Software-defined networking (SDN) is a networking architecture with improved efficiency achieved by moving networking decisions from the data plane to provide them critically at the control plane. In a traditional SDN, typically, a single controller is used. However, the complexity of modern networks due to their size and high traffic volume with varied quality of service requirements have introduced high control message communications overhead on the controller. Similarly, the solution found using multiple distributed controllers brings forth the 'controller placement problem' (CPP). Incorporating switch roles in the CPP modelling during network partitioning for controller placement has not been adequately considered by any existing CPP techniques. This article proposes the controller placement algorithm with network partition based on critical switch awareness (CPCSA). CPCSA identifies critical switch in the software defined wide area network (SDWAN) and then partition the network based on the criticality. Subsequently, a controller is assigned to each partition to improve control messages communication overhead, loss, throughput, and flow setup delay. The CPSCSA experimented with real network topologies obtained from the Internet Topology Zoo. Results show that CPCSA has achieved an aggregate reduction in the controller's overhead by 73%, loss by 51%, and latency by 16% while improving throughput by 16% compared to the benchmark algorithms.
With the rapid development of Internet of Things (IoT) technology, Terminal Devices (TDs) are more inclined to offload computing tasks to higher-performance computing servers, thereby solving the problems of insufficient computing capacity and rapid battery consumption of TD. The emergence of Multi-access Edge Computing (MEC) technology provides new opportunities for IoT task offloading. It allows TDs to access computing networks through multiple communication technologies and supports more mobility of terminal devices. Review studies on IoT task offloading and MEC have been extensive, but none of them focus on IoT task offloading in MEC. To fill this gap, this paper provides a comprehensive and in-depth understanding of the algorithms and mechanisms of multiple IoT task offloading in the MEC network. For each paper, the main problems solved by the mechanism, technical classification, evaluation methods, and supported parameters are extracted and analyzed. Furthermore, shortcomings of current research and future research trends are discussed. This review will help potential and new researchers quickly understand the panorama of IoT task offloading approaches in MEC and find appropriate research paths.
The proliferation of the Internet of Things (IoT) devices has led to a surge in Internet traffic characterized by variabilities in Quality of Service (QoS) demands. Managing these devices and traffic effectively proves challenging, particularly within conventional IoT network architectures lacking centralized management. However, the advent of Software-Defined Networking (SDN) presents intriguing opportunities for network management, capable of addressing challenges in traditional IoT architectures. SDN's ability to provide centralized network management through a programmable controller, separate from data forwarding elements, has led researchers to incorporate SDN features with IoT (SDIoT) and Wireless Sensor Networks (SDWSN) ecosystems. However, despite the SDN support, these networks encounter challenges related to load-imbalance routing issues, as the SDN controller may be constrained while certain access points serving end users become overloaded. In response to these challenges, various load-balancing routing solutions have been proposed, each with distinct objectives. However, a comprehensive study that classifies and analyzes these solutions based on their weaknesses and postmortem challenges is currently lacking. This paper fills this gap by providing an in-depth classification of existing solutions. The study categorizes the problems addressed by different schemes and summarizes their findings. Furthermore, it discusses the shortcomings of current studies, and postmortem challenges associated with integrating SDN with IoT, and suggests future research directions.