現在位置首頁 > 博碩士論文 > 詳目
論文中文名稱:基於深度學習之通道響應定位演算法 [以論文名稱查詢館藏系統]
論文英文名稱:A Deep Neural Network-Based Indoor Positioning Method using Channel State Information [以論文名稱查詢館藏系統]
院校名稱:臺北科技大學
學院名稱:電資學院
系所名稱:電子工程系研究所
畢業學年度:105
畢業學期:第二學期
出版年度:106
中文姓名:吳冠賢
英文姓名:Guan-Sian Wu
研究生學號:104368031
學位類別:碩士
語文別:中文
口試日期:2017/07/07
論文頁數:23
指導教授中文名:曾柏軒
指導教授英文名:Po-Hsuan Tseng
口試委員中文名:謝欣霖;許裕彬;黃琴雅;曾柏軒
中文關鍵詞:深度神經網路通道狀態資訊指紋辨識室內定位
英文關鍵詞:DeepNeuralNetwork(DNN)ChannelStateInformationFingerprintingIndoorPositioning
論文中文摘要:藉由量測的通道狀態信息(CSI),傳送端和接收端之間的多重路徑環境,可被當成室內位置的指紋(FP)通道特徵。由於其開源和低成本特性,基於CSI 的指紋識別的方法,受到越來越多的關注。近來,已經有提出一系列使用基於深度學習的方法,利用通道狀態信息(CSI)的室內指紋特性來增強室內定位的性能。在本文中,我們提出了一種使用 CSI 的深層神經網絡室內指紋系統,稱為 DNNFi。提出的 DNNFi,在不同的參考點上,不需要使用多個深層
Autoencoder 架構,只需使用單個 Deep NeuralNetwork 便能辨識環境。DNNFi可以允許我們更快的即時線上運算,並能減少儲存權重及偏差所需的參數。本文還考慮接收信號強度(RSS),採用貪心學習方法逐層預先訓練權重。Softmax函數用於確定參考點上位置的機率,並且基於參考點的加權平均值來估計位置。提供實驗結果中,顯示與傳統的 CSI /RSS 定位指紋識別方法相比,DNNFi 可以有效降低位置誤差。
論文英文摘要:The channel state information (CSI) measurement, which characterizes the multipath channel between the transmitter and the receiver, can serve as a received position signature for indoor position fingerprinting (FP). CSI based FP methods have received increasing attention due to its open access and low-cost properties. Recently, a series of novel deep learning-based indoor FP methods using CSI have been proposed to enhance the indoor localization performance. In this paper, we present a deep neural network-based indoor FP system using CSI, which is termed DNNFi. The proposed DNNFi, which maintains a single DNN instead of multiple deep autoencoders at different reference points, allows a faster computation for the online inference and a lower memory usage for the weights/biases. With the consideration of the received signal strength in the data pre-processing, a stack of autoencoders is utilized to pre-train the weights layer-by-layer. The softmax function is adopted to decide the probabilities of the receiver position being on these reference points, which can be used to estimate the receiver position. Experimental results are presented to confirm that DNNFi can effectively reduce location error compared with the conventional CSI/RSS positioning FP approaches.
論文目次:Chinese Abstract ...i
English Abstract ...ii
Acknowledgement ...iii
Contents ...iv
List of Figures ...v
List of Tables ...vi
1 Introduction ...1
2 SystemModel ...6
2.1 ChannelModel ...6
2.2 FingerprintingSystemArchitecture ...7
3 ProposedDeepNeuralNetwork-BasedFingerprinting(DNNFi) ...9
3.1 Offline/Online-Phase:Pre-processing ...9
3.2 Offline-Phase:DNNTraining ...11
3.3 OnlinePhase:DNNInferenceforLocationEstimation ...13
4 PerformanceEvaluation ...14
5 Conclusion ...24
Bibliography ...25
論文參考文獻:[1] X. Wang,L.Gao,S.Mao,andS.Pandey,“CSI-basedFingerprintingforIndoorLocalization:
A DeepLearningApproach,” IEEE Trans.Veh.Technol., vol.66,no.1,pp.763–776,Jan
2017.
[2] P.H.TsengandK.T.Lee,“AFemto-AidedLocationTrackingAlgorithminLTE-AHet-
erogeneous Networks,” IEEE Trans.Veh.Technol., vol.66,no.1,pp.748–762,Jan2017.
[3] C. Feng,W.S.A.Au,S.Valaee,andZ.Tan,“Received-signal-strength-basedIndoorPo-
sitioning usingCompressiveSensing,” IEEE TransactionsonMobileComputing, vol.11,
no. 12,pp.1983–1993,2012.
[4] Z. Yang,Z.Zhou,andY.Liu,“FromRSSItoCSI:IndoorLocalizationviaChannelRe-
sponse,” ACMComput.Surv., vol.46,no.2,pp.1–32,Dec.2013.
[5] M. YoussefandA.Agrawala,“TheHorusWLANlocationdeterminationsystem,”in Pro-
ceedingsofthe3rdinternationalconferenceonMobilesystems,applications,andservices.
ACM,2005,pp.205–218.
[6] D. Halperin,W.Hu,A.Sheth,andD.Wetherall,“ToolRelease:Gathering802.11nTraces
with ChannelStateInformation,” ACMSIGCOMMCCR, vol.41,no.1,p.53,Jan.2011.
[7] Y. Xie,Z.Li,andM.Li,“PrecisePowerDelayProfilingwithCommodityWiFi,”in Pro-
ceedingsofthe21stAnnualInternationalConferenceonMobileComputingandNetworking,
2015, pp.53–64.
[8] C. Nerguizian,C.Despins,andS.Affes,“Geolocationinmineswithanimpulseresponse
fingerprintingtechniqueandneuralnetworks,” IEEE Trans.WirelessCommun., vol.5,no.3,
pp. 603–611,Mar.2006.
[9] K. Wu,J.Xiao,Y.Yi,D.Chen,X.Luo,andL.Ni,“CSI-BasedIndoorLocalization,” IEEE
Trans.ParallelDistrib.Syst., vol.24,no.7,pp.1300–1309,July2013.
[10] X. Wang,L.Gao,andS.Mao,“CSIPhaseFingerprintingforIndoorLocalizationwitha
Deep LearningApproach,” IEEE InternetofThingsJournal, vol.PP,no.99,pp.1–1,2016.
[11] G. E.Hinton,S.Osindero,andY.-W.Teh,“AFastLearningAlgorithmforDeepBelief
Nets,” Neuralcomputation, vol.18,no.7,pp.1527–1554,2006.
[12] G. E.HintonandR.R.Salakhutdinov,“ReducingtheDimensionalityofDatawithNeural
Networks,” science, vol.313,no.5786,pp.504–507,2006.
[13] D. Erhan,Y.Bengio,A.Courville,P.-A.Manzagol,P.Vincent,andS.Bengio,“WhyDoes
UnsupervisedPre-trainingHelpDeepLearning?” Journal ofMachineLearningResearch,
vol.11,no.Feb,pp.625–660,2010.
[14] J. Dong,X.Mao,C.Shen,andY.Yang,“Unsupervisedfeaturelearningwithsymmetrically
connected convolutionaldenoisingauto-encoders,” CoRR, vol.abs/1611.09119,2016.
[Online]. Available:http://arxiv.org/abs/1611.09119
[15] P.Vincent,H.Larochelle,I.Lajoie,Y.Bengio,andP.-A.Manzagol,“StackedDenoising
Autoencoders:LearningUsefulRepresentationsinaDeepNetworkwithaLocalDenoising
Criterion,” Journal ofMachineLearningResearch, vol.11,no.Dec,pp.3371–3408,2010.
論文全文使用權限:不同意授權