Adversarial Design Framework for Self-Driving Networks (ADVISE)

点击翻译按钮获取中文摘要

The networking community is currently engaged in designing more automated and ``self-driving'' communication networks that overcome today's manually managed networks. These networks exploit the flexibilities introduced by emerging software-defined and virtualized communication technologies, to implement more demand-aware networks which meet the stringent requirements of new applications arising, e.g., in the context of 5G applications, such as low-latency tele-operation or high-bandwidth machine-to-machine type communication.This project proposes a network design framework relying on new methodologies to realize the vision of such self-driving networks, by combining adapted machine learning and artificial intelligence with approaches providing formal correctness and performance guarantees. We consider the lack of rigorous guarantees by existing algorithms based on artificial intelligence as one of the key issues which can prevent the adoption of self-driving networks: communication networks have become a critical infrastructure of our digital society and hence need to provide dependability and deterministic performance guarantees. In particular, we consider adversarial and game-theoretic approaches to test and optimize networks, to leverage the performance benefits from machine learning approaches while at the same time provide rigorous worst-case guarantees.The PIs are well-prepared for this project, through the unique combination of expertise in machine learning and network algorithms, as also demonstrated in their recent joint publications leading to this project.