No abstract

No abstract

Recent research on Software-Defined Networking (SDN) strongly promotes the adoption of distributed controller architectures. To achieve high network performance, designing a scheduling function (SF) to properly dispatch requests from each switch to suitable controllers becomes critical. However, existing literature tends to design the SF targeted at specific network settings. In this paper, a reinforcement-learning-based (RL) approach is proposed with the aim to automatically learn a general, effective, and efficient SF. In particular, a new dispatching system is introduced in which the SF is represented as a neural network that determines the priority of each controller. Based on the priorities, a controller is selected using our proposed probability selection scheme to balance the trade-off between exploration and exploitation during learning. In order to train a general SF, we first formulate the scheduling function design problem as an RL problem. Then a new training approach is developed based on a state-of-the-art deep RL algorithm. Our simulation results show that our RL approach can rapidly design (or learn) SFs with optimal performance. Apart from that, the trained SF can generalize well and outperforms commonly used scheduling heuristics under various network settings.

No abstract

No abstract

Cloud computing provides on-demand access to computational resources while outsourcing infrastructure and service maintenance. Edge computing could extend cloud computing capability to areas with limited computing resources, such as rural areas, by utilizing low-cost hardware, such as singleboard computers. Cloud data centre hosted machine learning algorithms may violate user privacy and data confidentiality requirements. Federated learning (FL) trains models without sending data to a central server and ensures data privacy. Using FL, multiple actors can collaborate on a single machine learning model without sharing data. However, rural network outages can happen at any time, and the quality of a wireless network varies depending on location, which can affect the performance of the Federated Learning application. Therefore there is a need to have a platform that maintains service quality independent of infrastructure status. We propose a self-adaptive system for rural FL, which employs the Greedy Nominator Heuristic (GNH) based optimisation to orchestrate application workflows across multiple resources that make up a rural computing environment. GNH provides distributed optimization for workflow placement. GNH utilises resource status to reduce failure risks and costs while still completing tasks on time. Our approach is validated using a simulated rural environment -composed of multiple decentralized controllers sharing the same infrastructure and running a shared FL application. Results show that GNH outperforms three algorithms for deployment of FL tasks: random placement, round-robin load balancer and simple greedy algorithm.

No abstract

Controller placement problem (CPP) is a key issue for Software-Defined Networking (SDN) with distributed controller architectures. This problem aims to determine a suitable number of controllers deployed in important locations so as to optimize the overall network performance. In comparison to communication delay, existing literature on the CPP assumes that the influence of controller workload distribution on network performance is negligible.In this paper, we tackle the CPP that simultaneously considers the communication delay, the control plane utilization, and the controller workload distribution. Due to this reason, our CPP is intrinsically different from and clearly more difficult than any previously studied CPPs that are NP-hard. To tackle this challenging issue, we develop a new algorithm that seamlessly integrates the genetic algorithm (GA) and the gradient descent (GD) optimization method. Particularly, GA is used to search for suitable CPP solutions. The quality of each solution is further evaluated through GD. Simulation results on two representative network scenarios (small-scale and large-scale) show that our algorithm can effectively strike the trade-off between the control plane utilization and the network response time.