Efficient Service Provisioning in Wireless Heterogeneous Networks with Ubiquitous Computing

在物联网普及和移动设备激增的推动下，我们见证了各种计算密集型应用的出现，如增强现实、自动驾驶和智能城市。受终端设备有限的能量和计算能力的限制，未来的无线网络将通过融合云计算和边缘计算等各种计算设施，在适应这些应用方面发挥至关重要的作用。此外，为满足随时随地的服务需求，无人驾驶飞行器(UAV)或高空平台(HAPS)等非地面网络将整合到网络架构中，在农村或服务不足的地区提供按需经济高效的覆盖。因此，未来的无线网络将成为一个高度异构化、动态化、复杂化的通信与计算融合的系统。为了以可扩展、自适应且经济高效的方式支持具有多样化服务质量(Qos)要求的广泛应用，有许多挑战需要解决。首先，系统中的巨大异构性使得分析和管理非常复杂，例如，各种无线链路和地面和空中基站(BSS)的覆盖、不同的计算服务器以及不同的资源。其次，需要高效地服务于在延迟、吞吐量和安全性方面具有不同的服务质量要求的多样化和动态的服务请求。第三，无线网络中的高动态性和多维随机性给服务提供带来了巨大的挑战，例如，无线信道条件的时变、流量的时空波动、边缘服务器不断变化的工作负载以及突发的用户请求。最后但并非最不重要的一点是，这种异类网络将导致通信和计算的大量能源消耗，从而导致高昂的运营成本(OPEX)。这项研究计划旨在通过在未来融合通信和计算的无线异质网络中开发新的网络管理解决方案来解决基本的研究挑战。这一创新的研究计划将主要研究三个主题：i)具有服务质量保证的无线异质网络中可扩展的计算任务调度和资源管理；ii)可持续和自适应的网络控制；iii)用于异质服务提供的动态网络切片和功能编排。通过应用最优化、人工智能、随机过程、马尔可夫决策过程、博弈论等，预计本研究计划将开发一套新的系统解决方案和方法，用于泛在计算的无线异质网络中的资源管理和服务提供。它还将有助于培训无线网络和边缘计算方面的高素质人才，并使他们能够为加拿大的高科技产业做出贡献。

Driven by the popularity of Internet of things and proliferation of mobile devices, we have witnessed the emergence of various computation-intensive applications, e.g., augmented reality, autonomous driving, and smart cities. Constrained by limited energy and computing capacities of end devices, future wireless networks will play a vital role in accommodating those applications by incorporating various computing facilities such as cloud computing and edge computing. In addition, to meet service demand anytime anywhere, non-terrestrial networks such as unmanned aerial vehicles (UAV) or high-altitude platforms (HAPs) will be integrated in the network architecture to provide on-demand cost-effective coverage in rural or underserved areas. Consequently, future wireless networks will become a highly heterogeneous, dynamic, and complex system with convergence of communication and computing. To support a wide range of applications with diversified quality of service (QoS) requirements in a scalable, adaptive, and cost-effective way, there are many challenges to be addressed. First, great heterogeneity in the system makes analysis and management very complex, e.g., various wireless links and coverage of ground and aerial base stations (BSs), distinct computing servers, and heterogeneous resources. Second, diversified and dynamic service requests with various QoS requirements in terms of latency, throughput and security, need to be served efficiently. Third, high dynamics and multi-dimensional randomness in wireless networks pose great challenges for service provisioning, e.g., time-changing wireless channel conditions, spatial-temporal fluctuation of traffic, varying workloads of edge servers, and burst user requests. Last but not least, this heterogeneous network will incur considerable energy consumption for communication and computing, resulting in high operating expenses (OPEX). This research program aims at addressing the fundamental research challenges by developing novel network management solutions in future wireless heterogonous networks with convergence of communication and computing. This innovative research program will mainly investigate three research themes: i) Scalable computational task scheduling and resource management in wireless heterogeneous networks with QoS guarantee; ii) sustainable and adaptive network control; and iii) dynamic network slicing and function orchestration for heterogeneous service provisioning. By applying optimization, artificial intelligence, stochastic processes, Markov decision process, game theory, etc., it is anticipated that this research program will develop a new set of systematic solutions and methodologies for resource management and service provisioning in wireless heterogeneous networks with ubiquitous computing. It will also contribute to the training of highly qualified personnel in wireless networks and edge computing, and enable them to contribute to the Canadian high technology industry.