Optimization models and methods for telecommunication network design with varying and uncertain demands

为了经济地建设和运营现代电信网络，在规划阶段就必须更好地考虑到需求预测的不确定性和网络随时间的演变。该项目的目标是为这项任务开发数学模型和优化方法，这些方法远远优于目前使用的方法。首先，我们计划根据混合整数规划模型构建各种技术（IP/MPLS、SDH、OTH、WDM）的规划要求。对于这些模型，我们打算开发原始和对偶的解决方案的方法。我们的目标是能够计算出具有成本效益的网络，并为多种需求场景提供质量保证。理想情况下，所采用的方法为我们提供了信息，以证明最优的计算网络配置。在与网络运营商和设备供应商的合作中，我们打算定义和发布实际相关的参考场景，以评估我们的算法。我们的波兰项目合作伙伴将专注于单一时间点需求预测的不确定性，而ZIB将专注于需求随时间的演变。双方将共同研究基本模型的理论背景和参考情景的定义。最终，共同的计算研究计划调查不确定的需求预测和需求随时间的演变之间的相互作用。

To build and operate a modem telecommunication network economically, the uncertainty of demand forecasts and the evolution of the network over time must already be better taken into account during the planning phase. The goal of this project is to develop mathematical models and optimization methods for this task which strongly outperform those used today. First, we are planning to structure the planning requirements for various technologies (IP/MPLS, SDH, OTH, WDM) in terms of mixed integer programming models. For these models, we intend to develop primal and dual solution methods. The goal is to be able to compute cost-effective networks with a quality guarantee for several demand scenarios. Ideally, the employed methods provide us with information to prove optimality ofthe computed network configurations. In cooperation with network operators and equipment vendors, we intend to define and publish practically relevant reference scenarios in order to evaluate our algorithms. While our Polish project partners will concentrate on the uncertainty of demand forecasts at a single point in time, ZIB will focus on the evolution of demands over time. Both partners will work together on the theoretical background for basic models and on the definition of reference scenarios. Eventually, common computational studies are planned to investigate the interaction between uncertain demand forecasts and the evolution of demands over time.