CAJ | 학술논문

针对传统单节点调制识别存在阴影和多径问题,提出了一种利用无线传感器网络进行分布式协同调制识别的方法。首先由相互协作的传感器节点提取信号的五个典型特征构成特征矢量,将特征矢量发送到中心节点,在中心节点通过粒子群算法对人工神经网络进行参数优化,最后利用训练好的神经网络进行分类识别,得到识别结果。对6种信号进行了仿真,结果表明该方法在信噪比大于5 dB且测试样本数大于50时,最终的识别率超过了90%,验证了将粒子群算法应用于神经网络参数优化进行分类识别可以有效提升识别性能。
침대전통단절점조제식별존재음영화다경문제,제출료일충이용무선전감기망락진행분포식협동조제식별적방법。수선유상호협작적전감기절점제취신호적오개전형특정구성특정시량,장특정시량발송도중심절점,재중심절점통과입자군산법대인공신경망락진행삼수우화,최후이용훈련호적신경망락진행분류식별,득도식별결과。대6충신호진행료방진,결과표명해방법재신조비대우5 dB차측시양본수대우50시,최종적식별솔초과료90%,험증료장입자군산법응용우신경망락삼수우화진행분류식별가이유효제승식별성능。
The traditional modulation identification with single node has problems of shadow effect and multipath fading, in view of which a kind of distributed collaborative modulation identification method based on wireless sensor network is presented. Firstly, characteristic vectors are formed by five typical signal characteristics extracted by the mutual cooperative sensor nodes and then sent to the central node. Secondly, parameters of artificial neural network are optimized by the particle swarm algorithm in the cen- tral node. Finally, classification is carried on by the trained neural network and result is got. 6 kinds of signals are simulated, and the results show that the final recognition rate of this method is more than 90% under the condition that the signal-to-noise ratio is higher than 5 dB and test samples are more than 50. It illustrates that identification performance can be improved effectively when particle swarm algorithm is used in parameters optimization of neural network.