Large Scale Optimization with Applications in Communication Networks

我的研究项目是大规模优化方法的理论研究，并将其应用于通信网络和人工智能中出现的不同工程问题。它建立在过去6年取得的一些成果的基础上，有些成果涉及国家或国际合作。有三个研究轴线：(I)分解方法和元启发式的理论进展；(Ii)在通信网络中出现的问题的应用，例如混合线路速率光网络的设计，通过例如能量高效和可生存的逻辑拓扑的能效设计和管理，多播业务的提供和(Iii)在人工智能中的应用。第一个轴的动机是需要解决越来越大的优化问题，并保证生成的解的质量。第二个轴的应用与互联网流量的持续增长以及按需应用的稳步增加有关，这导致了通信网络中具有挑战性的优化问题。解决方案总是需要考虑一些现有技术的容量限制(例如，电子路由器的处理速度与要交换/路由的业务量)。此外，必须控制能源消耗，以避免未来几年不可持续的电力需求。事实上，技术和运营效率的进步速度必须快于交通增长的速度。在运营效率方面，我们建议优化混合线速网络的设计和管理，提高带宽利用率以降低能耗，优化组播流量的提供和保护。

My research program deals with theoretical research on large scale optimization methods with applications to different engineering problems arising in communication networks and artificial intelligence. It is built upon some of the results obtained in the previous 6 years, with some involving national or international collaborations. There are three research axes: (i) theoretical advances on decomposition methods and meta-heuristics; (ii) applications to problems arising in communication networks such as the design of mixed-line-rate optical networks, energy-efficient designs and management through, e.g., energy-efficient and survivable logical topologies, provisioning of multicast traffic and (iii) applications in artificial intelligence. The first axis is motivated by the need to solve ever larger optimization problems, with guarantees on the quality of the generated solutions. The applications of the second axis are related to the continued growth of Internet traffic, and the steady increase in the usage of on-demand applications which leads to challenging optimization problems in communication networks. Solutions always need to take into account the capacity limits for some of the existing technology (e.g., processing speed of the electronic routers vs. amount of the traffic to be switched/routed). As well, energy consumption must be tamed in order to avoid unsustainable power requirements in the next coming years. Indeed, the rate of advances in technological and operational efficiency must increase faster than the rate of traffic growth. In terms of operational efficiency, we propose to optimize the design and management of mixed-line rate networks, to enhance the bandwidth usage in order to reduce the energy consumption, and to optimize the provisioning and the protection of the multicast traffic.