ID 原文 译文
4593 最后给出智慧城市、工业互联网等区块链前沿应用方向，提出区块链技术挑战与研究展望。 Finally, some application directions of blockchain frontiers, technology challenges and research prospects includingSmart Cities and Industrial Internet were given.
4594 为了满足多样化的业务需求，基于架构按需节点和多核光纤的软件定义多维光网络将实现可定制化的网络服务。 To meet diverse business needs, both architecture on demand (AoD) nodes and multi-core fibers (MCF) basedsoftware-defined multi-dimensional optical networks will enable customizable network services.
4595 针对多维网络资源管控问题，首先，介绍了软件定义多维光网络的架构； Aiming at the problem of management and control for multi-dimensional network resources, the architecture of software-defined mul-ti-dimensional optical networks was introduced firstly.
4596 接着，评述了软件定义多维光网络的使能技术； Then, the enabling technologies to support the implementation ofsoftware-defined multi-dimensional optical networks were presented.
4597 最后，展望了软件定义智能多维光网络的发展方向。 Finally, the development direction of soft-ware-defined intelligent multi-dimensional optical networks was proposed and analyzed.
4598 针对人工智能（AI）技术在光通信网络中的应用进行探讨。 The applications of artificial intelligence (AI) technique in optical communication networks were explored.Some representative AI applications and potential risks due to the failure of the AI technique were discussed.
4599 主要介绍了 AI 技术代表性应用和光通信网络开放性导致的 AI 技术失效的潜在风险，并针对这些风险提出了一些应对策略，主要包括通过单元化小型化的AI 系统建模，与传统经典网络建模和规划方法相结合，提高 AI 技术的有效性和可解释性。 To address these risks, methods including systematic AI modeling through unitizing and miniaturizing sub-systems and cooperationwith traditional network modeling and planning methods were proposed, which were expected to help improve the effec-tiveness and practicality of the application of the AI technique.
4600 同时，针对 AI 技术可能失效和遭受攻击提出了基于网络保护的应对策略。 Finally, to recover a system from the failure of its em-ployed AI technique or attacks, some protection strategies were proposed.
4601 为了全面有效地识别 DNS 隐蔽信道，对多种 DNS 隐蔽信道软件的实现方式进行了研究，提出了一种基于改进的卷积神经网络的 DNS 隐蔽信道识别方法。 In order to effectively identify the multiple types of DNS covert channels, the implementation of different sortsof DNS covert channel software was studied, and a detection based on the improved convolutional neural network was proposed.
4602 基于真实的校园网流量进行了实验，结果表明，所提方法可检测出全部 22 种数据交互模式的 DNS 隐蔽信道，并且具有识别未知的 DNS 隐蔽信道流量的能力。 The experimental results, grounded upon the campus network traffic, show that the detection can identifytwenty-two kinds of data interaction modes of DNS covert channels and is able to identify the unknown DNS covertchannel traffic.