【摘要】 无线算力网络充分挖掘并利用了移动终端的算力资源，极大地提升了计算过程的灵活性与实时性。旨在面向确定性性能需求和随机性传输环境的矛盾，深入分析无线算力网络特性，研究和展望无线算力网络的确定性技术。从算力网络与确定性技术的发展现状切入，涵盖网络架构、业务场景，从算力信息收集、算力资源调度、算力服务平台三方面探讨支撑确定性技术的算力管控体系，重点介绍了随机网络演算、极值理论、随机优化、机器学习等对抗随机性的确定性技术理论基础，并针对无线算力网络中随机无线环境与通信计算耦合的技术挑战，详细分析和展望了面向无线算力网络的确定性技术方案。这些技术的发展将进一步推动算网融合，提升网络在新兴行业应用的服务能力，为业务提供确定性性能保障。更多还原

【Abstract】 The wireless computing power networks fully utilize the computing power resources of mobile terminals, significantly enhancing the flexibility and real-time performance of computing. This paper aims to address the contradiction between the deterministic performance demand and the random transmission environment, conduct an in-depth analysis of the characteristics of wireless computing power networks, and introduce the research on the deterministic technologies for wireless computing power networks. Starting from the current development status of computing power networks and deterministic network technologies, it covers network architecture and application scenarios. Then, we discuss the computing power management system that supports deterministic technology from three aspects: computing power information collection, computing power resource scheduling, and computing power service platform. We introduce the theoretical foundations of deterministic technologies for combating the randomness, such as stochastic network calculus, extreme value theory, stochastic optimization, and machine learning. Given the technical challenges of random wireless environments and communication-computing coupling in wireless computing power networks, it analyzes and prospects the deterministic network technology solutions for wireless computing power networks. The development of these technologies will further promote the integration of computing and networking, enhance the network capabilities in emerging industry applications, and provide deterministic performance guarantees for services.更多还原