The papers in this special section focus on cognitive software defined networks and applications. Next generation networks (NGNs) are xpected to utilize internal and external sources of data through information and wireless communication techniques. Particularly, the demand for autonomic network management, orchestrations and optimization is as intense as ever, even though significant research has been needed. Software Defined Networks (SDNs) have been proposed to address QoS requirements for NGNs including high throughput, high mobility, low latency, heterogeneity and scalability. SDN has improved the user experience by providing high-performance communications between the network nodes, reconstructing the network structure, and optimizing the networking coverage, system security, communication latency, etc. The control intelligence is moved out of devices in a logically centralized controller, which interacts with data plane devices through standard interfaces. However, the existing applications in the SDN attract more attention to develop new learning algorithms, enhanced protocols and is even used in sensor power line communication for data transmission. The Cognitive Learning algorithms are the best solution to some particular applications. The cognitive software defined network (CSDN) presents to combine the efficiencies of SDN with new cognitive learning algorithms and enhanced protocols to automatize SDN. Its research and implementation are based on autonomic network management and control concepts. Such a combination of SDNs with autonomic frameworks and cognitive algorithms is better to solve the issues of traditional SDNs. This architecture of CSDN enables up-to-date control schemes to be developed and deployed to enable new smart networking services.

中文翻译：

---

客座社论介绍认知软件定义的网络和应用程序特别节

本节中的论文重点介绍认知软件定义的网络和应用程序。下一代网络（NGN）有望通过信息和无线通信技术利用内部和外部数据源。尤其是，即使需要大量研究，对自治网络管理，业务流程和优化的需求也与以往一样强烈。已经提出了软件定义网络（SDN）来解决NGN的QoS要求，包括高吞吐量，高移动性，低延迟，异构性和可伸缩性。SDN通过在网络节点之间提供高性能通信，重构网络结构以及优化网络覆盖范围，系统安全性，通信等待时间等，改善了用户体验。将控制智能移出逻辑集中控制器中的设备，该控制器通过标准接口与数据平面设备进行交互。但是，SDN中的现有应用吸引了更多的注意力来开发新的学习算法，增强的协议，甚至被用于传感器电源线通信中以进行数据传输。认知学习算法是某些特定应用程序的最佳解决方案。认知软件定义网络（CSDN）提出将SDN的效率与新的认知学习算法和增强的协议相结合，以实现SDN的自动化。它的研究和实现基于自主网络的管理和控制概念。SDN与自主框架和认知算法的这种结合更好地解决了传统SDN的问题。