Service Provisioning and Resource Management for Next-Generation Heterogeneous Wireless Networks

未来的无线网络，如5G及以后，预计将以异构网络架构、密集部署的业务接入点和具有不同服务质量（QoS）要求的多样化服务请求为特征。未来无线网络的这些特征将基于网络基础设施、资源和控制的进步。在基础设施一级，地面和卫星通信的一体化预计将在全球范围内提供无处不在的服务。在资源层面，通信、缓存和计算资源的整合，特别是在无线网络的边缘，有望通过拉近服务与用户的距离，缓解网络数据流量爆炸式增长带来的压力。在网络控制方面，网络切片和网络功能虚拟化作为一种很有前途的资源管理方式，可以满足多样化的业务需求。然而，要实现这一未来无线网络的愿景，仍然存在许多挑战。首先，由于实际边缘网络中业务需求的多样化和动态性，自适应和可扩展的网络切片和网络功能部署变得非常具有挑战性。此外，现有的研究并没有提供足够的解决方案，在一个综合的地面卫星系统中，服务交付需要一个复杂的决策过程，考虑成本，用户QoS和网络负载平衡。最后但并非最不重要的一点是，由于用户服务需求可能是不确定的，网络先验地不知道，因此有效的网络资源管理需要预测用户请求并主动联合分配通信、缓存和计算资源。因此，我们的研究计划将研究三个研究主题：综合地面卫星系统中的合作服务交付，异构边缘网络中的可扩展和自适应网络切片，以及基于机器学习的服务请求和联合资源管理。利用人工智能、机制设计和优化的强大工具，我们的研究将有助于下一代异构无线网络中智能可靠的服务提供和资源管理。

Future wireless networks, e.g., 5G and beyond, are expected to feature heterogeneous network architecture, dense deployment of service access points, and diverse service requests with various Quality of Service (QoS) requirements. These characteristics of future wireless networks will be based upon advancements in network infrastructure, resources and control. At the infrastructure level, the integration of terrestrial and satellite communications is anticipated to provide ubiquitous service coverage over the globe. On the level of resources, the integration of communication, caching and computing resources, especially on the edge of the wireless network, is expected to alleviate pressure from explosive growth in network data traffic by bringing services closer to users. In regards to network control, network slicing and network function virtualization are emerging as promising resource management approaches to fulfill diversified service requirements. Nevertheless, many challenges toward achieving this vision of future wireless networks remain. First, adaptive and scalable network slicing and network function deployment become very challenging due to diversified and dynamic service requests in a practical edge network. Moreover, existing research does not offer adequate solutions for service delivery in an integrated terrestrial-satellite system that demands a complex decision making process that considers cost, user QoS, and network load balancing. Last but not the least, as user service demands can be uncertain and unknown to the network a priori, effective network resource management calls for the prediction of user requests and proactive joint allocation of communication, caching, and computing resources. Accordingly, our research program will investigate three research topics: cooperative service delivery in integrated terrestrial-satellite systems, scalable and adaptive network slicing in heterogeneous edge networks, and machine learning based service request and joint resource management. With powerful tools from artificial intelligence, mechanism design, and optimization, our proposed research will contribute to intelligent and reliable service provisioning and resource management in next-generation heterogeneous wireless networks.