IPSJ SIG Technical Report Vol.2013-HCI-155 No.6 Vol.2013-UBI-40 No /11/5 Wi-Fi Wi-Fi Wi-Fi Wi-Fi Wi-Fi Wi-Fi,, MLLR Low cost Wi-Fi Indoor Positi

Size: px

Start display at page:

Download "IPSJ SIG Technical Report Vol.2013-HCI-155 No.6 Vol.2013-UBI-40 No /11/5 Wi-Fi Wi-Fi Wi-Fi Wi-Fi Wi-Fi Wi-Fi,, MLLR Low cost Wi-Fi Indoor Positi"

Bruno Hubbard
5 years ago
Views:

1 Wi-Fi Wi-Fi Wi-Fi Wi-Fi Wi-Fi Wi-Fi,, MLLR Low cost Wi-Fi Indoor Positioning Model for change of scan data Ono Hiroshi Maekawa Takuya Abstract: This paper proposes a new method for constructing a low cost Wi-Fi indoor positioning model. Generally, a Wi-Fi indoor positioning model is built by using scan data observed at each reference point in a short period. However, we cannot deal well with unstable signal strength from Wi-Fi APs when we use such scan data as training data. In addition, we cannot distinguish stable Wi-Fi APs from unstable APs (e.g., mobile APs), which are inappropriate for using as beacon, with such scan data. When an indoor positioning model is built by using scan data at each reference point in a long period, we need not to care about the problems. However, the approach requires a large cost as regards collecting training scan data. In this work, we collect scan data in a long period at only few reference points and complement signal strength models of the other reference points with short-term scan data by utilizing the long-term scan data. That is, a signal strength model of a reference point with long-term scan data is adapted to a model of a reference point with short-term scan data. In addition, we try to distinguish stable Wi-Fi APs with unstable APs from the long-term scan data. Keywords: Wi-Fi fingerprinting, indoor positioning, MLLR adaptation 1. Wi-Fi LAN 1 Graduate School of Information Science and Technology, Osaka University, Suita, Osaka, Japan Wi-Fi GPS Wi-Fi 1

2 Wi-Fi ( ) ( ) Wi-Fi Wi-Fi Wi-Fi (AP) ( ) ( ) Wi-Fi ( 1 ) Wi-Fi AP Wi-Fi ( 2 ) AP AP AP AP 2 ( 1 ) AP AP AP ( 2 ) Wi-Fi [1] AP Wi-Fi Naive Bayes SVM k (knn) [2] Wi-Fi AP Nishio [3] AP AP AP AP 2.2 Jiang [4] WiFi Pulkkinen [5] WiFi WiFi Chai [6] WiFi Wang [7] PDR Pedestrian Dead-Reckoning 2

3 Wi-Fi Wi-Fi 安定した AP の検出の観測密な観測安定した AP 時刻 2.3 Wi-Fi S. Chen [8] Y.C. Chen [9] Yin [10] Wi-Fi Wi-Fi Wi-Fi Pan [11] Wi-Fi Manifold co-regularization 2.4 Ohmura maximum-likelihood linearregression (MLLR) maximum a posteriori (MAP) [12], [13] [14] の観測近傍参照点の決定最も近距離の長期間参照点を近傍参照点とする RP1 に関する近傍参照点モデル適応 RP1 近傍参照点モデル短期間参照点短期間参照点モデル Fig. 1 疎な観測 RP2, RP3 に関する近傍参照点 RP2 RP3 AP3 長期間参照点近傍参照点のモデルを短期間参照点のモデルに適応させる 1 不安定な AP Outline of proposed method 時刻モデル適応後の短期間モデル AP AP AP AP 3

4 AP 3.2 AP AP AP F i AP AP F i = N i N N N i i AP AP AP AP 2 Wi-Fi AP Wi-Fi [8] 観測頻度電波強度 (db) 2 Fig. 2 Histogram of signal strength 2 :GMM GMM EM EM Expectation Step(E ) Maximization Step(M ) GMM x k LD k (x) = M π i N (x µ i, Σ i ) i=1 M GMM π i i µ i Σ i i GMM MLLR (GMM) ˆµ = Aµ + b µ A k k k AP b AP A b () 4

5 ˆµ i = µ i (µ MAXπi µ MAXπi ) 縦 :6.3m 横 : 7.3m µ i i AP µ MAXπi i AP µ MAXπi i AP 3.5 Wi-Fi GMM) knn GMM Wi-Fi GMM GMM) k k i=1 P os(x) = LD i(x) P os i k i=1 LD i(x) (1) 1 k x LD i (x) x i P os i i Wi-Fi Google Nexus One HTC Nexus One Wi-Fi OS Android 1 3 参照点長期間 Wi-Fi 電波強度取得点 3 Fig. 3 Floor plan of experimental environment Nexus One Wi-Fi 1 (1 Wi-Fi Nexus One 2 (1 Wi-Fi ) AP AP AP 4.2 : AP 4 AP 4 AP AP AP AP 5

6 A P 数平均推定誤差 (cm) 観測頻度 4 AP Fig. 4 Histogram of observation frequency 全て閾値 0.5 閾値 0.7 閾値 0.9 フロア内 5 : AP Fig. 5 Result of detection of stable APs AP AP AP AP 15cm AP AP AP 8 AP AP AP AP AP 平均推定誤差 (cm) 閾値 0.5 閾値 0.7 閾値 : Fig. 6 Result of model adaptation 従来手法提案手法 4.3 : AP cm 20cm 1 AP Table 1 Number of APs used in each method AP Wi-Fi AP AP AP AP 30cm 6

7 A-STEP AS251Z02124H [1] Kaemarungsi, K.: Modeling of indoor positioning systems based on location fingerprinting (2013). [2] Chaudhuri, K., Sanghi, D. and Bhagwat, P.: Location determination of a mobile device using IEEE b access point signals, IEEE Wireless Communications & Networking Conference (WCNC), pp (2003). [3] Nishio, N., Fukuzaki, Y. and Azumi, T.: Detecting wifi base station behavior inappropriate for positioning method in participatory sensing logs, UbiComp(Adjunct Publication) 2013, pp (2013). [4] Jiang, Y., Pan, X., Li, K., Lv, Q., Dick, R. P., Hannigan, M. and Shang, L.: ARIEL: Automatic wi-fi based room fingerprinting for indoor localization, Proc. of the 2012 ACM Conference on Ubiquitous Computing (UbiComp 12), pp (2012). [5] Pulkkinen, T., Roos, T. and Myllymäki, P.: Semisupervised Learning for WLAN Positioning, Proc. of the 21st International Conference on Artificial Neural Networks and Machine Learning (ICANN 2011), Vol. 6791, pp (2011). [6] Chai, X. and Yang, Q.: Reducing the calibration effort for location estimation using unlabeled samples, Proc. of Third IEEE International Conference on Pervasive Computing and Communications (PerCom 2005), pp (2005). [7] Wang, H., Sen, S., Elgohary, A., Farid, M., Youssef, M. and Choudhury, R. R.: Unsupervised indoor localization, Proc. of the 10th International Conference on Mobile Systems,Applications and Services (MobiSys 12), pp (2012). [8] Chen, S., Chen, Y. and Trappe, W.: Exploiting environmental properties for wireless localization and location aware applications, Proc. of Sixth Annual IEEE International Conference on Pervasive Computing and Communications (PerCom 2008), pp (2008). [9] Chen, Y.-C., Chiang, J.-R., Chu, H.-h., Huang, P. and Tsui, A. W.: Sensor-assisted wi-fi indoor location system for adapting to environmental dynamics, Proc. of the 8th ACM International Symposium on Modeling, Analysis and Simulation of Wireless and Mobile Systems (MSWiM 05), pp (2005). [10] Yin, J., Yang, Q. and Ni, L.: Adaptive temporal radio maps for indoor location estimation, Proc. of Third IEEE International Conference on Pervasive Computing and Communications (PerCom 2005), pp (2005). [11] Pan, S. J., Kwok, J. T., Yang, Q. and Pan, J. J.: Adaptive localization in a dynamic WiFi environment through multi-view learning, Proc. of the 22nd National Conference on Artificial Intelligence (AAAI 07), Vol. 2, pp (2007). [12] Gauvain, J. and Lee, C.: Maximum a posteriori estimation for multivariate gaussian mixture observations of markov chains, IEEE Trans. on Speech and Audio Processing, Vol. 2(2), pp (2002). [13] Leggetter, C. and Woodland, P.: Maximum likelihood linear regression for speaker adaptation of continuous density hidden Markov models, Computer Speech & Language, Vol. 9(2), pp (1995). [14] Ohmura, R., Hashida, N. and Imai, M.: Preliminary evaluation of personal adaptation techniques in accelerometerbased activity recognition, In Int l Symo. on Wearable Computers: Late Breaking Results (2009). 7

CER7027B / CER7032B / CER7042B / CER7042BA / CER7052B CER8042B / CER8065B CER1042B / CER1065B CER1242B / CER1257B / CER1277B

CER7027B / CER7032B / CER7042B / CER7042BA / CER7052B CER8042B / CER8065B CER1042B / CER1065B CER1242B / CER1257B / CER1277B 一般機器用 For Consumer Products 汎用パワーインダクタ Common Power Inductors CER-B series RoHS CER727B / CER732B / CER742B / CER742BA / CER752B CER842B / CER865B CER42B / CER65B CER242B / CER257B / CER277B 特徴 DC-DC コンバータ用インダクタとして最適