基于峰度值估计的无线信号频谱感知方法

doi:10.12382/bgxb.2024.0441

摘要/Abstract

摘要：

在现代无线通信环境下,由于噪声环境复杂多变,且信号在不同感知周期内的占空比变化较大,导致信号频谱感知能力的下降,甚至造成非授权用户对授权用户的干扰。针对此问题,提出一种基于峰度值估计的智能无线信号频谱感知方法。以典型的非高斯噪声分布(McLeish分布)作为通用背景噪声,依据多尺寸跳跃连接的思想构建深度神经网络框架,结合注意力机制捕获目标信号的多尺寸特征,在感知周期内占空比不确定的条件下,完成对目标信号峰度值的估计,通过对估计值的判决,实现在不同噪声模型下对无线信号的感知。仿真结果表明:在信噪比SNR≥-10dB条件下,当虚警概率P_f=0.02,感知占空比0.5≤η<1时,其平均检测概率达到了84.3%以上;当噪声功率估计误差ε≤2、P_f=0.01时,其平均检测概率达到96.1%以上。证明所提方法具有较强的抗占空比和噪声功率不确定性的能力,具备一定的理论研究意义和工程实用价值。

关键词: 频谱感知, 卷积神经网络, 信号检测, McLeish分布

Abstract:

The complexity and variability of noise in modern wireless radio communication environment and the great diversity of the duty cycles of signals in different sensing periods cause the decline in the sensing ability of signal spectrums,even lead to interference to authorized users by unauthorized users.An intelligent wireless signal spectra sensing method based on the estimation of kurtosis is proposed to solve the above problems.A deep neural network framework is constructed based on the idea of multi-scale skip connections,which usestypical non-Gaussian noise distribution (McLeish distribution) as the general background noise.The multi-scale features of target signal are captured by means of the attention mechanism.The kurtosis value of target signal is estimated under the condition of uncertain duty cycle in the sensing period.The wireless signals under different noise models are sensed by judging the estimated value.The simulated results indicate that the average detection probability reaches over 84.3% when P_f=0.02 and duty cycle 0.5≤η<1 under the condition of SNR≥-10dB.And it reaches over 96.1% when noise power estimation error ε≤2 and P_f=0.01.It is proven that the proposed method has strong resistance to duty cycle and noise power uncertainty,and has certain theoretical research significance and engineering practical value.

Key words: spectrum sensing, convolutional neural network, signal detection, McLeish distribution

中图分类号:

TN97

王洋, 冯永新, 钱博, 宋碧雪. 基于峰度值估计的无线信号频谱感知方法[J]. 兵工学报, 2025, 46(7): 240441-.

WANG Yang, FENG Yongxin, QIAN Bo, SONG Bixue. Kurtosis-based Spectrum Sensing Method for Wireless Signals[J]. Acta Armamentarii, 2025, 46(7): 240441-.

图/表 21

图1 峰度值估计频谱感知框架设计

Fig.1 Design of spectra sensing framework for kurtosis estimation

图2 多尺寸特征拟合网络框架

Fig.2 Multidimensional feature fitting network framework

表1 目标感知信号的主要参数

Table 1 Main parameters of target sensing signals

调制信号	过采样率	扩频增益/dB	伪码长度
DSSS	4	18.06	4096
FHSS	4	18.06	4096
QPSK	4

表2 训练集和测试集的主要参数

Table 2 Main parameters of training and testing sets

数据集	样本总数	噪声参数	信噪比/dB	样本尺寸
训练集	19.2万	v=1,3,5,∞ σ²=1	-15~0	2×4096
测试集1	4.8万	v=1,3,5,∞ σ²=1,σ²≤2	-15~0	2×4096 2×1024
测试集2	4.8万	v=1,3,5,∞ 0.5≤η<1	-15~0	2×4096

表3 网络训练参数

Table 3 Network training parameters

优化器	学习率	权重衰减	训练次数	验证方法	早停策略
Adam	0.001	10^-4	50	十折交叉	是

表4 特征融合模块的特征尺寸

Table 4 Feature sizes of feature fusion module

模块名称	特征尺寸
模块名称	1×8	1×16	1×32	1×64	1×128
SSFF	2	3
BSFF				2	3
MSFF	1	1	1	1	1

图3 不同尺寸特征融合模块性能对比

Fig.3 Performance comparison of feature fusion modules with different sizes

表5 不同特征融合模块平均检测概率对比

Table 5 Comparison of detection probabilities of different feature fusion modules %

模块名称	平均检测概率
模块名称	v→∞ SNR≤-12	v→∞ SNR>-12	v=3 SNR<-13	v=3 SNR≥-13
SSFF	61.64	96.01	72.90	99.59
BSFF	67.77	93.40	81.05	96.75
MSFF	66.44	95.00	79.57	99.49

图4 不同信号在不同信噪比下检测概率对比

Fig.4 Comparison of detection probabilities for different signals under different signal-to-noise ratios

图5 不同信噪比下平均检测概率对比

Fig.5 Comparison of average detection probabilities under different signal-to-noise ratios

图6 不同虚警概率下平均检测概率对比

Fig.6 Comparison of average detection probabilities under different false alarm probabilities

图7 不同信噪比下平均检测概率对比

Fig.7 Comparison of average detection probabilities under different signal-to-noise ratios

图8 不同虚警概率下平均检测概率对比

Fig.8 Comparison of average detection probabilities under different false alarm probabilities

图9 不同信噪比下平均检测概率对比

Fig.9 Comparison of average detection probabilities under different signal-to-noise ratios

图10 不同虚警概率下平均检测概率对比

Fig.10 Comparison of average detection probabilities under different false alarm probabilities

图11 不同信噪比下平均检测概率对比

Fig.11 Comparison of average detection probabilities under different signal-to-noise ratios

图12 不同虚警概率下平均检测概率对比

Fig.12 Comparison of average detection probabilities under different false alarm probabilities

表6 不同数量训练样本主要参数

Table 6 Main parameters of different numbers of training sets

数据集	样本总数/10⁴	噪声参数	信噪比/dB	样本尺寸
训练集1	1.92	v=∞ σ²=1 v=∞ σ²=1	-15~0	2×4096
训练集2	9.6	v=∞ σ²=1 v=∞ σ²=1	-15~0	2×4096

图13 不同信噪比下平均检测概率对比(训练集1和训练集2)

Fig.13 Comparison of average detection probabilities under different signal-to-noise ratios(training sets 1 and 2)

图14 不同虚警概率下平均检测概率对比(训练集1和训练集2)

Fig.14 Comparison of average detection probabilities under different false alarm probabilities(training sets 1 and 2)

表7 方法复杂度对比

Table 7 Comparison of method complexity

名称	参数量	计算量	训练时长/ms	推理时长/ms
本文方法	1.23×10⁶	2.73×10⁸	4491.11	853.33
SSCL	1.07×10⁶	2.09×10⁸	4470.14	847.82
SSDNN	1.11×10⁵	2.98×10⁷	4228.19	823.19

参考文献 28

[1]	张航领, 周顺勇, 胡琴, 等. 认知无线电频谱感知技术研究综述[J]. 无线电工程, 2024, 54(11):2527-2536.
	ZHANG H L, ZHOU S Y, HU Q, et al. A review of research on cognitive radio spectrum sensing technology[J]. Radio Engineering, 2024, 54(11):2527-2536. (in Chinese)
[2]	HAYKIN S, SETOODEH P. Cognitive radio network:the spectrum supply chain paradigm[J]. IEEE Transactions on Cognitive Communications & Networking, 2017, 1(1):3-28.
[3]	DE LIMA C, BELOT D, BERKVENS R, et al. Convergent communication,sensing and localization in 6G systems:an overview of technologies,opportunities and challenges[J]. IEEE Access, 2021,9:26902-26925.
[4]	SINHA K, TRIVEDI Y N. Spectrum sensing for dynamic primary user with multiple transitions under Rayleigh distributed channel attenuation in additive Laplacian noise[J]. Physical Communication, 2024,62:102259.
[5]	OH H, NAM H. Energy detection scheme in the presence of burst signals[J]. IEEE Signal Processing Letters, 2019, 26(4):582-586. doi: 10.1109/LSP.2019.2900165
[6]	CHAUDHARY N, MAHAJAN R. Identification of spectrum holes using energy detector based spectrum sensing[J]. International Journal of Information Technology, 2021, 13(3):1243-1254.
[7]	BRITO A, SEBASTIÃO P, VELEZ F J. Hybrid matched filter detection spectrum sensing[J]. IEEE Access, 2021,9:165504-165516.
[8]	KABEEL A A, HUSSEIN A H, KHALAF A A M, et al. A utilization of multiple antenna elements for matched filter based spectrum sensing performance enhancement in cognitive radio system[J]. AEU-International Journal of Electronics and Communications, 2019,107:98-109.
[9]	KUMAR A, NANDHAKUMAR P. OFDM system with cyclostationary feature detection spectrum sensing[J]. ICT Express, 2019, 5(1):21-25.
[10]	MEHRABIAN A, ZAIMBASHI A. Robust and blind eigenvalue-based multiantenna spectrum sensing under IQ imbalance[J]. IEEE Transactions on Wireless Communications, 2018, 8(17):5581-5591.
[11]	LIU C, WANG J, LIU X M, et al. Maximum eigenvalue-based goodness-of-fit detection for spectrum sensing in cognitive radio[J]. IEEE Transactions on Vehicular Technology, 2019, 68(8):7747-7760.
[12]	O’SHEA T J, ROY T, CLANCY T C. Over the air deep learning based radio signal classification[J]. IEEE Journal of Selected Topics in Signal Processing, 2018, 12(1):168-179.
[13]	CHEN S Y, ZHANG Y, HE Z W, et al. A novel attention cooperative framework for automatic modulation recognition[J]. IEEE Access, 2020,8:15673-15686.
[14]	何遵文, 侯帅, 张万成, 等. 通信特定辐射源识别的多特征融合分类方法[J]. 通信学报, 2021, 42(2):103-112. doi: 10.11959/j.issn.1000-436x.2021028
	HE Z W, HOU S, ZHANG W C, et al. Multi-feature fusion classification method for communication specific emitter identification[J]. Journal on Communications, 2021, 42(2):103-112. (in Chinese) doi: 10.11959/j.issn.1000-436x.2021028
[15]	普运伟, 刘涛涛, 郭江, 等. 基于卷积神经网络和模糊函数主脊坐标变换的雷达辐射源信号识别[J]. 兵工学报, 2021, 42(8):1680-1689. doi: 10.3969/j.issn.1000-1093.2021.08.012
	PU Y W, LIU T T, GUO J, et al. Radar emitter signal recognition based on convolutional neural network and coordinate transformation of ambiguity function main ridge[J]. Acta Armamentarii, 2021, 42(8):1680-1689. (in Chinese)
[16]	LEE W, KIM M, CHO D H. Deep cooperative sensing:cooperative spectrum sensing based on convolutional neural networks[J]. IEEE Transactions on Vehicular Technology, 2019, 68(3):3005-3009.
[17]	LIU C, WANG J, LIU X M, et al. Deep CM-CNN for spectrum sensing in cognitive radio[J]. IEEE Journal on Selected Areas in Communications, 2019, 10(37):2306-2321.
[18]	LIU C, WEI Z, NG D W K, et al. Deep transfer learning for signal detection in ambient backscatter communications[J]. IEEE Transactions on Wireless Communications, 2020, 20(3):1624-1638.
[19]	TEKBIYIK K, AKBUNAR Ö, EKTI A R, et al. Spectrum sensing and signal identification with deep learning based on spectral correlation function[J]. IEEE Transactions on Vehicular Technology, 2021, 70(10):10514-10527.
[20]	ZHU X M, WANG T J, BAO Y P, et al. Signal detection in generalized Gaussian distribution noise with Nakagami fading channel[J]. IEEE Access, 2019,7:23120-23126.
[21]	YILMAZ F. McLeish distribution:performance of digital communications over additive white McLeish noise (AWMN) channels[J]. IEEE Access, 2020,8:19133-19195.
[22]	Gao Y, Ma S, Shi J, et al. An optimised multi-channel neural network model based on CLDNN for automatic modulation recognition[J]. International Journal of Wireless and Mobile Computing, 2023, 24(2):144-158.
[23]	ZHANG Z B, HUANG L H, LIU W. Hierarchical residual convolutional neural network for modulation classification[J]. IEEE Access, 2020,8:195010-195021.
[24]	王洋, 冯永新, 宋碧雪, 等. DP-DRCnet卷积神经网络信号调制识别算法[J]. 兵工学报, 2023, 44(2):545-555. doi: 10.12382/bgxb.2021.0620
	WANG Y, FENG Y X, SONG B X, et al. A modulation recognition algorithm of DP-DRCnet convolutional neural network[J]. Acta Armamentarii, 2023, 44(2):545-555. (in Chinese) doi: 10.12382/bgxb.2021.0620
[25]	LONG J, SHELHAMER E, DARRELL T. Fully convolutional networks for semantic segmentation[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2017, 39(4):640-651.
[26]	WOO S, PARK J, LEE J Y, et al. CBAM:convolutional block attention module[C]//Proceeding of Computer Vision-ECCV 2018.Munich, Germany:Springer, 2018.
[27]	LI X Y, ZHAO Z J, ZHANG Y P, et al. Spectrum sensing algorithm based on self-supervised contrast learning[J]. Electronics, 2023, 12(6):1317.
[28]	ZHANG Y P, ZHAO Z J. Limited data spectrum sensing based on semi-supervised deep neural network[J]. IEEE Access, 2021,9:166423-166435.