Benefits of Deep Reinforcement Learning for Resource Allocation in Optical Networks

在这个项目中，将研究动态弹性光网络中的几个资源分配问题，目的是确定DRL代理可以在保持可接受的计算复杂性的情况下提供相对于现有解决方案的显著优势的场景和配置。主要研究目的是了解在光网络中的资源分配问题中给予DRL优势的特征：1.相对于迄今已知的最佳启发式，识别使DRL方法在解决光网络中的资源分配问题方面具有优势的特征；以及2.识别允许相对于当前解决方案改进DRL代理的性能的资源分配问题的基本参数。

In this project several resource allocation problems in dynamic elastic optical networks will be studied with the aim of identifying the scenarios and configurations where DRL agents can offer a significant advantage over existing solutions, whilst keeping an acceptable computational complexity. The main research objective is to understand the characteristics that give DRL an advantage in the problem of resource allocation in optical networks: 1. Identify the characteristics that give a DRL approach an advantage in solving resource allocation problems in optical networks with respect to the best heuristics known to date; and 2. Identify the fundamental parameters of the resource allocation problems that allow for improved performance of DRL agents with respect to current solutions.