基于WIFI信息熵特征分析的室内人员行为监测技术

Indoor Personnel Behavior Monitoring Technology Based on WIFI Information Entropy Feature Analysis

现有的室内监控技术存在一些限制,例如盲区和高成本.相反,无线传感技术因其非视距和被动式的优点受到广泛研究.现有的基于无线信号的检测方法大多采用信号强度和信道状态信息(CSI)的测量值.然而,实验证明从同一设备接收的信号随不同天线有差异.计算出不同天线间相位差时频图的熵值可以作为有效特征提高检测率.利用无线设备在3种常见室内环境中采集的数据提取相位差特征,并与现有的特征提取方法进行比较.同时选取支持向量机(SVM)和神经网络分类方法对比.结果表明,该方法具有较高的准确性和鲁棒性.

There are some limitations among existing indoor monitoring technologies,such as blind areas and high costs.On the contrary,wireless sensing technology has been widely researched for its nonline of sight and passive.The existing detection methods based on wireless signals mostly adopt the measured values of signal strength and channel state information(CSI).However,the experiment proves that signals received from the same device vary with different antennas.The calculated entropy value of the phase difference of the time-frequency map between different antennas can be used as effective feature to improve the detection accuracy.The phase difference features extracted from the data collected through the wireless devices in three common indoor environments were used to make a comparison with existing feature extraction methods.Meanwhile Support Vector Machine(SVM)and neural network classification methods were selected for comparison.The results show that the accuracy and robustness are higher.