Resource allocation for Edge Computing in Next Generation Wireless Networks

无线网络在满足近期和未来应用的性能要求方面面临着前所未有的挑战。尽管通过管理稀缺的通信资源来提高无线通信能力的研究正在进行，但目前的无线网络基础设施在处理用户日益增长的期望和需求方面遇到了困难。幸运的是，与基本上有限的频谱和电力资源不同，无线网络可以利用越来越丰富、低成本和可扩展的计算和内存资源。因此，在网络边缘集成缓存和计算能力成为无线网络可持续发展的重要关键之一。事实上，这种集成可以改善许多系统性能指标，例如频谱和能源效率、端到端延迟、运营和资本支出以及用户体验质量。

Wireless networks are facing unprecedented challenges in fulfilling the performance requirements of recent and future applications. Despite ongoing research efforts in enhancing wireless communication capabilities through the management of scarce communication resources, current wireless networking infrastructures encounter difficulties to deal with the users' increasing expectations and needs. Fortunately, unlike the fundamentally limited spectrum and power resources, wireless networks can make use of computing and memory resources that are more and more abundant, low-cost and scalable. Therefore, the integration of caching and computing capabilities at the network edge becomes one of the important keys for wireless networks' sustainability. In fact, this integration results in improving numerous system performance metrics, such as spectral and energy efficiency, end-to-end latency, operating and capital expenditures and users' quality of experience. The general objective of this research program is to optimize wireless network performance by proposing innovative approaches for the allocation of communication, caching and computing resources available at the wireless network edge. More specifically, novel resource allocation schemes will be proposed and evaluated in the context of several emerging wireless networking architectures, taking into account the new constraints, opportunities and challenges dictated by the interaction of communications, caching and computing. Specifically, we will consider two emerging technologies, namely mmWave and vehicle-to-everything communications. To this end, theoretical tools such as those from combinatorial and continuous optimization, game theory and machine learning theory will be used to formulate and solve centralized and distributed resource allocation problems. Optimal and benchmark schemes will be first developed before designing low complexity heuristic and approximation algorithms. The performance of the proposed schemes will be then assessed using both analytical and simulation tools. The integration of caching and computing capabilities at the network edge brings various benefits for many network actors. For instance, network providers benefit from a considerable reduction in the backhaul congestion, whereas mobile device users see their quality of experience enhanced and their battery life prolonged. Therefore, the proposed research is expected to give a better understanding of the benefits resulting from the integration of caching and computing at the edge by providing theoretical and simulation results and the design of especially adapted algorithmic solutions. The anticipated research results will be of interest for various Canadian wireless networks industry actors. This research program will prepare several undergraduate and graduate students to meet the needs of the Canadian industry for highly qualified personnel in emerging areas of wireless communications and networks.