ID 原文 译文
14345 首先分析了AC( Aho-Corasick) 快速统计算法及基于位置差的长序列拼接算法，并指出了这两个算法存在的不足。 The Aho-Corasick( AC) fast statistical algorithm and thelong sequence splicing algorithm based on position difference are analyzed，and the shortcomings of whichare pointed out．
14346 采用数组代替二叉树存储比特流中模式序列的位置信息，并通过构造使得数组下标值与二叉树节点值保持一致的关系式，有效降低了对模式序列进行计数及筛选的时间复杂度。 An array instead of the binary tree is used to store the position information of the patternsequence in the bit stream，and a relational expression is constructed to keep the value of the array subscript value consistent with that of binary tree node，which reduces the time complexity of counting and filtering the pattern sequence effectively．
14347 进一步，将数组元素依照对应频繁序列出现位置的先后顺序从左到右进行重新排列，提出了基于位置差的特征序列挖掘改进算法。 Furthermore，the array elements are rearranged from left to right according to the order in which the corresponding pattern sequences appear，and an improved algorithm basedon position difference is proposed．
14348 最后，使用 Wireshark 截取 100 个地址解析协议( Address Ｒesolution Protocol，AＲP)数据包进行仿真验证。 Finally，Wireshark is used to intercept 100 Address Ｒesolution Protocol( AＲP) packets for simulation verification．
14349 结果表明，与原算法相比，改进算法的时间复杂度至少可降低一个数量级。 The results show that the time complexity of the improved algorithm can be reduced by at least an order of magnitude compared with that of the original algorithm．
14350 针对物联网终端设备类型复杂、数量繁多，传统的识别方法难以对终端进行精准识别的问题，提出了一种新的终端自动识别方法。 For the problem that the types of Internet of Things( IoT) terminals are complex and numerous，and the traditional methods are difficult to accurately identify the terminal，a new automatic terminal identification method is proposed．
14351 该方法利用 ZMap 工具在网络中进行扫描，对扫描出的主机发送“get”请求以获取终端的登录页面，并根据超文本标记语言源代码、超文本传输协议中响应头字段和导航栏信息构建终端特征数据库，并利用长短期记忆( Long Short-Term Memory，LSTM) 算法生成三级分类器，实现对终端进行类型、品牌及版本的自动识别。 This method uses the ZMap tool to scan the network and sends a“get”requestto the scanned host to obtain the login page of the terminal． Based on the source code of Hyper Text MarkupLanguage( HTML) and the response header fields and navigation bar information in Hyper Text TransferProtocol( HTTP) ，an IoT terminal feature database is established． In the meantime，the long short-termmemory( LSTM) algorithm is used to generate a three-level classifier to realize automatic identification ofthe terminal type，brand and version．
14352 实验结果表明，该方法在 F1-score 上表现出优异的结果，最高可达 98． 4% ，对于具体品牌“TP-Link”交换机识别的准确率最高为 99% ，召回率最高为 98． 7% 。由此可见，该方法具有精准识别终端属性的优点。 The experimental results illustrate that this method has excellent results on F1-score，with a value of up to 98． 4% ． The recognition accuracy rate of the specific brand“TPLink”switch is up to 99% ，and the recall rate is up to 98． 7% ． It can be seen that this method has the advantage of accurately identifying terminal attributes．
14353 为了有效利用边缘云的计算资源，尽可能降低任务卸载时的平均等待时延，提出了一种满足边缘计算服务器容限阈值和任务卸载成功率约束条件下的多个边缘计算服务器相互协作的资源分配方案，通过单位时间总代价指标优化边缘计算服务器个数。 To efficiently utilize the computing resources of edge cloud and reduce the average waiting delayfor task offloading as much as possible，this paper proposes a resource allocation scheme in which multipleedge computing servers cooperate with each other when edge computing servers’threshold and task offloading success rate satisfy constraints． The number of edge computing servers is optimized by the total cost perunit time.
14354 将此方案建模为一个整数优化问题，之后设计了一种最小代价算法求解此优化问题，得到约束条件下的单位时间总代价以及任务平均等待时延。 This scheme is modeled as an integer optimization problem． Then，the minimum cost algorithm isdesigned to solve it and obtain the total cost per unit time along with the average waiting delay under theconstraints．