With increasing digitization and the emergence of the Internet of Things, more and more devices communicate with each other, resulting in a drastic growth of communication networks. Consequently, managing these networks, too, becomes harder and harder. Thus, Software-defined Networking (SDN) is employed, simplifying the management and configuration of networks by introducing a central controlling entity, which makes the network programmable via software and ultimately more flexible. As the SDN controller may impose scalability and elasticity issues, distributed controller architectures are utilized to combat this potential performance bottleneck. However, these distributed architectures introduce the need for constant synchronization to keep a centralized network view, and controller instances need to be placed in appropriate locations. As a result, thoroughly designing SDN-enabled networks with respect to a multitude of performance metrics, e. g., latency and induced traffic, is a challenging task. To assist in this process, we train a performance prediction model based on properties which are available during the network planning phase. We utilize a simulation-based approach for data collection to cover a large parameter space, simulating a variety of networks and controller placements for two opposing SDN architectures. On basis of this dataset, we apply Machine Learning (ML) to solve the performance prediction as a regression problem.
Wide Area Network (WAN) research benefits from the availability of realistic network topologies, e. g., as input to simulations, emulators, or testbeds. With the rise of Machine Learning (ML) and particularly Deep Learning (DL) methods, this demand for topologies, which can be used as training data, is greater than ever. However, public datasets are limited, thus, it is promising to generate synthetic graphs with realistic properties based on real topologies for the augmentation of existing data sets. As the generation of synthetic graphs has been in the focus of researchers of various applications fields since several decades, we have a variety of traditional model-dependent and modelindependent graph generators at hand, as well as DL-based approaches, such as Generative Adversarial Networks (GANs). In this work, we adapt and evaluate these existing generators for the WAN use case, i. e., for generating synthetic WANs with realistic geographical distances between nodes. Moreover, we investigate a hierarchical graph synthesis approach, which divides the synthesis into local clusters. Finally, we compare the similarity of synthetic and real WAN topologies and discuss the suitability of the generators for data augmentation in the WAN use case.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.