6G wireless networks and beyond

Current State

Research into 6G wireless networks is rapidly advancing, aiming to address the limitations of 5G and meet future demands for enhanced connectivity, intelligence, and efficiency. The vision for 6G includes global coverage, multi-Tb/s speeds, ultra-low latency, and pervasive AI integration (You et al., 2020; Zhang et al., 2019). Key technological pillars are emerging, such as terahertz (THz) communication, reconfigurable intelligent surfaces (RIS), and AI-native network architectures (Chaccour et al., 2022; Nwufo et al., 2025; Wu et al., 2022).

Strongest Evidence

THz communication is a cornerstone, offering vast bandwidth for high-resolution sensing and significant capacity gains, though it faces challenges with dynamic channels (Chaccour et al., 2022; Papasotiriou et al., 2021). RIS technology is gaining traction for its ability to enhance coverage and energy efficiency by intelligently manipulating wireless signals (Nwufo et al., 2025; Hamedoon et al., 2025). AI is not merely an application but is becoming integral to 6G network design, enabling intelligent management, resource allocation, and signal detection (Letaief et al., 2019; Wu et al., 2022; Babu et al., 2024). The concept of AI-native networks, moving beyond incremental AI applications, is a significant paradigm shift (Thomas et al., 2024).

Unresolved Uncertainties

While the potential of THz communication is clear, overcoming the challenges of highly uncertain and dynamic channels remains a significant hurdle (Chaccour et al., 2022). The practical implementation and scalability of RIS in complex, large-scale networks require further investigation. Furthermore, developing truly AI-native wireless networks, which move beyond data-driven, black-box AI models to incorporate causal reasoning, presents substantial technical challenges (Thomas et al., 2024). Ensuring energy efficiency and security in these highly complex, integrated networks is also a critical area of ongoing research (Huang et al., 2019; You et al., 2020).

Why the Topic Matters

6G networks are crucial for enabling the intelligent information society of 2030 and beyond, supporting emerging applications that demand unprecedented levels of connectivity, intelligence, and performance (Zhang et al., 2019; Dogra et al., 2020). The convergence of communication and sensing, facilitated by technologies like THz, will unlock new capabilities for environmental interaction and data acquisition (Chaccour et al., 2022). The deep integration of AI will transform network management, leading to more autonomous, efficient, and adaptable systems, ultimately driving innovation across various sectors from smart cities to advanced robotics (Letaief et al., 2019; Wu et al., 2022).

Consensus Map

Points of Agreement

• 6G networks will significantly exceed 5G capabilities, offering global coverage, enhanced spectral/energy/cost efficiency, and improved intelligence and security (You et al., 2020; Dogra et al., 2020; Zhang et al., 2019).
• Artificial Intelligence (AI) and Machine Learning (ML) are fundamental to 6G, enabling intelligent network management, resource allocation, and supporting ubiquitous AI services from core to edge devices (Letaief et al., 2019; Wu et al., 2022; Thomas et al., 2024).
• Terahertz (THz) communication (0.1–10 THz) is a cornerstone technology for 6G, providing vast bandwidth for significant capacity gains and high-resolution environmental sensing (Chaccour et al., 2022; Papasotiriou et al., 2021).
• Reconfigurable Intelligent Surfaces (RIS) are crucial for next-generation wireless networks due to their energy efficiency and low-cost characteristics, addressing resource scarcity and improving connectivity and coverage (Nwufo et al., 2025; Hamedoon et al., 2025).
• 6G architecture will integrate space, air, ground, and underwater networks to provide ubiquitous and unlimited wireless connectivity (Zhang et al., 2019; Huang et al., 2019).

Active Disagreements

• While AI is universally accepted as critical, the specific implementation and challenges of creating truly "AI-native" wireless networks are still being debated, with some highlighting limitations of data-driven AI and advocating for causal reasoning (Thomas et al., 2024).

Uncertain Claims

• The precise timeline for the widespread commercialization and deployment of 6G networks and its advanced features remains an estimate, generally projected for 2030 and beyond (You et al., 2020).
• The full extent of performance gains and reliability of THz systems, given their highly uncertain and dynamic channels, is still under investigation, requiring robust fading models and deployment strategies (Chaccour et al., 2022; Papasotiriou et al., 2021).

Evidence Gaps

• More experimental validation is needed for proposed 6G technologies, particularly for RIS-assisted systems and novel AI-driven signal detection methods, to move beyond theoretical and simulation-based performance assessments (Papasotiriou et al., 2021; Babu et al., 2024; Hamedoon et al., 2025).
• Comprehensive studies on the energy consumption and environmental impact of large-scale 6G deployments, especially concerning the integration of diverse network segments (space, air, ground, underwater), are needed to ensure "green" network evolution (Huang et al., 2019).
• Further research is required to develop robust security and privacy mechanisms for AI-native 6G networks, given the increased complexity and data reliance (You et al., 2020).

The vision for 6G wireless networks extends beyond 5G, aiming for deployment by 2030 with enhanced capabilities such as global coverage, improved spectral, energy, and cost efficiency, and support for new services (You et al., 2020). These networks are expected to deliver multi-terabyte per second (Tb/s) speeds and incorporate ubiquitous intelligence (Zhang et al., 2019).

One of the cornerstones of 6G is terahertz (THz) communication (0.1–10 THz), which offers substantial bandwidth for significant wireless capacity gains and high-resolution environmental sensing (Chaccour et al., 2022). However, operating at these high frequencies presents challenges, primarily due to high path loss (Chaccour et al., 2022).

Artificial intelligence (AI) is integral to 6G, with proposals for AI-native network slicing architectures to enable intelligent network management and virtualization (Wu et al., 2022). This integration facilitates the synergy between AI and network slicing, crucial for managing complex 6G environments (Wu et al., 2022). The overall 6G architecture is envisioned as a large-dimensional, autonomous system integrating space, air, ground, and potentially underwater networks (Zhang et al., 2019; Dogra et al., 2020).

The landscape of 6G wireless networks is characterized by several directional trends aimed at achieving unprecedented performance and supporting novel applications beyond 5G. A significant trend is the increasing focus on terahertz (THz) communication (0.1–10 THz) as a cornerstone for future 6G systems. THz bands offer vast bandwidth, promising substantial capacity gains and enabling high-resolution environmental sensing, effectively merging communication and sensing functionalities (Chaccour et al., 2022; Papasotiriou et al., 2021). However, challenges related to dynamic and uncertain THz channels require further validation for reliable deployment (Chaccour et al., 2022).

Artificial Intelligence (AI) integration is another dominant trend, moving towards "AI-native" network architectures rather than merely "AI for wireless" paradigms (Thomas et al., 2024; Letaief et al., 2019). This involves AI-native network slicing for intelligent management and support of emerging AI services (Wu et al., 2022), and AI-driven signal detection methods for complex MIMO systems (Babu et al., 2024). The shift towards causal reasoning in AI is emerging as a critical method to overcome limitations of data-driven AI, enhancing adaptability and robustness in next-generation wireless networks (Thomas et al., 2024).

Reconfigurable Intelligent Surfaces (RIS) are gaining traction as an enabling technology to manage scarce wireless resources and improve coverage and energy efficiency (Nwufo et al., 2025). RIS-assisted non-orthogonal multiple access (NOMA) is being explored for IoT networks, addressing challenges in connectivity, coverage, and energy dissipation (Hamedoon et al., 2025). These surfaces promise significant performance strides by intelligently reconfiguring the wireless environment.

Emerging methods also include the development of green 6G networks that prioritize energy efficiency alongside high Quality of Service, integrating terrestrial, space, aerial, and underwater networks for ubiquitous access (Huang et al., 2019). The vision for 6G encompasses global coverage, enhanced spectral/energy/cost efficiency, and improved intelligence and security (You et al., 2020; Dogra et al., 2020; Zhang et al., 2019).

Signals that need more validation include the practical implementation and channel modeling for THz systems, the full potential and challenges of AI-native network slicing, and the comprehensive integration of diverse network components (space, air, ground, underwater) into a seamless, intelligent 6G architecture.

Synthesis

Sixth-generation (6G) wireless networks are envisioned to transcend the capabilities of 5G, providing global coverage, enhanced spectral and energy efficiency, and native intelligence (You et al., 2020; Dogra et al., 2020). The core vision involves integrating space, air, ground, and even underwater networks to achieve ubiquitous connectivity (Zhang et al., 2019; Huang et al., 2019). A fundamental shift is the deep integration of Artificial Intelligence (AI), moving towards AI-native architectures that manage network resources, such as network slicing, autonomously and efficiently (Letaief et al., 2019; Wu et al., 2022). Key enabling technologies at the forefront are Terahertz (THz) communications, which unlock vast bandwidth for multi-terabit-per-second data rates, and Reconfigurable Intelligent Surfaces (RIS), which promise to create smart, programmable wireless environments by intelligently controlling signal propagation (Chaccour et al., 2022; Nwufo et al., 2025).

Methods/Materials

Research in 6G heavily relies on advanced computational modeling and simulation to design and validate new architectures and protocols. This includes developing AI-driven resource allocation algorithms for complex systems like RIS-assisted Non-Orthogonal Multiple Access (NOMA) networks (Hamedoon et al., 2025). For instance, AI-based signal detectors using feed-forward neural networks (FFNN) are proposed to manage the high complexity of massive MIMO systems (Babu et al., 2024). A critical area is the development of accurate channel models for new frequency bands, such as experimentally validated fading models for the THz spectrum, which are essential for assessing system feasibility (Papasotiriou et al., 2021). Furthermore, there is a push towards new AI paradigms, like causal reasoning, to overcome the limitations of traditional data-driven models and build more adaptive, robust networks (Thomas et al., 2024).

Applications

The technological advancements in 6G are expected to enable a wide range of futuristic applications. The massive bandwidth from THz communications will support holographic displays, immersive extended reality (XR), and real-time digital twins. The integration of communication and sensing (ISAC) is a defining feature, allowing for high-resolution environmental sensing, gesture recognition, and tracking directly through the communication infrastructure (Chaccour et al., 2022). AI-native networks will facilitate intelligent transportation systems, smart cities, and large-scale Internet of Things (IoT) deployments with unprecedented efficiency and reliability (Hamedoon et al., 2025; Zhang et al., 2019).

Limitations

Despite the promising vision, 6G development faces significant hurdles. THz communication suffers from a short range and high susceptibility to blockages, making reliable links challenging (Chaccour et al., 2022). The massive deployment of technologies like RIS and integrated networks introduces immense complexity in resource management and optimization (Nwufo et al., 2025). Current AI models are often data-hungry, energy-intensive, and operate as 'black boxes,' which limits their adaptability and trustworthiness in critical network functions (Thomas et al., 2024). Furthermore, achieving a 'green 6G' with improved energy efficiency while delivering superior performance remains a primary challenge (Huang et al., 2019).

Update Triggers

This review should be updated upon the emergence of several key developments:

• Standardization Milestones: Official 6G standardization efforts by bodies like 3GPP.
• Hardware Breakthroughs: Advances in energy-efficient THz transceivers and low-cost, large-scale RIS manufacturing.
• AI Paradigm Shifts: Maturation of causal AI, federated learning, or other novel AI techniques demonstrating practical success in network management.
• Large-Scale Testbeds: Publication of results from large-scale, real-world 6G testbeds validating integrated sensing, communication, and AI functionalities.

Near-Term Work

1. AI-Native Network Architectures: Develop and refine AI-driven solutions for dynamic network slicing, resource allocation, and signal detection (Wu et al., 2022; Babu et al., 2024). A key focus is shifting from purely data-driven models to causal reasoning AI, which can adapt to dynamic network conditions with less reliance on massive training data (Thomas et al., 2024).
2. Terahertz (THz) Channel Modeling: Improve and experimentally validate channel models for the THz band (0.1–10 THz). This involves characterizing molecular absorption, fading, and blockage effects to design reliable communication protocols (Chaccour et al., 2022; Papasotiriou et al., 2021).
3. Reconfigurable Intelligent Surface (RIS) Optimization: Design efficient algorithms for optimizing the phase shifts of RIS elements to enhance signal coverage and system capacity. Research should focus on practical, low-complexity solutions for joint active beamforming at the base station and passive beamforming at the RIS (Nwufo et al., 2025; Hamedoon et al., 2025).

Validation Needs

• Experimental Testbeds: Physical testbeds are crucial for validating THz communication links and RIS performance beyond simulations. Experimental validation of fading models, as demonstrated by Papasotiriou et al. (2021), is essential.
• Integrated System Prototypes: Development of prototypes that combine multiple 6G technologies (e.g., RIS-assisted THz networks, AI-managed space-air-ground networks) is needed to evaluate system-level performance and interoperability (Zhang et al., 2019).
• Real-World Performance Metrics: Validation must move beyond theoretical metrics like spectral efficiency to include energy efficiency, cost efficiency, security, and quality of experience (QoE) under realistic network loads (You et al., 2020).

Datasets and Benchmarks

• Standardized Channel Datasets: Publicly available, large-scale datasets of channel measurements for THz and RIS-enabled environments are urgently needed to train and benchmark AI models for tasks like channel estimation and resource management.
• Network Traffic Benchmarks: Creation of benchmark scenarios representing future 6G use cases (e.g., holographic communication, tactile internet, massive IoT) is required to test the scalability and robustness of AI-native network slicing and management solutions (Wu et al., 2022).

Translation Barriers

1. Complexity and Computational Cost: The computational demand of AI models for real-time network control and the complexity of managing millions of RIS elements are significant hurdles for practical deployment (Letaief et al., 2019).
2. Hardware and Cost: Manufacturing cost-effective and energy-efficient hardware for THz transceivers and large-scale RIS arrays remains a major barrier to widespread adoption (Chaccour et al., 2022).
3. Security and Privacy: The introduction of ubiquitous AI and intelligent surfaces creates new attack vectors. Ensuring robust security and data privacy in highly dynamic, autonomous networks is a critical challenge (Dogra et al., 2020).

High-Value Experiments

• AI-Driven Joint Communication and Sensing (JCAS): Conduct experiments using THz systems to perform high-resolution environmental sensing (e.g., imaging, localization) simultaneously with data transmission. This would validate a core feature of 6G networks.
• End-to-End AI-Native Network Slice: Deploy and test an end-to-end network slice managed entirely by an AI agent, from resource allocation to QoS assurance, under dynamic traffic conditions to demonstrate the feasibility of zero-touch networks.
• Large-Scale RIS Deployment: An experiment deploying a large RIS in a complex urban environment to quantify the real-world gains in coverage extension and capacity for mmWave and THz frequencies.