logo

SCIENTIA SINICA Informationis, Volume 49, Issue 7: 868-885(2019) https://doi.org/10.1360/N112018-00030

WLAN indoor target intrusion detection algorithm based on adaptive-depth ray tree

More info
  • ReceivedFeb 7, 2018
  • AcceptedMay 18, 2018
  • PublishedMay 10, 2019

Abstract

With the wide deployment of wireless local area network (WLAN) and general support of WLAN protocol by various intelligent terminals, the intrusion detection with respect to the indoor target can be realized by using the existing WLAN infrastructure. To achieve this goal, the adaptive-depth ray tree based quasi 3D ray-tracing model is constructed to model the received signal strength (RSS) propagation property under the indoor silence and intrusion scenarios. Then, the RSS mean, variance, maximum, minimum, range, and median are allied to construct the training database for the probabilistic neural network (PNN). Finally, after the training, the PNN is utilized to perform the multi-classification decision with respect to the newly-collected RSS data, and consequently achieve the indoor target intrusion detection and area localization. The experimental results indicate that the proposed algorithm is featured with high intrusion detection rate as well as low database construction cost.


Funded by

国家自然科学基金(61771083,61704015)

长江学者和创新团队发展计划(IRT1299)

重庆市科委重点实验室专项经费

重庆市基础科学与前沿技术研究(cstc2017jcyjAX0380,cstc2015jcyjBX0065)

重庆市高校优秀成果转化(KJZH17117)


References

[1] Huggins K R, Mcgrath M A, Zheng Y F, et al. Computer vision localization based on pseudo-satellites. IEEE Aerospace and Electronic Systems Magazine, 2010, 25: 4--10. Google Scholar

[2] Jo K, Chu K, Sunwoo M. Interacting Multiple Model Filter-Based Sensor Fusion of GPS With In-Vehicle Sensors for Real-Time Vehicle Positioning. IEEE Trans Intell Transp Syst, 2012, 13: 329-343 CrossRef Google Scholar

[3] Wei S, Zhang W, Deng C. BriGuard: a lightweight indoor intrusion detection system based on infrared light spot displacement. IET Sci Measurement Tech, 2015, 9: 306-314 CrossRef Google Scholar

[4] Li Y P, Yang J, Li X, et al. Ultrasonic intruder detection system for home security. Intelligent Control and Automation, 2006: 1108--1115 doi: 10.1007/978-3-540-37256-1_143. Google Scholar

[5] Bo K, Kvarstein B, Hagen R, et al. Pelvic floor muscle exercise for the treatment of female stress urinary incontinence: reliability of vaginal pressure measurements of pelvic floor muscle strength. Neurourology Urodynamics, 2010, 9: 471--477. Google Scholar

[6] Yuan W, Wu N, Wang H, et al. Factor graph approach for joint passive localization and receiver synchronization in wireless sensor networks. In: Proceedings of IEEE Conference on Annual International Symposium on Personal, Indoor, and Mobile Radio Communications, Valencia, 2016. 1--5. Google Scholar

[7] Youssef M, Mah M, Agrawala A. Challenges: device-free passive localization for wireless environments. In : Proceedings of ACM International Conference on Mobile Computing and Networking, Montreal, 2007. 222--229. Google Scholar

[8] Jin S, Choi S. A Seamless Handoff With Multiple Radios in IEEE 802.11 WLANs. IEEE Trans Veh Technol, 2014, 63: 1408-1418 CrossRef Google Scholar

[9] Wang Q, Yigitler H, Jantti R. Localizing Multiple Objects Using Radio Tomographic Imaging Technology. IEEE Trans Veh Technol, 2016, 65: 3641-3656 CrossRef Google Scholar

[10] Condell J, Deak D, Deak G. Detection of multi-occupancy using device-free passive localisation. IET Wireless Sens Syst, 2014, 4: 130-137 CrossRef Google Scholar

[11] Savazzi S, Nicoli M, Carminati F. A Bayesian Approach to Device-Free Localization: Modeling and Experimental Assessment. IEEE J Sel Top Signal Process, 2014, 8: 16-29 CrossRef ADS Google Scholar

[12] Liu Z Y, Guo L X, Tao W. Full automatic preprocessing of digital map for 2.5D ray tracing propagation model in urban microcellular environment. Waves Random Complex Media, 2013, 23: 267-278 CrossRef Google Scholar

[13] Specht D F. Probabilistic neural network and general regression neural network. Fuzzy Logic Neural Netw Handbook, 1996, 3: 301--344. Google Scholar

[14] Mahmoud S F, Wait J R. Geometrical optical approach for electromagnetic wave propagation in rectangular mine tunnels. Radio Sci, 2016, 9: 1147--1158. Google Scholar

[15] Ang T, Tan S, Tan H. Analytical methods to determine diffraction points on multiple edges and cylindrical scatterers in UTD ray-tracing. Microw Opt Tech Lett, 2015, 22: 304--309. Google Scholar

[16] Dersch U, Zollinger E. Propagation mechanisms in microcell and indoor environments. IEEE Trans Veh Technol, 1994, 43: 1058-1066 CrossRef Google Scholar

[17] Farace P, Righetto R, Deffet S. Technical Note: A direct ray-tracing method to compute integral depth dose in pencil beam proton radiography with a multilayer ionization chamber. Med Phys, 2016, 43: 6405-6412 CrossRef PubMed ADS Google Scholar

[18] Coco S, Laudani A, Mazzurco L. A Novel 2-D Ray Tracing Procedure for the Localization of EM Field Sources in Urban Environment. IEEE Trans Magn, 2004, 40: 1132-1135 CrossRef ADS Google Scholar

[19] Sabar N R, Ayob M, Kendall G. A dynamic multiarmed bandit-gene expression programming hyper-heuristic for combinatorial optimization problems.. IEEE Trans Cybern, 2015, 45: 217-228 CrossRef PubMed Google Scholar

[20] Xu Y, Erricolo D, and Uslenghi P L E. A novel approach to 3D propagation in urban environments. In: Proceedings of IEEE Antennas and Propagation Society International Symposium, San Antonio, 2002. 338--341. Google Scholar

[21] Erceg V, Rustako A J, Roman R S. Diffraction around corners and its effects on the microcell coverage area in urban and suburban environments at 900 MHz, 2 GHz, and 4 GHz. IEEE Trans Veh Technol, 1994, 43: 762-766 CrossRef Google Scholar

[22] Hata M, Nagatsu T. Mobile location using signal strength measurements in a cellular system. IEEE Trans Veh Technol, 1980, 29: 245-252 CrossRef Google Scholar

[23] El-Sallabi H M, Vainikainen P. Improvements to Diffraction Coefficient for Non-Perfectly Conducting Wedges. IEEE Trans Antennas Propagat, 2005, 53: 3105-3109 CrossRef ADS Google Scholar

[24] Dutt V, Chaudhry V, Khan I. Different approaches in pattern recognition. Computer Science and Engineering, 2011, 1: 32--35. Google Scholar

[25] Queiroz A, Trintinalia L C. An analysis of human body shadowing models for ray-tracing radio channel characterization. In: Proceedings of IEEE Conference on Microwave and Optoelectronics, Galinhas, 2016. 1--5. Google Scholar

[26] Chendong D, Zhengjia H, Hongkai J. A sliding window feature extraction method for rotating machinery based on the lifting scheme. J Sound Vib, 2007, 299: 774-785 CrossRef ADS Google Scholar

[27] Battiti R, Colla A M. Democracy in neural nets: Voting schemes for classification. Neural Networks, 1994, 7: 691-707 CrossRef Google Scholar

[28] Stehman S V. Selecting and interpreting measures of thematic classification accuracy. Remote Sens Environ, 1997, 62: 77-89 CrossRef ADS Google Scholar

[29] Youssef M, Mah M, Agrawala A. Challenges: Device-free passive localization for wireless environments. In: Proceedings of ACM International Conference on Mobile Computing and Networking, Quebec, 2007. 222--229. Google Scholar

  • Table 1   The calculation process of each layer of PNN neurons
    Functional layer Number of neurons Calculation process
    Input layer Signal feature dimension $d$ (1) Compute ${z_k}~=~{\boldsymbol~x}_k^{\rm{T}}{\boldsymbol~x}~(k~=~1,{\rm{~}}~\ldots~,~n)$, in which ${\boldsymbol~x}~=~({{{x_1}}~/~{\sqrt~{\sum\nolimits_{j~=~1}^d~{x_j^2}~}~}},~\ldots~,$ ${{{x_d}}~/~{\sqrt~{\sum\nolimits_{j~=~1}^d~{x_j^2}~}~}})$ and ${{\boldsymbol~x}_k}~=~({{{x_{k1}}}~/~{\sqrt~{\sum\nolimits_{j~=~1}^d~{x_{kj}^2}~}~}},{\rm{~}}~\ldots~,{\rm{~}}{{{x_d}}~/~{\sqrt~{\sum\nolimits_{j~=~1}^d~{x_{kj}^2}~}~}})$ are normalization vectors of test sample and $k$-th training sample, respectively, $n$ is the size of training samples. (2) Assign the connection weights $\omega_{jk}$ of the $j$-th neuron in the input layer and the $k$-th neuron in the model layer, and the assignment process is described in Algorithm 3.
    Model layer $n$ (1) Compute kernel density function $\varphi~({z_k})~=~\exp~(\frac{{{z_k}~-~1}}{{{\delta~^2}}})$, in which $\delta$ is the smoothing factor. (2) Assign the connection weights ${a_{ki}}~(i~=~1,{\rm{~}}~\ldots~,{\rm{~}}c)$ of the $k$-th neuron in the model layer and the $i$-th neuron in the summation layer, in which $c$ is the number of states. If ${\boldsymbol~x}_k$ belongs to $i$-th state, then $a_{ki}=1$, otherwise $a_{ki}=0$, as described in Algorithm 3.
    Summation layer $c$ Compute the conditional probability of ${\boldsymbol~x}$ belonging to the $i$-th state ${g_i}({\boldsymbol{x)}}~=~\frac{{{P_i}}}{{{N_i}}}\sum\nolimits_{k~\in~\{~1,{\rm{~}}~\ldots~,{\rm{~}}n\}~~\cup~{a_{ki}}~=~1}~{\varphi~({z_k})}$, in which $P_i$ is the priori probability of the $i$-th state and, $N_i$ is the number of training samples that belong to the $i$-th state.
    Output layer $1$ Compute$\max~\{~{g_i}({\boldsymbol~x}),{\rm{~}}i~\in~\{~1,{\rm{~}}~\ldots~,{\rm{~}}c\}~\}$, in which the $i$ that corresponds to the maximum value of $g_i({\boldsymbol~x})$ is the output state of PNN.
  • Table 2   Average time cost for ray modeling between each pair of AP and MP
    Performance index Ref. [18] Ref. [12] The proposed method
    Time overhead (s) 6.03 7.25 3.41
  • Table 3   Performance comparison of different intrusion detection methods
    Performance index MA MV Radio tomographic imaging Neural network The proposed method
    Missed detection probability (%) 14.46 14.26 7.08 5.23 3.42
    False detection probability (%) 13.85 10.40 12.47 9.01 7.96
    Successful detection probability (%) 95.8 94.5 97.40
    Average time overhead 2.35 2.46 9.77 6.57 6.62
    for single detection (s)

Copyright 2019 Science China Press Co., Ltd. 《中国科学》杂志社有限责任公司 版权所有

京ICP备18024590号-1