摘要
针对基于单步频谱预测的认知通信模型中预测和感知过程会占用大量的数据传输时间,设计基于多步预测的宽带频谱感知系统,提出基于多步预测的离线粗感知和基于滤波器组的宽带频谱在线细感知相结合的方法,介绍基于混合神经网络的多步频谱预测和宽带频谱感知的具体实现方法。通过仿真实验可得,预测后基于滤波器组的宽带频谱感知性能优于直接采用能量感知,预测和感知的子带空闲率接近于真实值,节省了感知时间。
For cognitive communication model based on single step spectrum prediction,the prediction and sensing process can take up a large amount of data transmission time.The wideband spectrum sensing system based on multi-step prediction was designed to save sensing time.A combination of off-line coarse sensing based on multi-step prediction and wideband spectrum online fine sensing based on filter bank was proposed.The realization methods of multi-step spectrum prediction based on hybrid neural network and wideband spectrum fine sensing were presented.The simulation results show that the performance of wideband spectrum sensing based on filter bank after multi-step prediction is better than energy detection,and the sub-channel idle rate of prediction and sensing is close to the real value.Furthermore,the sensing time is saved.
作者
王焕
吴斌
WANG Huan;WU Bin(Institute of Electronic Engineering,China Academy of Engineering Physics,Mianyang 621900,China;School of Information Engineering,Southwest University of Science and Technology,Mianyang 621010,China)
出处
《计算机工程与设计》
北大核心
2021年第7期1816-1823,共8页
Computer Engineering and Design
基金
四川省科技计划基金项目(2019YJ0309)
四川省教育厅基金项目(18ZB0611)
西南科技大学龙山学术人才科研支持计划基金项目(17LZX319)
西南科技大学龙山学术人才科研支持团队基金项目(17LZXT12)。
关键词
认知无线电
多步频谱预测
频谱粗感知
宽带频谱感知
滤波器组
cognitive radio
multi-step spectrum prediction
coarse spectrum sensing
wideband spectrum sensing
filter bank
