The application of intelligent agents is currently undergoing accelerated evolution and deployment. Based on technological capabilities and ecosystem strengths, different entities are pursuing varied implementation pathways and product forms, which faces many challenges. This paper first systematically outlines the two mainstream technical pathways: interaction based on Graphical User Interface (GUI) recognition and operation, and invocation based on Application Programming Interface (API) authentication. It then provides a thorough analysis of the impacts exerted by these different pathways on cybersecurity, privacy boundaries, and the evolution of industrial ecosystems, while also analyzing different technical choices, ecosystem models, and regulatory practices. Finally, considering the characteristics of China’s market structure and application ecosystem, the paper proposes a governance framework aimed at promoting the compliant empowerment of industrial innovation by intelligent agents and supporting the high-quality development of digital economy.
In the context of the significant breakthroughs in generative artificial intelligence technology, the mobile internet industry is transforming from an application-driven model to an agent-driven one. Intelligent terminal manufacturers and Internet service enterprises are competing to establish large model access portals, forming three typical ecological construction modes: deep integration of operating systems, super-App service aggregation, and the reconstruction of terminal native systems. Taking Doubao’s hardware-software integration attempt, Tongyi Qianwen’s full-scenario service access, and domestic mobile phone manufacturers’AIOS as examples, this paper deeply analyzes the current industrial problems such as access portal competition, data fragmentation and service invocation barriers. To promote the high-quality development of the digital economy, it is necessary to promote the construction of technical standards for end -cloud collaboration, clarify the ecological boundaries between terminals and services, and build a new paradigm of open, interconnected, secure, and trustworthy industrial collaboration.
With the migration of artificial intelligence (AI) technologies toward the edge, intelligent terminals are evolving from cloud service access devices into computing nodes with local intelligence capabilities. This paper systematically sorts out the development path of intelligent terminal AI computing technologies, analyzing its evolutionary trajectory from cloud-based inference to edge-side inference, from lightweight models to the deployable large models, and from single-point optimization to system-level collaboration. It summarizes core optimization technologies including model compression, compilation optimization, hardware acceleration and system-level energy efficiency management, and discusses key challenges such as edge-friendly model paradigms, unified compilation and runtime systems, AI-native operating systems, and local intelligent agent construction, providing a useful reference for research and industrial practices in this field.
Intelligent terminals are accelerating towards being lightweight, low-power, highly secure, and customizable. The authorization constraints, power consumption bottlenecks, and lack of flexibility in traditional chip architectures can no longer meet the innovative demands of multiple scenarios. The Reduced Instruction Set Computing-V (RISC-V) architecture, with its core advantages of being open-source and free, modular and customizable, low-power, and highly scalable, has broken the monopoly of traditional architectures and become a key support for the transformation and upgrading of the intelligent terminal industry. By focusing on technological breakthroughs, ecosystem building, and industrial collaboration, an optimization path is proposed to enhance the adaptability and synergy between RISC-V and intelligent terminals, helping to promote high-quality development of the industry.
With the integrated development of the 5th generation mobile communication technology and artificial intelligence technologies, smart terminals have been widely applied across various industries, leading to frequent security incidents and multiple threats. This paper first analyzes the development dynamics of mobile terminal security at home and abroad, then systematically examines the key security technologies for intelligent terminals from five core dimensions: chips, operating systems, applications, data and privacy, and artificial intelligence. It further delves into the evolutionary processes of security technologies in chips, operating systems and applications. Finally, it dissects the current challenges faced by the security of China’s intelligent terminals from the aspects of standards, technologies, industries and applications, proposing development recommendations to provide practical references for advancing smart terminal security technologies in China and offering technical support for building a trusted digital infrastructure.
Against the background of compliance-driven development, this paper systematically analyzes the current state of the intelligent terminal security protection systems and the core challenges in compliance implementation. It deconstructs the three-tier in-depth defense technical framework covering the hardware security layer, system kernel layer, and application & data layer of intelligent terminals, and clarifies the compliance adaptation direction of core security technologies. Finally, this paper constructs a multi-dimensional compliance practice verification system, providing theoretical and practical references for the secure and compliant development of the intelligent terminal industry.
Infrastructure-less communication is one of the recent innovations in smartphone connectivity. It enables devices to communicate between each other within a certain range without relying on traditional network infrastructure. This paper analyzes the development of infrastructure-less communication, covering key aspects such as technical principles, primary application scenarios, functionality and performance, security, challenges, and regulatory suggestions. The aim is to provide reference information for future development directions and regulatory measures in this field.
The deep integration of Artificial Intelligence (AI) and Augmented Reality (AR) technologies is driving AI glasses to progressively transform from proof-of-concept to commercialization, demonstrating significant application potential and commercial value across fields such as consumer electronics, industrial manufacturing, and healthcare. Based on a systematic review of the latest industry trends, this paper provides an in-depth study of the recent advancements in five key technological areas of AI glasses: hardware miniaturization, software algorithm optimization, human factors engineering, large model integration, and multimodal interaction. It also offers a comprehensive analysis of the industrial chain structure, market growth potential, current state of the content ecosystem, and mass-production challenges.
To address the challenge of slowing market growth for traditional TV products in core family scenarios, this study investigates and proposes a novel remote real-time communication device based on spatial perception and emotional synchronization. It systematically analyzes the pivotal role of 4K/8K ultra-high-definition technology in this context and explores the engineering challenges and key technical pathways for end-to-end ultra-high-definition real-time interaction. The results demonstrate that ultra-high-definition technology can significantly enhance the system’s immersion and multimodal perception accuracy while enabling new interaction methods, serving as the critical foundation for realizing the design vision of this device. The mature application of 4K/8K technology can provide core driving force for the device, thereby opening new pathways for scenario innovation in the display industry and injecting fresh vitality into the future information society.
The integration of artificial intelligence (AI) and augmented reality (AR) technologies propels digital twin cities towards dynamic interaction and evolution, facilitating the optimization of urban governance paradigms. Drawing on Kevin Kelly’s Mirror Worlds theory, this paper constructs an integrated governance analysis framework encompassing “perception-cognition-decision-making-interaction”, dissecting the intrinsic mechanisms by which the integration of AI and AR empowers digital twin city governance. An empirical study is conducted using smart glasses as a terminal, incorporating scenarios of fire emergency response and community governance. The research reveals that the deep integration of the two can drive urban governance towards phased optimization, transitioning from passive response to limited active intervention, from empirical decision-making to the initial stages of scientific decision-making, and from fragmented management to local system coordination. Finally, the paper explores challenges and countermeasures related to data privacy and algorithmic fairness, providing theoretical and practical references for the development of digital twin cities.
To address the communication and operational requirements of remote orthopedic surgical robots in practical engineering deployment, this paper constructs a system-level test environment based on cellular communication networks. Focusing on key dimensions including system security, functionality, and service performance, a testing indicator system and corresponding test methods oriented toward typical remote surgical scenarios are designed. Under controllable network conditions, critical processes such as remote surgical planning, navigation coordination, and audio-video services are validated and analyzed. The results demonstrate that, with appropriate network parameter configurations, cellular communication networks can effectively support the stable operation of remote orthopedic surgical robot systems, while different services exhibit differentiated sensitivities to variations in network performance.
This paper proposes a hybrid access method for multi-access points and multi-cloud pools based on 5G technology, by constructing a redundant active-standby network architecture and combining the active-standby switching mechanism of wired links with software-defined networking in a wide area network (SD-WAN) superimposed 5G slicing, it solves the hybrid cloud access problems caused by the replacement of old and new systems and the requirement for business continuity. Focusing on sub-second switching between primary and backup links, the method realizes efficient connection between multi-access points and multi-cloud pools through a hierarchical service guarantee mechanism. Verified by practical tests, on the premise of ensuring business continuity, this method can significantly reduce network management complexity and operational costs, and may serve as an effective access solution for high-reliability network scenarios..
The advancement of brain-computer interface (BCI) technology has sparked academic discussions on the protection of brain privacy. Brain privacy exhibits novel characteristics distinct from traditional privacy, including diversity in objects, passive expression, and sensitivity in content, so it is necessary to construct differentiated protection pathways. The objects of brain privacy protection are stratified according to the technological stages, encompassing raw brain signals at the physical layer, basic physiological information at the decoding layer, and high-level cognitive content at the cognitive layer. Each object carries distinct value orientations, requiring a hierarchical protection approach. Raw brain signals at the physical layer and basic physiological information at the decoding layer should be incorporated into the framework of sensitive personal information protection, balancing protection and utilization through enhanced “specific purpose + separate consent” rules and the introduction of anonymized processing certification systems. For high-level cognitive content at the cognitive layer, absolute protection within the realm of privacy rights is essential, which can be ensured through privacy by design principles, technological reinforcement measures, and stringent legal responsibilities.