基于蚁群算法的骨干网络发现  

Discovering the backbone network with a novel designed ant colony algorithm

作　　者：吕芳 柏军[1] 黄俊恒[1,2] 王佰玲 LYU Fang;BAI Jun;HUANG Junheng;WANG Bailing(School of Computer Science and Technology,Harbin Institute of Technology,Weihai 264209,China;Research Institute of Cyberspace Security,Harbin Institute of Technology,Weihai 264209,China)

机构地区：[1]哈尔滨工业大学(威海)计算机科学与技术学院,山东威海264209 [2]哈尔滨工业大学(威海)网络空间安全研究院,山东威海264209

出　　处：《通信学报》2020年第11期74-85,共12页Journal on Communications

基　　金：国家重点研发计划基金资助项目(No.2018YFB2004201);国家区域创新中心科技专项基金资助项目(No.2017QYCX14);中央高校基本科研业务费专项资金资助项目(No.HIT.NSRIF.2020098)。

摘　　要：针对交互网络中实体的非法、异常活动日趋隐蔽,复杂的交互关系又加剧了网络分析难度的问题,提出一种发现骨干网络的蚁群模型。该模型基于路径寻优理论模拟实体间交互关系,量化信息交互路径的显著性,实现网络规模约减。首先,利用网络中心性理论,提出了交互网络中蚂蚁初始位置选择策略;然后,设计了拟合信息交互行为的蚁群模型路径转移机制;最后,利用自适应的信息素动态更新机制引导信息流通路径优化。在真实金融交互网络上的实验结果表明,所提模型在最优解质量和性能上均优于传统蚁群算法,且相比于贪心算法具有更好的覆盖率和准确率。For the problem that in interactive network,the illegal and abnormal behaviors were becoming more hidden,moreover,the complex relation in real interactive network heightens the difficulty of detecting anomalous entities,an ant colony model was proposed for extracting the backbone network from the complex interactive network.The novel model simulated the relationships among entities based on the theory of path optimization,reduced the network size after quantifying the significance of each flow of information.Firstly,a strategy of initial location selection was proposed taking advantage of network centrality.Secondly,a novel path transfer mechanism was devised for the ant colony to fit the flow behavior of entities.Finally,an adaptive and dynamic pheromone update mechanism was designed for guiding the optimization of information flows.The experimental results show that the proposed model is superior to the traditional ant colony algorithm in both solving quality and solving performance,and has better coverage and accuracy than the greedy algorithm.

关 键 词：交互网络 蚁群算法 骨干网络 路径优化 动态信息素更新 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]