everywhere, security and secrecy requirements are also increasing. However, ensuring secrecy and security measures is becoming more difficult. Thanks to the full effective deployment of 6G and space networks, their boundaries are likely to be unpredictably wide. It is envisaged that networks at this scale will be designed with intelligent systems, such as AI, and decentralized systems, such as blockchain [56]. These systems are relatively new systems. It is necessary to expand the scope of the studies conducted to ensure security and confidentiality in these systems. In addition, it has not been clearly defined how measures should be taken if the potential possessed by AI systems is used as a counterattack. Data on how to prevent security and secrecy violations at the point of using the stored data in AI algorithms are also very limited. AI acts as both a poison and an antidote. These technologies, which will inevitably take their place in life, need innovative solutions at the point of data privacy. Quantum encryption methods can be used to protect security and secrecy in next-generation networks [68].

In 2018, Tang et al. worked on the integration of deep learning networks into networks for traffic control and routing management. In this study, the authors aimed to produce actions by evaluating historical data with deep learning methods. Traditional network routing protocols are unable to analyze historical data and generate actions based on it. The application of deep learning techniques to process this data and manage network traffic has become a trending area of research. In this research, they suggest a new routing technique that they propose to employ in future wired and wireless networks. They present a real-time deep learning-based smart network traffic control technique employing Deep Convolutional Neural Networks (deep CNNs) [143].

In 2020, Kato et al. conducted research on machine learning techniques to be used in 6G networks. In this study, the difficulties that may be encountered in the integration of machine learning algorithms with 6G networks are examined. They presented the study from a broad perspective. They examined the challenges of machine learning in 6G applications under 10 headings from the perspectives of communication, learning, and computing [144].

In 2020, Tan et al. proposed a load balance algorithm called “reliable intelligent routing mechanism (RIRM)” to manage congestion in 5GC (5G Core) networks. This algorithm aims to select the shortest path to adjust the load balance on the network. To implement this algorithm, they added an RIRM traffic tracker that monitors all current routing information on the data plane and also added an RIRM traffic controller to the control plane of the 5GC network. The RIRM traffic tracker transmits the information it collects to the RIRM traffic controller. As a result of this flow, they try to predict the best route. In the experimental results, it turned out to be 102 and 22 times better compared to well-known round-robin load balance algorithms by way of packet loss rate latency, respectively. Moreover, the mean data throughput was boosted by 13% [145].

In 2021, Saeed et al. carried out a detailed study on 6G satellite-aerial networks. They provided a detailed review of the various layers in satellite networks and high-altitude platform (HAP) networks and two different solution proposals, such as radio-frequency (RF) and free-space optics (FSOs). At the point of future communication technology, satellite networks are at a point that cannot be ignored, along with the advantage of coverage. However, according to terrestrial networks, the volume of challenges in these networks is growing [122].

In 2022, Goswami et al. introduced a novel approach by integrating Distributed Artificial Intelligence (DAI) with neural networks to ensure energy-efficient routing and address the challenges of fast communication within intra-clusters of nodes in intelligent transportation systems (ITSs). Additionally, they propose a new method to combine DAI with Self-Organizing Maps (SOMs). This technique aims to minimize overall energy consumption and reduce computational challenges within the network [146].

A summary of the current articles on routing approaches in 6G and satellite networks is listed chronologically in

Table 5. Summary of existing articles on routing approaches in 6G and satellite networks.

Year	Article	Major Features and Characteristics	Network Type	Key Contributions	Challenges Addressed
2018	[143]	Real-time deep learning-based traffic control using Deep CNNs.	6G	Dynamic traffic routing with AI-driven adaptability.	Scalability in high-density networks; real-time processing demands.
2020	[144]	Examines ML integration challenges in 6G networks.	6G	Identifies gaps in interoperability and latency for AI/ML-driven routing.	Computational overhead; compatibility with legacy systems.
2020	[145]	Proposes RIRM (Reliable Intelligent Routing Mechanism) for 5GC congestion management.	6G	AI-based load-balancing for ultra-reliable low-latency communication (URLLC).	Congestion in high-density scenarios; reliability under mobility.
2021	[122]	Reviews satellite/HAPs network layers; compares RF and FSO solutions.	Satellite	Highlights trade-offs between RF (robustness) and FSO (high bandwidth).	Signal attenuation; atmospheric interference in FSO.
2022	[146]	Integrates Distributed AI with neural networks for energy-efficient routing in ITS.	6G/Satellite	Optimizes intra-cluster communication for rapid node mobility.	Energy efficiency; handling dynamic topologies.

. It reveals distinct routing priorities for 6G and satellite networks. The reviewed articles highlight distinct routing paradigms for 6G and satellite networks, each addressing unique challenges. In 6G networks, AI-driven approaches dominate: ref. [143] employs Deep CNNs for real-time traffic control, optimizing scalability but facing computational overhead, while ref. [144] identifies interoperability and latency hurdles in ML integration. Ref. [145] provides an RIRM algorithm to tackle congestion in 5G Core networks, ensuring reliability for URLLC applications like autonomous vehicles. Conversely, satellite routing [122] prioritizes robustness through hybrid RF/FSO solutions, where RF withstands weather disruptions, but FSO offers higher bandwidth, albeit with atmospheric interference. Cross-cutting themes emerge in [146] regarding a Distributed AI framework for energy-efficient routing in ITS, which is relevant to both domains due to shared challenges like dynamic topologies and energy constraints. Notably, 6G emphasizes low-latency terrestrial routing, while satellite networks focus on long-distance, delay-tolerant solutions. Future integration of these paradigms (e.g., 6G-satellite NTNs) could leverage AI to harmonize terrestrial speed with satellite coverage, though handover management and resource allocation remain critical gaps.

7. Conclusions

Technology is developing with unpredictable momentum. The demands for this development are both increasing and deepening. Future communication technologies demand Tbps-level speeds, end-to-end latency rates under milliseconds, a connection density of ${10}^7$ per square kilometer, and mobility of up to 1000 km per hour, with access to all these features from any location at any time. To fulfill these requirements, the availability and coverage of satellite networks is a key point. There is no suspicion that both 6G networks and satellite networks will progress arm-in-arm at a point of transformation in communication. However, routing protocols in these networks are also one of the important processes that affect throughput and demands. At this point, new routing approaches with high flexibility, invulnerability, and the ability to adapt to variable conditions are needed. This paper presents a comprehensive analysis of 6G, satellite networks, and routing techniques based on these networks. A detailed literature review of 6G and satellite networks, algorithms, key performance indicators, key enablers, challenges (and their control), and a detailed review of future directions was carried out.

To begin with, this paper presents the history of network generations from 1G to 6G. Specifically, the paper focuses on satellite and 6G networks and routing techniques and management. Then, the classification of routing techniques and the limits of these routing approaches are explained in detail. The methods of routing algorithms in 6G networks have been investigated, and the studies carried out have been presented. Satellite networks, routing protocols in satellite networks, and proposed routing algorithms for LEO networks have been examined from a broad perspective. This paper presents the challenges in routing management caused by the fact that the dynamic environments of 6G and satellite networks are very dominant; the reasons for these challenges are presented, and information about solution proposals is provided. Finally, the essential/priority parameter that should be included in order for routing algorithms to be more effective in the future and various technological approaches, such as AI/ML/DRL, new channels, new antennas, and security—which have great potential at this point—are discussed.

While this survey provides a comprehensive overview of routing protocols for 6G networks, future research should include a systematic and quantitative benchmarking of the described methods. This includes evaluating the performance of these protocols under realistic 6G scenarios in terms of latency, scalability, mobility support, and energy efficiency. Comparative analysis frameworks and experimental validations will be essential to understand trade-offs and guide the development of optimized routing solutions.