于海濤+李治軍+姜守旭
摘要: 關(guān)鍵詞: 中圖分類號: 文獻標志碼: A文章編號: 2095-2163(2017)06-0148-04
Abstract: The receiving signal strength indicator (RSS) as a mainstream solution is often used for locating system and fingerprint positioning system based on ranging. However, RSS is often affected by multiple size effects and noise signals, and its location performance is not stable. In recent years, many commercial WiFi devices have supported access to the physical layer's channel status information (CSI). CSI is a more finegrained indicator of signal characteristics than RSS. Compared to RSS, CSI analyses the characteristics of multiple subcarrier signals to avoid the effects of multipath effect and noise. The CSI has opened up new spaces for WiFi based indoor location technology, and has been concerned by researchers. For this purpose, this paper carries out the research on the indoor location method based on RSS and CSI hybrid fingerprint.
0引言
隨著WiFi網(wǎng)絡(luò)的密集部署以及智能移動設(shè)備的普及,基于WiFi通訊的無線網(wǎng)絡(luò)變得越來越重要。在無線網(wǎng)絡(luò)環(huán)境下,人類活動會影響通訊信號及信號特征,所以通訊信號除了用于滿足正常的通信需求外,還可以通過分析信號來挖掘出人類活動信息的內(nèi)容,從而更好地利用無線網(wǎng)絡(luò),室內(nèi)定位就是其典型應(yīng)用之一。目前,利用WiFi信號進行室內(nèi)定位的方法主要可以分為三類:指紋法(fingerprinting-based)、測距法(ranging-based)、到達角度法(angle of arrival (AOA)-based)。其中,測距法通過計算待定位目標與至少三個不同AP之間的距離并利用幾何模型進行定位,而測距法又可以分為兩類:基于信號強度、基于時間(TOF)。進一步研究可知,基于信號強度方法利用多個接受信號訓(xùn)練信號強度衰落模型中的參數(shù),從而得到距離;基于時間方法與之類似,也是通過計算信號傳播時間求出距離。但是,上述兩種方法需要AP與定位目標之間存在LOS通訊路徑。本文的室內(nèi)定位研究選用了基于RSS與CSI的混合指紋,使用混合指紋進行定位相比其他基于單一指紋信息(RSS或CSI)的定位方法有很多好處。由于多徑效應(yīng)的影響,RSS信息不穩(wěn)定,即使在固定位置采集得到的RSS信息也會隨時間不斷劇烈變化,并且RSS并沒有包含OFDM下多子載波的相應(yīng)多徑信息。OFDM系統(tǒng)中,相比RSS信息,CSI利用了不同子載波的信號傳輸過程信息,從而可以降低多徑效應(yīng)的影響。通過細粒度的CSI指紋法,可以在不增加數(shù)據(jù)采集成本的前提下,改善室內(nèi)定位精度。因此本次研究利用CSI和RSS混合指紋來進行室內(nèi)定位的設(shè)計實現(xiàn)。
1RSS初步定位
1.1spike剔除
如圖1所示,不同顏色的折線代表不同AP對應(yīng)beacon包的RSS值,橫軸為時間,縱軸為信號強度。從圖1中可以看出,原始RSS數(shù)據(jù)基本保持穩(wěn)定,但是存在某些不規(guī)律的信號突變,而這些信號突變往往導(dǎo)致RSS大幅度降低,研究將這類大幅變化稱為spike。這些spike并不能真實反映信號強度在空間上的分布。無論在離線指紋數(shù)據(jù)庫建立階段,還是在線采集樣本指紋時,都需要去除spike的影響。所以就需要識別spike并剔除其影響。為此提出了一個簡單的基于滑動時間窗統(tǒng)計的spike檢測與恢復(fù)方法。時間窗長度為1 s,統(tǒng)計時間窗內(nèi)最小RSS與其他RSS均值的差值。若差值的絕對值大于一定的閾值,就可判定該最小RSS對應(yīng)的beacon受到spike影響,則去除該beacon的RSS值,并恢復(fù)為當前時間窗內(nèi)其它beacon的RSS均值。實驗效果如圖2所示,恢復(fù)后的RSS數(shù)據(jù)在保留了原有大部分數(shù)據(jù)的同時,去除了spike的影響。
1.2缺失beacon對應(yīng)RSS恢復(fù)
由于802.11n中載波偵聽機制(CSMA/CA)的存在,在信道高負載無線網(wǎng)絡(luò)環(huán)境下,由于在一大段時間內(nèi)的信道繁忙而導(dǎo)致AP的beacon缺失。實際生活中,大量WiFi設(shè)備無法及時偵測到AP也是由以上原因所導(dǎo)致。如圖3所示,不同顏色的折線代表不同AP對應(yīng)的beacon的RSS信號隨時間的變化,圖3表明:三個AP對應(yīng)的beacon在622 s之后的近2 s內(nèi)缺失,2 s的beacon缺失將會對實時要求較高的室內(nèi)定位產(chǎn)生較大的影響。為了避免beacon缺失引發(fā)的后果,從而盡量減少未偵測AP信息帶來的損失,需要對其相應(yīng)AP的RSS信息進行恢復(fù)。
圖4是對某一網(wǎng)格內(nèi)的AP信號進行主成分分析的結(jié)果,可以看出該網(wǎng)格內(nèi)的不同AP信號強度具有鮮明的線性相關(guān)性、數(shù)據(jù)低秩性。所以,研究可以利用基于矩陣分解的低秩數(shù)據(jù)回復(fù)算法對丟失beacon的AP的RSS信號提供恢復(fù)處理。為此,則選取了基于奇異值分解的算法。為了盡量減小計算時間,過程中首先利用未丟失的AP的RSS組成的向量與指紋數(shù)據(jù)庫中相應(yīng)AP的RSS向量進行比較,選取余弦距離較小的top-k個指紋參與矩陣分解。最終可得本文設(shè)計給出的方法恢復(fù)得到的RSS相對誤差為20.7%。endprint
1.3離線階段
1.4在線階段
2CSI精確定位
2.1深度神經(jīng)網(wǎng)絡(luò)結(jié)構(gòu)
利用CSI進行精確定位的時候用到了深度神經(jīng)網(wǎng)絡(luò)系統(tǒng),這里選用的是tensorflow系統(tǒng)神經(jīng)網(wǎng)絡(luò),考慮到神經(jīng)網(wǎng)絡(luò)的強大的學(xué)習(xí)能力,原有的3*3*30的270維度特征的建模在精確度上仍有所欠缺,因此重點擇取深度學(xué)習(xí)進行特征學(xué)習(xí),其中的數(shù)據(jù)輸入是270維的CSI數(shù)據(jù)特征,通過把標簽換成對應(yīng)的CSI輸入數(shù)據(jù),這樣就開始了深度學(xué)習(xí)訓(xùn)練??梢允褂帽碚鲾?shù)據(jù)內(nèi)部特征的深度網(wǎng)絡(luò)DFDN。對于每一個APi及單位區(qū)域 j, 均可以得到表征數(shù)據(jù)的內(nèi)部特征的深度神經(jīng)網(wǎng)DFDN(i, j)。圖5即完整展示了深度神經(jīng)網(wǎng)絡(luò)的訓(xùn)練過程。由圖5可知,該網(wǎng)絡(luò)共有6層,其中每一層的相關(guān)設(shè)置都在ubuntu的tensorflow深度學(xué)習(xí)框架下面獲得定制實現(xiàn)。
4結(jié)束語
本文提出了一種基于RSS與CSI混合指紋室內(nèi)定位研究方法。展開來說,本次研究首先給出了基于RSS初步定位的設(shè)計解析和功能實現(xiàn);同時,又重點探討了基于CSI精確定位的分析模式與方法流程。在此基礎(chǔ)上,進一步論述展示了基于RSS與CSI混合指紋室內(nèi)定位的研發(fā)仿真結(jié)果。關(guān)于本課題的深入研究還在不斷的發(fā)展進程中,本文的研究成果也可為后續(xù)的同類研究提供有益的借鑒與參考。
參考文獻:
[1] LIU Hui, DARABI H, BANERJEE P, et al. Survey of wireless indoor positioning techniques and systems[J]. IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews), 2007, 37( 6):1067-1080.
[2] DAYEKH S, AFFES S, KANDIL N, et al. Cooperative localization in mines using fingerprinting and neural networks[C]//2010 IEEE Wireless Communication and Networking Conference. Sydney, NSW, Australia:IEEE, 2010:1-6.
[3] WU Zhili, LI C H, NG J K ,et al. Location estimation via support vector regression[J]. IEEE Transactions on Mobile Computing, 2007, 6(3):311-321.
[4] WU Kaishun, XIAO Jiang, YI Youwen, et al. CSI-based indoor localization[J]. IEEE Transactions on Parallel and Distributed Systems, 2013,24(7):1300-1309.
[5] HALPERIN D, HU Wenjun, SHETH A, et al. Predictable 802.11 packet delivery from wireless channel measurements[C]//Proceedings of the ACM SIGCOMM 2010 conference. New Delhi, India:ACM, 2010:159-170.
[6] WANG Xuyu, GAO Lingjun, MAO Shiwen, et al. DeepFi: Deep learning for indoor fingerprinting using channel state information[C]//2015 IEEE Wireless Communications and Networking Conference (WCNC). New Orleans, LA, USA:IEEE, 2015:1666-1671.
[7] WANG Xuyu, GAO Li^ngjun, MAO Shiwen. PhaseFi: Phase fingerprinting for indoor localization with a deep learning approach[C]//2015 IEEE Global Communications Conference (GLOBECOM) . San Diego, CA, USA:IEEE, 2015:1-6.
[8] WANG Xuyu, GAO Lingjun, MAO Shiwen. CSI phase fingerprinting for indoor localization with a deep learning approach[J]. IEEE Internet of Things Journal,2016,3(6):1113-1123.
[9] XIONG J, JAMIESON K. Arraytrack: A finegrained indoor location system[J]. USENIX Symposium on Networked Systems Design and Implementation, 2013(279976): 71-84.
[10]HAN D, JUNG S, LEE M, et al. Building a practical WiFibased indoor navigation system[J]. IEEE Pervasive Computing, 2014,13(2): 72-79.endprint
[11]MIROWSKI P, WHITING P, STECK H. Probability kernel regression for WiFi localisation[J]. Journal of Location Based Services, 2012,6(2): 81-100.
[12]CHO H, SANG W K. Mobile robot localization using biased chirpspreadspectrum ranging[J]. IEEE transactions on industrial electronics,2010,57(8): 2826-2835.
[13]DHITAL A, CLOSAS P, FERNNDEZPRADES C. Bayesian filtering for indoor localization and tracking in wireless sensor networks[J]. Eurasip Journal on Wireless Communications and Networking, 2012,2012(1): 1-13.
[14]ZHANG Xinyu, SHIN K G. Cooperative carrier signaling: Harmonizing coexisting WPAN and WLAN devices[J]. IEEE/ACM Transactions on Networking, 2013, 21(2):426-439.
[15]CHEBROLU K, DHEKNE A. Esense: Communication through energy sensing[C]// Proceedings of the 15th annual international conference on Mobile computing and networking. Beijing, China: ACM, 2009:85-96.
[16]KIM S M, TIAN He. FreeBee: Crosstechnology communication via free sidechannel[C]//Proceedings of the 21st Annual International Conference on Mobile Computing and Networking. Paris, France:ACM, 2015:317-330.
[17]BHARADIA D, JOSHI K R, KOTARU M, et al. Backfi: High throughput WiFi backscatter[C]//Proceedings of the 2015 ACM Conference on Special Interest Group on Data Communication. London, United Kingdom:ACM,2015: 283-296.
[18]IYER V, TALLA V, KELLOGG B, et al. Intertechnology backscatter: Towards Internet connectivity for implanted devices[C]// Proceedings of the 2016 ACM SIGCOMM Conference. Florianopolis, Brazil:ACM, 2016:356-369.
[19]ISHIDA S, IZUMI K, TAGASHIRA S, et al. WiFi APRSS monitoring using sensor nodes toward anchorfree sensor localization[C]//2015 IEEE 82nd Vehicular Technology Conference (VTC2015-Fall). Boston, MA, USA:IEEE, 2015:1-5.
[20]NIU Jianwei, WANG Bowei, SHU Lei, et al. ZIL: An energyefficient indoor localization system using ZigBee radio to detect WiFi fingerprints[J]. IEEE Journal on Selected Areas in Communications,2015, 33 (7): 1431-1442.
[21]WANG Yufei, WANG Qixin, ZENG Zheng, et al. WiCop: Engineering WiFi temporal whitespaces for safe operations of wireless body area networks in medical applications[C]// 2011 IEEE 32nd Real-Time Systems Symposium. Vienna, Austria:IEEE, 2011:170-179.
[22]BAHL P, PADMANABHAN V N. Radar: An inbuilding RFbased user location and tracking system[C]// INFOCOM 2000. Nineteenth Annual Joint Conference of the IEEE Computer and Communications Societies. Tel Aviv, Israel:IEEE, 2000,2:775-784.endprint
[23]CHAI Xiaoyong, YANG Qiang. Reducing the calibration effort for probabilistic indoor location estimation[J]. IEEE Transactions on Mobile Computing, 2007,6(6): 649-662.
[24]LI B, WANG Y, LEE H K, et al. Method for yielding a database of location fingerprints in WLAN[J]. IEE ProceedingsCommunications,2005, 152(5): 580-586.
[25]KUO S P, TSENG Y C. Discriminant minimization search for largescale RFbased localization systems[J]. IEEE Transactions on mobile computing,2011, 10(2): 291-304.
[26]PERING T, AGARWAL Y, GUPTA R, et al. Coolspots: Reducing the power consumption of wireless mobile devices with multiple radio interfaces[C]//Proceedings of the 4th international conference on Mobile systems, applications and services. Uppsala, Sweden:ACM, 2006:220-232.
[27]LI Qiang, ZHANG Ying, LIN Jingran, et al. Fullduplex bidirectional secure communications under perfect and distributionally ambiguous eavesdropper's CSI[J]. arXiv preprint arXiv:1705.07337v2, 2017.
[28]LI R T H, CHUNG H S H, CHAN T K M. An active modulation technique for singlephase grid connected CSI[J]. IEEE Transactions on Power Electronics,2007,22(4):1373-1382.endprint