| 摘要: |
| 针对通信协议未知或部分已知条件下询问和应答信号配对难、配对准确度不高等问题，利用询问与应答信号间时序的模式规律，提出了一种询问应答信号配对规则提取方法，实现了问答信号精准配对。以非合作侦察获得的询问和应答信号特征参数时间序列为基础，首先根据询问、应答节点定位信息和侦察节点位置信息估计时间窗，在此基础上以询问信号到达时间为起始，在应答信号序列中搜索时间窗范围内的应答信号，交叉构建询问-应答信号2-项候选项集；再采用DBSCAN(Density-based Spatial Clustering of Applications with Noise)算法对候选项集进行特征聚类并挖掘频繁项集，通过频繁项集中问答信号频率、到达时间、脉冲宽度等参数，提取信号时序和特征配对规则;最后利用规则进行问答信号配对。计算机仿真和实采数据验证了该方法的有效性，当设置时间窗大小符合信号传播时间差且规则偏差小于50%时，置信度可达90%以上。 |
| 关键词: 敌我识别 协同问答 问答配对 时间序列 频繁项集 |
| DOI： |
|
| 基金项目: |
|
| A pairing rule mining method of cooperative question and answer signals |
| LONG Huimin,YU Bo,FENG Zhibin |
| (Southwest China Institute of Electronic Technology,Chengdu 610036,China;Unit 93114 of PLA,Beijing 100195,China;Unit 93147 of PLA,Chongqing 402760,China) |
| Abstract: |
| For the requirement of pairing mining of cooperative question&answer(Q&A) signals under the condition of unknown communication protocol,a pairing rule mining method based on Q&A signal timing pattern mining is proposed.Based on the time series of characteristic parameters of Q&A signals obtained by non-cooperative reconnaissance,the time window is estimated according to the location information of question node,answer node and reconnaissance node.On this basis,by taking the arrival time of question signals as start,the answer signals are searched within the time window in the answer signal sequence and cross-construction of the Q&A signal 2-candidate set is performed.Then, the Density-based Spatial Clustering of Applications with Noise(DBSCAN) algorithm is used to cluster the characteristics of the candidate set and mine the frequent itemset,and signal timing and feature pairing rules are extrcted through the Q&A signal frequency,arrival time,pulse width and other parameters in the frequent itemset.Finally, the rules are applied to Q&A signal pairing.The effectiveness of this method is verified by computer simulation and real data.When the time window size is set to conform to the signal propagation time difference and the rule deviation is less than 50%,the confidence can reach more than 90%. |
| Key words: identification of friend or foe(IFF) cooperative Q&A;Q&A pairing time sequence frequent itemsets |
