Efficient Resource Allocation in Software Defined Wireless Networks

未来几代蜂窝网络将发生重大变化，以应对日益增长的网络用户数量以及他们对高数据速率业务的要求，包括低延迟要求在内的高服务质量。此外，蜂窝网络的能源消耗约占全球总能源消耗的0.5%。环境和经济要求提高能源效率，保持或降低能源消耗水平，同时将数据速率提高至少1000倍。在该项目中，我们开发了一种灵活的、流量和密度感知的新型无线接入网架构。该体系结构很好地反映了大规模MIMO基站、非正交多址（NOMA）的最新发展，并充分利用了软件定义无线电和网络的特点。在方法上，该项目将相应的无线电资源管理任务制定为可靠的编程问题，并以集中或分布式实现的方式解决它们。如果相应的资源分配支持，毫米波、WiFi和蜂窝等多种网络技术的共存可以为用户提供多种无缝连接。该方案还考虑了不完全信道和交通状态信息下的鲁棒资源分配。采用分数规划、凸序列逼近、半定松弛等最新优化工具求解多目标规划问题。从性能、复杂性和收敛性等方面分析了所有解决方案。开发的算法将为5G及以后网络的软件实现奠定基础。

Future generations of cellular networks will change significantly in order to cope with the increasing number of network users as well as their request for high data rate services with high quality of service including low delay requirements. Additionally, the energy consumptions of cellular networks is about 0.5% of the total energy consumption world wide. Environment and economics require to increase the energy efficiency and keep or lower the level of energy consumption while increase the data rate by a factor of at least 1000. In the project, we develop a new radio access network architecture which is flexible and traffic and density aware. The architecture reflects well the recent developments in terms of massive MIMO base stations, non-orthogonal multiple access (NOMA) and exploits the features of software defined radio and networking. Methodologically, the project formulates the corresponding radio resource management tasks as solid programming problems and solves them either centralized or in a distributed implementation. Coexistence of several network technologies as mmWave, WiFi and cellular allows to supply users with multiple seamless connectivity if the corresponding resource allocation supports it. The project also considers robust resource allocation for imperfect channel and traffic state information. The multi-objective programming problems are solved by using the latest optimization tools including fractional programming, convex sequential approximation, semidefinite relaxation. All solutions are analyzed in terms of performance, complexity and convergence. The developed algorithms will build a basis for software implementation in 5G and beyond networks.