Intelligent Networking and Computing for Next-Generation Wireless Applications

随着无线网络、边缘计算、大数据分析和人工智能（AI）技术的发展，超第五代（B5G）无线通信网络有望支持日益多样化的移动宽带和面向物联网（IoT）的无线应用，并对服务质量（QoS）提出严格和差异化的要求。未来的物联网通常不仅需要高无线传输性能，还需要高效的计算来支持不同类型的任务处理（例如，对象检测）。然而，由于数据量的爆炸式增长、网络复杂性的增加和时空流量的动态变化，在当前的网络模式下，满足通信和计算性能指标在技术上是具有挑战性的。建立该研究计划的长期目标是研究下一代无线网络设计的科学原理和开发工程解决方案，从而降低以用户为中心的服务交付的复杂性。未来五年的短期目标是研究智能网络和计算解决方案，以支持具有细粒度QoS的数据密集型和计算密集型无线应用。从网络部署、资源编排和协议设计三个方面提出了以下研究重点：1)动态网络部署：通过利用先进的流量预测和机器学习技术，以经济有效的方式开发主动和动态的空中网络部署策略，以覆盖时空通信需求变化；2)边缘智能支持的资源编排；开发边缘协调的自适应AI模型构建策略，优化通信和计算资源利用，以实现高任务处理效率。3)学习辅助网络协议设计：利用建模和学习技术，开发网络协议，以实现高性能和简化开销。这些协议将适应B5G网络动态，并为支持差异化业务而定制。该研究计划将为建立下一代无线系统提供新的基础见解，为通信、计算机工程/科学、人工智能应用等领域的技术和社会发展做出贡献。该计划还将有助于培养高素质人才（HQP），为学员提供尖端技术，技能和研究与开发（R&D）经验的知识，为他们的学术/工业生涯做好准备。该计划预计将导致高质量的出版物和专利，这将推动加拿大的技术和创新。

With the technological development of wireless networking, edge computing, big data analytics, and artificial intelligence (AI), the beyond fifth generation (B5G) wireless communication networks are envisioned to support increasingly diversified mobile broadband and Internet-of-Things (IoT)-oriented wireless applications with strict and differentiated quality-of-service (QoS) requirements. The future IoT often requires not only high wireless transmission performance but also efficient computation to support different types of task processing (e.g., object detection). However, due to explosive data volume, increased network complexity, and spatial/temporal traffic dynamics, it can be technically challenging to satisfy the communication and computing performance indicators under current networking paradigms. The long-term goal of establishing this research program is to investigate scientific principles and develop engineering solutions for the design of next-generation wireless networks such that user-centric service deliveries can be provided with reduced complexity. The short-term objectives for the next five years are to investigate intelligent networking and computing solutions to support data-intensive and computation-intensive wireless applications with fine-grained QoS. The proposed research includes the following three thrusts from the aspects of network deployment, resource orchestration, and protocol design: 1) Dynamic network deployment: To develop a proactive and dynamic aerial network deployment strategy, by leveraging advanced traffic prediction and machine learning techniques, to cover spatial and temporal communication demand variations in a cost-effective way, 2) Edge-intelligence-enabled resource orchestration: To develop edge-coordinated adaptive AI model structuring strategies, with optimized communication and computing resource utilization, for achieving high task processing efficiency, 3) Learning-assisted networking protocol design: To develop networking protocols for achieving high performance with simplified overhead by leveraging both modeling and learning techniques. The protocols will be adaptive to B5G network dynamics and be customized for supporting differentiated services. The proposed research program is expected to generate new fundamental insights for establishing the next-generation wireless systems, which will contribute to the technological and social development in the fields of telecommunications, computer engineering/science, and applications of AI. This program will also contribute to the training of highly qualified personnel (HQP), providing the trainees with knowledge of cutting-edge technologies, skills, and research & development (R&D) experience, which will prepare them for academic/industrial careers. This program is expected to lead to high quality publications and patents which will advance Canadian technology and innovation.