基于粒子群優(yōu)化多核支持向量數(shù)據(jù)描述的廣播式自動相關(guān)監(jiān)視異常數(shù)據(jù)檢測模型
doi: 10.11999/JEIT190767 cstr: 32379.14.JEIT190767
-
空軍工程大學(xué) 信息與導(dǎo)航學(xué)院 西安 710077
基金項目: 國家自然科學(xué)基金(61902426)
ADS-B Anomalous Data Detection Model Based on PSO-MKSVDD
-
School of Information and Navigation, Air Force Engineering University, Xi’an 710077, China
Funds: The National Natual Science Foundation of China (61902426)
-
摘要: 廣播式自動相關(guān)監(jiān)視(ADS-B)作為新一代空中交通管理(ATM)通信協(xié)議，是未來空管監(jiān)視系統(tǒng)的關(guān)鍵技術(shù)。目前，由于ADS-B采用明文格式廣播發(fā)送數(shù)據(jù)，其安全性問題受到挑戰(zhàn)。針對ADS-B易受到的欺騙干擾，該文將ADS-B位置數(shù)據(jù)和同步的二次雷達(dá)(SSR)數(shù)據(jù)作差，將兩者的差值作為樣本數(shù)據(jù)。利用多核支持向量數(shù)據(jù)描述(MKSVDD)訓(xùn)練樣本，得到了超球體分類器，此超球體分類器能檢測出ADS-B測試樣本中的異常數(shù)據(jù)。并且，通過粒子群算法(PSO)優(yōu)化了GaussLapl和GaussTanh兩種MKSVDD的懲罰因子、多核核函數(shù)系數(shù)以及核參數(shù)，提高了異常數(shù)據(jù)檢測性能。實驗結(jié)果表明，對于隨機位置偏移、固定位置偏移、拒絕服務(wù)(DOS)攻擊和重放攻擊，粒子群優(yōu)化多核支持向量數(shù)據(jù)描述(PSO-MKSVDD)模型能檢測出這4種攻擊類型的異常數(shù)據(jù)。且相較于其他機器學(xué)習(xí)和深度學(xué)習(xí)方法，該模型的適應(yīng)性更好，異常檢測的召回率和檢測率更優(yōu)。證明該模型可用于ADS-B異常數(shù)據(jù)的檢測。
-
關(guān)鍵詞:
- 廣播式自動相關(guān)監(jiān)視 /
- 空中交通管理 /
- 異常檢測 /
- 多核支持向量數(shù)據(jù)描述 /
- 粒子群優(yōu)化
Abstract: As a new generation of Air Traffic Management(ATM) communication protocol, Automatic Dependent Surveillance-Broadcast(ADS-B) is the key technology of ATM monitoring system in the future. At present, the security of ADS-B is challenged because it broadcasts data in plaintext format. Because ADS-B is susceptible to spoofing, the difference between ADS-B position data and synchronous Secondary Surveillance Radar(SSR) data is taken as sample data. Using Multi-Kernel Support Vector Data Description(MKSVDD) to train samples, a hypersphere classifier is obtained, which can detect anomalous data in ADS-B test samples. In addition, Particle Swarm Optimization (PSO) is used to optimize GaussLapl and GaussTanh MKSVDD penalty factors, coefficients of multi-kernel functions and kernel parameters.The performance of anomaly detection is improved. Experimental results show that PSO-MKSVDD can detect anomalous data of random position deviation, fixed position deviation, Denial Of Service(DOS) attack and replay attack. In addition, compared with other machine learning and deep learning methods, this model has better adaptability and better recall rate and detection rate of anomaly detection.It is proved that this model can be used to detect ADS-B anomalous data. -
表 1 樣本分類結(jié)果表
實際情況 預(yù)測結(jié)果 正例 負(fù)例 正例 ${\rm{TP}}$(真正例) ${\rm{FN}}$(假負(fù)例) 負(fù)例 ${\rm{FP}}$(假正例) ${\rm{TN}}$(真負(fù)例) 
下載: 導(dǎo)出CSV
表 2 異常檢測對比表(%)
SVDD GaussLapl GaussTanh 隨機位置偏移 召回率 94.0 95.2 94.8 檢測率 89.2 93.6 92.0 固定位置偏移 召回率 94.8 95.6 96.0 檢測率 94.4 96.4 97.2 DOS攻擊 召回率 94.8 96.0 95.2 檢測率 100.0 100.0 100.0 重放攻擊 召回率 94.8 96.0 95.6 檢測率 98.4 99.2 98.9
下載: 導(dǎo)出CSV
表 3 各種異常檢測方法結(jié)果對比(%)
LSTM SVDD LSTM-encoder-decoder seq2seq GaussLapl GaussTanh 隨機位置偏移 召回率 85.6 94.0 90.3 91.7 95.2 94.8 檢測率 87.0 89.2 89.8 90.6 93.6 92.0 固定位置偏移 召回率 84.2 94.8 93.8 91.0 95.6 96.0 檢測率 72.1 94.4 79.4 82.4 96.4 97.2 DOS攻擊 召回率 87.5 94.8 93.7 94.4 96.0 95.2 檢測率 92.6 100.0 95.2 95.6 100.0 100.0 重放攻擊 召回率 85.7 94.8 92.0 91.6 96.0 95.6 檢測率 88.2 98.4 93.6 94.4 99.2 98.9
下載: 導(dǎo)出CSV
-
STROHMEIER M, SCHAFER M, LENDERS V, et al. Realities and challenges of nextgen air traffic management: The case of ADS-B[J]. IEEE Communications Magazine, 2014, 52(5): 111–118. doi: 10.1109/MCOM.2014.6815901 ZHANG Jun, LIU Wei, and ZHU Yanbo. Study of ADS-B data evaluation[J]. Chinese Journal of Aeronautics, 2011, 24(4): 461–466. doi: 10.1016/s1000-9361(11)60053-8 MCCALLIE D, BUTTS J, and MILLS R. Security analysis of the ADS-B implementation in the next generation air transportation system[J]. International Journal of Critical Infrastructure Protection, 2011, 4(2): 78–87. doi: 10.1016/j.ijcip.2011.06.001 STROHMEIER M, LENDERS V, and MARTINOVIC I. On the security of the automatic dependent surveillance-broadcast protocol[J]. IEEE Communications Surveys & Tutorials, 2015, 17(2): 1066–1087. doi: 10.1109/comst.2014.2365951 WESSON K D, HUMPHREYS T E, and EVANS B L. Can cryptography secure next generation air traffic surveillance?[J/OL]. IEEE Security & Privacy. http://radionavlab.ae.utexas.edu/images/stories/files/papers/adsb_for_submission.pdf, 2014. BAEK J, BYON Y J, HABLEEL E, et al. Making air traffic surveillance more reliable: A new authentication framework for automatic dependent surveillance-broadcast (ADS-B) based on online/offline identity-based signature[J]. Security and Communication Networks, 2015, 8(5): 740–750. doi: 10.1002/sec.1021 MONTEIRO M. Detecting malicious ADS-B broadcasts using wide area multilateration[C]. The IEEE/AIAA 34th Digital Avionics Systems Conference, Prague, Czekh, 2015. doi: 10.1109/DASC.2015.7311579. NIJSURE Y A, KADDOUM G, GAGNON G, et al. Adaptive air-to-ground secure communication system based on ADS-B and wide-area multilateration[J]. IEEE Transactions on Vehicular Technology, 2016, 65(5): 3150–3165. doi: 10.1109/TVT.2015.2438171 SUN J, ELLERBROEK J, and HOEKSTRA J M. Modeling aircraft performance parameters with open ADS-B data[C]. The 12th USA/Europe Air Traffic Management Research and Development Seminar, Seattle, USA, 2017. HABLER E and SHABTAI A. Using LSTM encoder-decoder algorithm for detecting anomalous ADS-B Messages[J]. Computers & Security, 2018, 78: 155–173. doi: 10.1016/j.cose.2018.07.004 丁建立, 鄒云開, 王靜, 等. 基于深度學(xué)習(xí)的ADS-B異常數(shù)據(jù)檢測模型[J]. 航空學(xué)報, 2019, 40(12): 323220. doi: 10.7527/S1000-6893.2019.23220DING Jianli, ZOU Yunkai, WANG Jing, et al. ADS-B anomaly data detection model based on deep learning[J]. Acta Aeronautica et Astronautica Sinica, 2019, 40(12): 323220. doi: 10.7527/S1000-6893.2019.23220 王振昊, 王布宏. 基于SVDD的ADS-B異常數(shù)據(jù)檢測[J]. 河北大學(xué)學(xué)報: 自然科學(xué)版, 2019, 39(3): 323–329. doi: 10.3969/j.issn.1000-1565.2019.03.015WANG Zhenhao and WANG Buhong. ADS-B anomaly data detection based on SVDD[J]. Journal of HeBei University:Natural Science Edition, 2019, 39(3): 323–329. doi: 10.3969/j.issn.1000-1565.2019.03.015 TAX D M J and DUIN R P W. Support vector data description[J]. Machine Language, 2004, 54(1): 45–66. doi: 10.1023/b:Mach.0000008084.60811.49 G?NEN M, and ALPAYDIN E. Multiple kernel learning algorithms[J]. Journal of Machine Learning Research, 2011, 12: 2211–2268. ?Z??üR-AKYüZ S and WEBER G W. On numerical optimization theory of infinite kernel learning[J]. Journal of Global Optimization, 2010, 48(2): 215–239. doi: 10.1007/s10898-009-9488-x 殷禮勝, 唐圣期, 李勝, 等. 基于整合移動平均自回歸和遺傳粒子群優(yōu)化小波神經(jīng)網(wǎng)絡(luò)組合模型的交通流預(yù)測[J]. 電子與信息學(xué)報, 2019, 41(9): 2273–2279. doi: 10.11999/JEIT181073YIN Lisheng, TANG Shengqi, LI Sheng, et al. Traffic flow prediction based on hybrid model of auto-regressive integrated moving average and genetic particle swarm optimization wavelet neural network[J]. Journal of Electronics &Information Technology, 2019, 41(9): 2273–2279. doi: 10.11999/JEIT181073 錢亞冠, 盧紅波, 紀(jì)守領(lǐng), 等. 基于粒子群優(yōu)化的對抗樣本生成算法[J]. 電子與信息學(xué)報, 2019, 41(7): 1658–1665. doi: 10.11999/JEIT180777QIAN Yaguan, LU Hongbo, JI shouling, et al. Adversarial example generation based on particle swarm optimization[J]. Journal of Electronics &Information Technology, 2019, 41(7): 1658–1665. doi: 10.11999/JEIT180777 https://opensky-network.org/. -
圖(13) / 表(3)
計量
- 文章訪問數(shù): 3520
- HTML全文瀏覽量: 1016
- PDF下載量: 90
- 被引次數(shù): 0
下載:
