[1]岳文静,崔恒瑞,陈 志.基于卷积神经网络的自适应频谱感知模型[J].计算机技术与发展,2021,31(05):62-66.[doi:10. 3969 / j. issn. 1673-629X. 2021. 05. 011]
 ,AdaptiveSpectrum SensingModelBasedonConvolutionalNeuralNetwork[J].,2021,31(05):62-66.[doi:10. 3969 / j. issn. 1673-629X. 2021. 05. 011]
点击复制

基于卷积神经网络的自适应频谱感知模型()
分享到:

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
31
期数:
2021年05期
页码:
62-66
栏目:
图形与图像
出版日期:
2021-05-10

文章信息/Info

Title:
AdaptiveSpectrum SensingModelBasedonConvolutionalNeuralNetwork
文章编号:
1673-629X(2021)05-0062-05
作者:
岳文静1崔恒瑞1陈 志2
1.南京邮电大学通信与信息工程学院,江苏南京210023
2.南京邮电大学计算机学院,江苏南京210023
Author(s):
YUEWen-jingCUIHeng-ruiCHENZhi2
1.SchoolofTelecommunicationsandInformationEngineering,NanjingUniversityofPostsandTelecommunications,Nanjing210023,China
2.SchoolofComputer,NanjingUniversityofPostsandTelecommunications,Nanjing210023,China
关键词:
认知无线系统频谱感知模型匹配卷积神经网络信噪比估计算法
Keywords:
cognitivewirelesssystemspectrumsensingmodelmatchingconvolutionalneuralnetworkSNRestimationalgorithm
分类号:
TP391.41
DOI:
10. 3969 / j. issn. 1673-629X. 2021. 05. 011
摘要:
频谱感知作为认知无线电系统中重要的一环,传统的频谱感知算法在信道质量好的情况下检测概率较高,但是易受噪声影响,当信噪比较低时,检测性能急剧下降。针对传统频谱感知算法的优缺点,提出了基于卷积神经网络模型的频谱预测,提高了低信噪比情况下的检测概率,由于信道是时变的,加入信道感应和模型匹配,提升系统的可用性。将仿真的信号序列映射为RGB图片,将图片输入到卷积神经网络中训练得到模型,利用信噪比估计算法将认知用户接收的检测信号与模型进行匹配,通过训练好的匹配模型进行频谱感知。仿真实验表明:在低信噪比情况下,当虚警概率相同时,卷积神经网络模型比传统能量检测法的检测概率有大幅度的提升。甚至在能量检测法信号采样点比卷积神经网络模型多的情况下,卷积神经网络模型依然有着非常好的性能。在信号与模型匹配阶段,二阶-四阶信噪比估计算法在 -6dB到 -10dB,理论信噪比与估计信噪比基本一致。所提出的模型不但显著提高了检测的准确率,而且通过信噪比匹配提高了模型的自适应和高可用性。
Abstract:
Spectrumsensingisanimportantpartofcognitiveradiosystem.Traditionalspectrum sensingalgorithm hasahighdetectionprobabilityunderwellchannelquality,butitissusceptibletonoise.Whenthesignal-to-noiseratioislow,thedetectionperformancedropssharply.Inviewoftheadvantagesanddisadvantagesoftraditionalspectrumsensingalgorithm,aspectrumpredictionbasedonconvolutionalneuralnetworkmodelisproposedtoimprovethedetectionprobabilityunderlowsignal-to-noiseratio.Sincethechannelistime-varying,channelsensingandmodelmatchingareaddedtoimprovetheavailabilityofthesystem.ThesimulatedsignalsequenceismappedintoRGBimageswhichareinputintotheconvolutionalneuralnetworktotrainthemodel.Thesignal-to-noiseratioestimationalgorithmisusedtomatchthedetectionsignalreceivedbythecognitiveuserwiththemodel,andthespectrumperceptioniscarriedoutthroughthetrainedmatchingmodel.Simulationshowsthatinthecaseoflowsignal-to-noiseratio,whenthefalsealarmprobabilityisthesame,thedetectionprobabilityoftheconvolutionalneuralnetworkmodelisgreatlyimprovedcomparedwiththetraditionalenergydetectionmethod.Eveninthecasewheretheenergydetectionmethodhasmoresignalsamplingpointsthantheconvolutionalneuralnetworkmodel,theconvolutionalneuralnetworkmodelstillhasexcellentperformance.Inthesignalandmodelmatchingphase,thesecond-fourthordersignal-to-noiseratioestimationalgorithmrangesfrom-6dBto-10dB,andthetheoreticalsignal-to-noiseratioisbasicallythesameastheestimatedsignal-to-noiseratio.Theproposedmodelnotonlysignificantlyimprovestheaccuracyofdetection,butalsoimprovestheadaptabilityandhighavailabilityofthemodelthroughsignal-to-noiseratiomatching.

相似文献/References:

[1]郭云玮 刘全 高俊.不同衰落信道下的协作感知性能研究[J].计算机技术与发展,2011,(05):13.
 GUO Yun-wei,LIU Quan,GAO Jun.Performance of Cooperative Spectrum Sensing over Different Fading Channels[J].,2011,(05):13.
[2]汪晓睿 刘全.认知无线电网络中频谱感知安全的研究进展[J].计算机技术与发展,2011,(12):155.
 WANG Xiao-rui,LIU Quan.A Survey on Spectrum Sensing Security Issues in Cognitive Radio Networks[J].,2011,(05):155.
[3]宗平 刘柳 乔秀泉[].认知无线电技术在ZigBee中的应用研究[J].计算机技术与发展,2012,(08):241.
 ZONG Ping,LIU Liu,QIAO Xiu-quan.Application Research of Cognitive Radio Technology in ZigBee[J].,2012,(05):241.
[4]刘洋,季薇,侯晓赟.一种改进的基于 OMP 重建的宽带频谱感知算法[J].计算机技术与发展,2013,(01):99.
 LIU Yang,JI Wei,HOU Xiao-yun.A Modified Spectrum Sensing Algorithm for Wideband Cognitive Radio Based on OMP[J].,2013,(05):99.
[5]张学军,唐小丽,田峰.基于改进D-S证据理论的频谱感知算法[J].计算机技术与发展,2014,24(06):44.
 ZHANG Xue-jun[][],TANG Xiao-li[],TIAN Feng[].Spectrum Sensing Algorithm Based on Improved D-S Evidence Theory[J].,2014,24(05):44.
[6]岳文静,陆海亚,蒲昱,等. 基于协作频谱感知的多天线中继系统功率分配[J].计算机技术与发展,2015,25(11):218.
 YUE Wen-jing,LU Hai-ya,PU Yu,et al. Power Allocation in MIMO Relay System Based on Cooperative Spectrum Sensing[J].,2015,25(05):218.
[7]李红,齐丽娜. 认知无线电中基于感知时间和K秩准则的优化[J].计算机技术与发展,2015,25(12):186.
 LI Hong,QI Li-na. Joint Optimization Based on Sensing Time and K-rank Criteria in Cognitive Radio[J].,2015,25(05):186.
[8]胡 敏,朱 琦.基于微分博弈的群智频谱感知算法[J].计算机技术与发展,2021,31(03):138.[doi:10. 3969 / j. issn. 1673-629X. 2021. 03. 024]
 HU Min,ZHU Qi.Spectrum Crowd-sensing Algorithm Based on Differential Games[J].,2021,31(05):138.[doi:10. 3969 / j. issn. 1673-629X. 2021. 03. 024]
[9]郭晓敏,朱 琦.面向多任务的频谱感知博弈算法[J].计算机技术与发展,2023,33(06):109.[doi:10. 3969 / j. issn. 1673-629X. 2023. 06. 017]
 GUO Xiao-min,ZHU Qi.Spectrum Sensing Game Algorithm for Multi-task[J].,2023,33(05):109.[doi:10. 3969 / j. issn. 1673-629X. 2023. 06. 017]

更新日期/Last Update: 2020-05-10