Space-Air-Ground Integrated Network (SAGIN) 星空地整合網路 represents a groundbreaking paradigm for next-generation communication networks, seamlessly integrating spaceborne, airborne, and ground-based components to establish a ubiquitous and robust communication ecosystem. SAGIN aims to address the increasing demands for global connectivity, high data throughput, and real-time communication across diverse and geographically dispersed regions. Its architecture typically consists of satellites in space, unmanned aerial vehicles (UAVs) 無人機 or high-altitude platforms (HAPs) 高空平台飛船 in the air, and traditional terrestrial networks on the ground. By leveraging the complementary characteristics of these layers, SAGIN ensures connectivity even in remote or disaster-affected areas where conventional terrestrial networks may be unavailable or compromised. At the forefront of SAGIN’s development is the deployment of Low-Earth Orbit (LEO) satellites 低軌衛星, which orbit the Earth at altitudes ranging from 500 to 2,000 kilometers. Unlike geostationary satellites, LEO satellites offer significantly reduced latency due to their proximity to the Earth’s surface, making them ideal for applications requiring real-time communication, such as autonomous vehicles, telemedicine, and augmented reality. Additionally, the constellation-based architecture of LEO satellites ensures global coverage by deploying a network of interconnected satellites. This enhances the system's capacity and resilience, enabling efficient load balancing and fault tolerance. LEO satellite communications also align with the growing trends in 6G technologies, facilitating seamless integration with terrestrial networks. Advanced technologies such as beamforming, multi-band communication, and machine learning algorithms further optimize LEO satellite operations, ensuring reliable and high-throughput communication.

(RIS Theory) The fixed BS infrastructure can beamform its mmWave/THz signals to the RIS 智慧反射面板, which may reflect them to arbitrary transmit directions. Blocking the line-of-sight (LOS) paths of mmWave/THz carriers may be circumvented with the aid of RISs. However, they create extra interference, which has to be carefully managed. In this context, it is imperative to jointly design the active beamforming at the BS and the passive phase shift based beamforming at the RIS in order to meet different requirements. RIS-empowerd multiuser transmissions include RIS-NOMA, RIS-FD, RIS-CoMP. An impediment of the conventional RIS solutions is that the transmitter and the user have to be within the same 180-degree half-plane, rather than roaming across the entire 360-degree full plane. By contrast, the STAR-RIS architecture 同時傳輸與反射之智慧反射面板, or termed as intelligent omni-surface (IOS), allows full-plane coverage by potentially harnessing full reflection, full transmission, as well as simultaneous transmission and reflection. These modes were discussedwith special emphasis on NTT DOCOMO's prototype. There are three different principles governing their operations, namely the so-called energy-splitting, partitioning and time-switching types, which have their different pros and cons. In the energy-splitting mode the signal impinging upon an element is partially reflected and transmitted. There is a huge variety of compelling applications scenarios and RIS types from 2019-2025 in communication domains:

• Active RIS/STAR: Amplifying signals with active element gain larger than one

• Hybrid RIS: Active and passive elements or sensing/comm. oriented elements

• STAR-RIS: Simultaneous transmission and reflection

• DS-STAR: Incident signals impinging from both sides

• MF-RIS: Energy harvesting, amplification and STAR capabilities (Self-sustainability)

• BD-RIS: Beyond diagonal RIS with full-coverage (multi-sectors, hexagonal)

• SIM: Multiple stacked RIS with refractive functionality (similar to relaying)

• Flexible RIS: Inspired from fluid/movable antenna systems, flexible intelligent metasurface (FAS/MA/FIM)

(RIS Implementation) 

1. i-Dris, Intelligent Rapid Deployment for Reconfigurable Intelligent Surface, resolving mmWave blockage and non-line-of-sight (NLOS) issues, has integrated Nokia 5G BS (28 GHz) and RIS with MIMO beamforming and 1 Gbps rate (100 MHz bandwidth). The RIS is equipped on auto-guided vehicle (AGV) performing the designed multi-time scaled temporal reinforcement learning for RIS deployment. The edge server monitoring throughput data is transferred to AGV-RIS. The weights for probabilistic deployment actions are computed based on historical data and current throughput reward. Results reveal that i-Dris achieves maximum throughput up to 900 Mbps (1.2-3.6x compared to existing solutions) under NLOS, multi-reflections, blockages, hand-held behavior, long distance, different beamwidths. Compared to academic/industrial solutions, i-Dris requires no fiber, control signals from BS, re-measurement, and offline-learning, which has 10 min deployment time compared to existing methods up to 1 hr. 

News: 電信三雄高層參展MWC 簽6G備忘錄、秀AI節能網路、遠傳挺進MWC 2023秀肌肉 發佈5G毫米波整合部署方案、遠傳MWC秀肌肉 展台灣首發B5G毫米波黑科技、遠傳全台首發 5G毫米波整合部署解決方案 遠傳與網通大廠光寶科技、國際材料大廠與頂尖學府國立陽明交通大學（方教授、沈博士團隊）共同研發的「B5G毫米波整合智慧反射板」同樣在MWC首次登場，為全台首發的5G毫米波室內佈建解決方案，可以強化被環境遮蔽區域的5G涵蓋範圍、包含大樓林立的深層室內，並可以根據環境需求，客製化調整無線電波角度，精準解決真實情境中複雜的盲區覆蓋問題。

2. Regarding sensing, human presence detection is a promising technology for detecting a human body’s presence in an area of interest. Channel state information (CSI) extracted from Wi-Fi can offer high precision environment detection based on its abundant signal characteristics, which can be applied for wireless sensing. It extracts meaningful information of spatial, motion and static features, and applies AI/ML techniques to identify human behaviors. Moreover, signal dead spots may occur due to multi-reflection pathloss. Therefore, RIS with higher dimensional antennas re-directs the signal paths to the desired measured areas. With higher resolution of RIS through adjusting its deployment and phase shift configuration on the metasurface, number of learning features will be largely escalated.

Integrated sensing and communication (ISAC) 整合感測與通訊 relying on artificial intelligence (AI) and machine learning (ML) 人工智慧與機器學習 have become key techniques in the intelligent autonomous vehicular network coordination (i-VNC) in intelligent transport systems (ITS), such as auto-pilot, platooning, connected and autonomous vehicles (CAV) and vehicle-to-everything. In i-VNC system, we propose joint robust intra-vehicle sensing control and inter-CAV, i.e., vehicle-to-everything (V2X), vehicle-to-vehicle (V2V) or vehicle-to-infrastructure (V2I) coordination, aiming for achieving soaring service performances of rate, latency, and security. Different sensors of radar, lidar and camera are employed to interact with environments for safe-driving policy decision, whilst communication units are equipped for inter-vehicle communication, which forms integrated sensing and communication (ISAC) as specified in 6G prospects. Issues of intra-vehicle control include wireless resource management, vehicular action control, and environmental data collected from sensors. Management and control are benefited by leveraging the reinforcement learning (RL) aided decision making under the variant and highly-mobile scenario. While, fusion-based deep learning (DL) architecture should be designed for multi-modal data from all sensors, which extract influential hidden features from different modality. As for inter-CAV in ISAC, wireless-oriented issues arise, such as latency-aware radio resource orchestration and reliable cooperative learning optimization under stringent use case of ultra-reliable and low latency communications (URLLC). Moreover, multi-agent learning allows an individual car to update its action after interacting with the environments, which may be potential and efficient as a game-manner solution for low-delay inter-CAV optimization but with low reliability.

Artificial intelligence, machine learning (AI/ML) and deep learning, 人工智慧/機器學習/深度學習 including supervised learning, unsupervised learning and reinforcement learning have found favour in solving challenging communications and network problems. Specific examples are constituted by radio interference management, resource allocation, multiple parameter optimization, network traffic prediction, computing resource assignment, and flexible configuration of network functions. In supervised learning, ground truth labels and fixed-size inputs constitute a deep layered neural network (NN). However, labeling is not required in unsupervised learning, which exploits the correlation between samples of the dataset. In reinforcement learning, an agent will interact with the environment and then updates the model based on the corresponding rewards. Note that deep learning can deal with comparably complex problems in a non-linear and non-convex manner than that utilizing machine learning. Therefore, we can efficiently manage both vertical and horizontal networks with the aid of deep neural networks. Moreover, in a multi-BS and multi-network scenarios, federated edge learning (FEL) can perform resource assignment collaboratively among local BSs and devices. Aided by FEL-empowered central server, hybrid TH/mmWave beamforming and AI-enhanced radio resource management have the capability to provide services by guaranteeing globally-optimal data rate, latency, utilization, fairness, security, and energy/spectrum efficiencies. The selected topics for AI/ML include:

1. Transformative convolutional network, UNet family

2. Attention-based Transformer family, multi-modality fusion

3. Reinforcement learning family

4. Federated learning, gossip learning, distributed learning family

Digital-Twin (DT) 數位雙生, as applications of AI/ML, is composed of a physical world and digital world, which benefit each other by exchanging informative data or models. Physical world can provide architecture and related information in real environments at small scale, whereas digital world provides policy adjustment for large-scale scenarios based on the input of physical information. Such information exchange may induce high overhead, depending on the amount of data or module of AI/ML. Such case will become particularly severe in a long-distance transmission, such as satellite networks. Moreover, partial radio resources, such as power, beam and channels will be leveraged and shared among other communication or sensing-oriented tasks. By guaranteeing sufficient task requirements (e.g., data throughput, sensing accuracy), DT can assist both physical/digital worlds to improve overall performances; Semantic Communications (SC) 語義通訊, as applications of AI/ML represents a transformative approach to Beyond-Shannon communication systems, aiming to enhance efficiency by transmitting not just raw data but the underlying meaning and intent of information. Unlike conventional communication methods that focus on achieving minimal error rates in data transfer, SC emphasizes the accurate interpretation and relevance of the transmitted content. By extracting and transmitting essential semantics, or meaning, from a message, this approach reduces the amount of data needing transmission, resulting in reduced bandwidth usage and lower latency. SC enables devices and systems to operate more collaboratively by prioritizing critical information and dynamically adjusting to the context, leading to more efficient decision-making and resource management. Additionally, by integrating AI/ML models, SC systems can learn to interpret diverse data, fuse multi-modal, and even anticipate user intentions, bringing a new level of intelligence and adaptability to communication networks.

6G new radios (NR) with advanced transmission and multiple access techniques for are pivotal for meeting the demands of ultra-high throughput, low latency, massive connectivity, and energy efficiency in next-generation wireless networks. These techniques collectively form the backbone of advanced transmission strategies, ensuring enhanced performance across diverse applications such as autonomous systems, smart cities, immersive AR/VR, and massive IoT. By integrating these innovations, future wireless networks aim to achieve unprecedented levels of reliability, capacity, and adaptability, supporting the evolution towards a fully connected and intelligent digital ecosystem.

• Movable/Fluid/Pinching Antennas (MA/FAS/PASS) are emerging antenna technologies that dynamically adapt their physical positions or electromagnetic properties to enhance wireless system performance. By enabling reconfigurable aperture geometries and location-aware radiation patterns, MA/FAS/PASS architectures offer improved spatial diversity, channel decorrelation, and adaptive coverage. These features provide new degrees of freedom for interference mitigation, capacity scaling, and resilient connectivity in next-generation communication systems.

Millimeter wave (mmWave) 毫米波 technology achieving multi-gigabits speed plays a significant role in beyond 5G and the next 6G wireless communication networks thanks to its huge spectrum utilization and beam-based directional transmissions. To tackle temporary ultra-high data demands of hotspot areas, three-dimensional (3D) heterogeneous network (HetNet) is designed with the integration of mmWave unmanned aerial vehicles (UAV) to provide resilient instantaneous control and data transmissions. However, some critical beam-related issues for mmWave implementation of UAVs/drones are addressed including robust initial beam alignment, mobility-aware beam tracking and beam refinement. In this research, we aim at developing robust and efficient beam control mechanisms by implementing a prototype of 3D flying heterogeneous communications. The backhaul connections operate at mmWave frequency between airship and UAV/drone, while fronthaul links adopt lower frequency bands such as Wi-Fi for multiuser data transmissions.

Terahertz (THz) 太赫波 is considered as a promising technology using wider hundreds-of-GHz bandwidth compared to millimeter wave (mmWave) transmissions, which is capable of supporting Tbps high speed data traffic. However, THz transmission possesses more severe path loss due to higher operating frequencies at 0.1-10 THz. Accordingly, new hybrid beamforming techniques should be enhanced by array-of-subarray (AoSA) equipped with tens-of-thousands of antenna elements. Due to short and medium distance-based wireless transmissions, a large number of THz-enabled base stations (BSs) should be deployed to support seamless coverage and high throughput performance. When using ultra-thin THz beams, beam training becomes compellingly imperative due to enormous number of beam training steps, which potentially induces high training latency overhead.

The evolution of WiFi standards, particularly WiFi 6 (802.11ax), WiFi 7 (802.11be), and the forthcoming WiFi 8, is driven by the need to optimize performance, efficiency, and reliability in increasingly dense and demanding wireless environments. These advancements address challenges such as higher data rates, reduced latency, and improved power efficiency, while ensuring compatibility with a growing ecosystem of smart devices. The progression from WiFi 6 to WiFi 8 highlights a trajectory of continual optimization, aiming to meet the demands of smart cities, 8K streaming, immersive gaming, and large-scale IoT ecosystems. By leveraging cutting-edge technologies, these advancements ensure that next-generation WiFi remains a cornerstone of global connectivity. 

• WiFi 6 (802.11ax) introduces features like Orthogonal Frequency Division Multiple Access (OFDMA) and 1024-QAM modulation, enabling efficient utilization of the spectrum and increased data rates. Target Wake Time (TWT) optimizes energy efficiency for IoT devices by scheduling communication times, reducing idle power consumption. Additionally, Basic Service Set (BSS) Coloring minimizes interference in high-density deployments, enhancing performance in congested areas such as stadiums and airports.

• WiFi 7 (802.11be) pushes optimization further with Extremely High Throughput (EHT), offering data rates beyond 30 Gbps. Key features include Multi-Link Operation (MLO), enabling simultaneous use of multiple bands (2.4 GHz, 5 GHz, 6 GHz) to improve reliability and reduce latency. 320 MHz channel bandwidth and 4096-QAM modulation significantly enhance throughput. Advanced interference management and deterministic latency mechanisms make WiFi 7 ideal for real-time applications like AR/VR and industrial automation.

• WiFi 8 (802.11bn) While still under development, WiFi 8 is expected to focus on AI-driven optimization for dynamic resource allocation, adaptive beamforming, and enhanced security protocols. Integration with terahertz frequencies and cognitive radio technologies will further expand capacity and enable ultra-low latency communication for next-generation applications.

• WiFi Sensing (802.11bf) enables analyzing variations in wireless signals caused by motion, presence, or physical changes, introducing sensing capabilities such as human activity recognition, gesture detection, and intrusion monitoring, all using existing WiFi infrastructure. By leveraging channel state information (CSI) and advanced signal processing, 802.11bf transforms WiFi networks into dual-purpose platforms that support both communication and ambient sensing, unlocking new applications in smart homes, healthcare, and security.