-
摘要:
无人机数据链通信受到各种自然与人为的干扰，信噪比（SNR）是信道状态和通信质量的有效评估指标。为解决传统估计算法信噪比估计精度不足的问题，提出了一种卷积神经网络（CNN）与长短时记忆（LSTM）网络结合的估计模型。利用仿真与实测相结合的方式，构建了一个包含不同信噪比、调制方式、衰落信道等信息的无人机通信信号数据集；在网络训练阶段，将样本序列进行分割，对分割后的每一部分序列使用CNN-LSTM网络提取深度特征，多次训练并保存模型参数；在测试阶段，利用构建好的测试集完成对算法的验证与测试，得到信噪比估计值。实验表明，相比于传统信噪比估计算法与单一网络结构的深度学习算法，所提算法的均方误差最低，实现了对信噪比的高精度估计。
-
关键词:
- 无人机 /
- 信噪比(SNR)估计 /
- 深度特征 /
- 卷积神经网络(CNN) /
- 长短时记忆(LSTM)网络
Abstract:UAV data link communication is subject to natural and artificial interferences. The signal-to-noise ratio (SNR) is an effective evaluation indicator of channel state and communication quality. In order to address insufficient SNR estimation accuracy involved in traditional estimation algorithm, an estimation model which combines convolutional neural networks (CNN) and long short term memory (LSTM) network is proposed. By means of both simulation and actual measurement, a data set of UAV communication signals is constructed with multiple SNRs, modulation modes, fading channels and other information included. In the network training phase, the sample sequence is segmented, CNN-LSTM is used to extract the deep feature of each part, and the model parameters are saved through multiple trainings. In the test phase, the constructed test set is used to verify and test the algorithm, and the SNR estimation value is obtained. Experiments show that compared with traditional SNR estimation algorithm and single-network deep learning method, the proposed algorithm can help achieve the lowest mean square error for different levels of SNR, thus achieving the high-precision estimation of SNR.
-
图 1 基于CNN-LSTM信噪比估计网络框架
Figure 1. Network framework of SNR estimation based on CNN-LSTM
图 2 基于分割窗的信号数据分割算法示意图
Figure 2. Schematic diagram of signal data segmentation algorithm based on split window
图 4 不同估计算法精度对比
Figure 4. Comparison of accuracy between CNN-LSTM network and traditional algorithms
图 5 不同网络结构的训练损失变化对比
Figure 5. Comparison of training loss change with different network structures
图 6 不同网络结构的测试精度对比
Figure 6. Comparison of test accuracy with different network structures
图 7 不同码元速率的测试精度对比
Figure 7. Comparison of test accuracy with different symbol transmission rates
表 1 通信信号数据集属性
Table 1. Attributes of communication signal data set
序号 字段名 定义 1 编号 数据样本的顺序 2 SNR 信噪比 3 调制类型 信息和载波结合的方式 4 载波频率 通信信号的工作频段 5 样本数 样本总共的条数 6 信号长度 信号的序列长度 7 码元速率 信道中码元传输的速率 
下载: 导出CSV
表 2 不同分割窗长度实验结果对比
Table 2. Comparison of experimental results of different split window lengths
分割窗长度 损失MSE SNR=-10dB SNR=-5dB SNR=0dB SNR=5dB SNR=10dB SNR=15dB SNR=20dB 无分割 4.417 1.542 0.911 0.796 0.726 0.423 0.315 1000 3.547 0.896 0.624 0.441 0.272 0.137 0.095 500 2.716 0.562 0.211 0.146 0.092 0.063 0.021 400 2.436 0.446 0.165 0.063 0.065 0.043 0.023 200 8.811 6.428 3.347 3.574 2.916 4.467 5.324 100 13.424 11.756 8.815 7.741 7.629 9.945 7.453 注：最小MSE值由下划线标记。
下载: 导出CSV
-
[1] 张文秋, 丁文锐, 刘春辉.一种无人机数据链信道选择和功率控制方法[J].北京航空航天大学学报, 2017, 43(3):583-591. https://bhxb.buaa.edu.cn/CN/abstract/abstract14020.shtmlZHANG W Q, DING W R, LIU C H.A channel selection and power control method for UAV data link[J].Journal of Beijing University of Aeronautics and Astronautics, 2017, 43(3):583-591(in Chinese). https://bhxb.buaa.edu.cn/CN/abstract/abstract14020.shtml [2] PAULUZZI D R, BEAUCIEU N C.A comparison of SNR estimation techniques for the AWGN channel[J].IEEE Transactions on Communications, 2000, 48(10):1681-1691. doi: 10.1109/26.871393 [3] BOUJELBEN M A, BELLILI F.EM algorithm for non-data-aided SNR estimation of linearly-modulated signals over SIMO channels[C]//Lobal Telecommunications Conference.Piscataway, NJ: IEEE Press, 2009: 4464-4469. [4] XIAO H F, SHI Y Q, SU W, et al.An investigation of non-data-aided SNR estimation techniques for analog modulation signals[C]//IEEE Sarnoff Symposium.Piscataway, NJ: IEEE Press, 2010: 351-355. [5] ÁLVAREZ-DÍAZ M, LÓPEZ-VALCARCE R, MOSQUERA C.SNR estimation for multilevel constellations using higher-order moments[J].IEEE Transactions on Signal Processing, 2010, 58(3):1515-1526. doi: 10.1109/TSP.2009.2036069 [6] STEPHENNE A, BELLILI F, AFFES S.Moment-based SNR estimation over linearly-modulated wireless SIMO channels[J].IEEE Transactions on Wireless Communications, 2010, 9(2):714-722. doi: 10.1109/TWC.2010.02.081719 [7] 华惊宇, 黄清, 滑翰, 等.一种移动环境下的信噪比估计算法及其在多普勒频移估计中的应用[J].通信学报, 2005, 26(5):135-140. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=txxb200505022HUA J Y, HUANG Q, HUA H, et al.A SNR estimation algorithm in mobile environment and its application in Doppler shift estimation[J].Transactions of Communications, 2005, 26(5):135-140(in Chinese). http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=txxb200505022 [8] 彭耿, 黄知涛, 陆凤波, 等.中频通信信号信噪比的快速盲估计[J].电子与信息学报, 2010, 32(1):102-106. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=dzkxxk201001019PENG G, HUANG Z T, LU F B, et al.Fast blind estimation of signal-to-noise ratio of IF communication signals[J].Journal of Electronics & Information Technology, 2010, 32(1):102-106(in Chinese). http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=dzkxxk201001019 [9] O'SHEA T J, HOYDIS J.An introduction to deep learning for the physical layer[J].IEEE Transactions on Cognitive Communications & Networking, 2017, 3(4):563-575. https://ieeexplore.ieee.org/document/8054694 [10] ZHANG D N, DING W R, ZHANG B C, et al.Automatic modulation classification based on deep learning for unmanned aerial vehicles[J].Sensors, 2018, 18(3):924-939. doi: 10.3390/s18030924 [11] 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, 2017, 12(1):168-179. https://ieeexplore.ieee.org/document/8267032 [12] O'SHEA T J, CORGAN J, CLANCY T C.Convolutional radio modulation recognition networks[C]//International Conference on Engineering Applications of Neural Networks.Berlin: Springer, 2016: 213-226. [13] LÉCUN Y, BOTTOU L, BENGIO Y, et al.Gradient-based learning applied to document recognition[J].Proceedings of the IEEE, 1998, 86(11):2278-2324. doi: 10.1109/5.726791 [14] SUN Y, WANG X, TANG X.Hybrid deep learning for face verification[C]//IEEE International Conference on Computer Vision.Piscataway, NJ: IEEE Press, 2013: 1489-1496. [15] SAK H, SENIOR A, BEAUFAYS F.Long short-term memory based recurrent neural network architectures for large vocabulary speech recognition[C]//15th Annual Conference of the International Speech Communication Association, 2014: 338-342. [16] GHOSH S, VINYALS O, STROPE B, et al.Contextual LSTM (CLSTM) models for large scale NLP tasks[EB/OL].(2016-05-31)[2018-10-15].https://www.researchgate.net/publication/301857393_Contextual_LSTM_CLSTM_models_for_Large_scale_NLP_tasks. [17] WANG J, CAO Z W.Chinese text sentiment analysis using LSTM Network based on L2 and Nadam[C]//IEEE International Conference on Communication Technology.Piscataway, NJ: IEEE Press, 2017: 1891-1895. [18] JOZEFOWICZ R, ZAREMBA W, SUTSKEVER I.An empirical exploration of recurrent network architectures[C]//Proceeding of the 32nd International Conference on Machine Learning, 2015, 37: 2342-2350. [19] GREFF K, SRIVASTAVA R K, KOUTNÍK J, et al.LSTM:A search space odyssey[J].IEEE Transactions on Neural Networks & Learning Systems, 2015, 28(10):2222-2232. -
下载:
点击查看大图
计量
- 文章访问数: 777
- HTML全文浏览量: 44
- PDF下载量: 710
- 被引次数: 0
