Indoor Positioning System Based on Received Signal Strength (RSS) Fingerprinting : Case in Politeknik Caltex Riau Muhammad Diono
Nana Rachmana
Telecommunication Engineering Politeknik Caltex Riau Pekanbaru, Indonesia
[email protected]
School of Electric and Informatic Engineering Institut Teknologi Bandung Bandung, Indonesia
[email protected]
Abstract— Location detection system utilizing wireless LAN technology is an interesting field to study. The technique is widely used by researchers to build an indoor location detection system is fingerprint. Fingerprint is an information gathering phase of the received signal strength (RSS) by coordinate measuring instrument in particular. The system is able to overcome the limitations of GPS is not capable of providing location information in an indoor area. The results were analyzed by using a location estimate 80 test data and 1040 data from the fingerprint that has been selected. Accuracy of the results of this test reached 72.5% for the coordinate accuracy with an average error distance of 2 meters. keywords—location detection, WLAN, Fingerprinting
I. INTRODUCTION Location detection system in indoor areas is an application that utilizes wireless technology to obtain information on the location of the indoor area[1]. The development of indoor positioning system is affected by the development of mobile devices that have wireless communication support. Location detection system in indoor areas can be used to detect the presence of people, the existence of an asset, and navigation. GPS location detection technology is very popular nowadays. GPS is able to provide accurate location information of the object. However, GPS has its limitations. GPS is not able to do so because GPS signals cannot penetrate the building structure[2]. Therefore, it needs new technologies and systems that can be used to build an application system in indoor areas. There are two things that must be considered in establishing the location detection system in indoor areas, namely the selection of the technology used and the technique of measurement or estimation of the location. There are two approaches in the selection of technology for building indoor positioning system. The first approach is by designing their own wireless technology[3]. The second approach is to use existing technologies. WLAN is one
technology that is widely used in building location detection system in indoor areas[4]. WLAN is a wireless technology that is often found in the campus area, offices, and shopping centers. Once the technology, the next process is to determine the measurement technique. Several measurement techniques have been used to establish the location detection system in indoor areas. Among others, proximity sensing, lateration, angulation, dread reckoning, and pattern recognition. II. RELATED RESEARCH Research on indoor positioning system has been carried out. These studies use a variety of technologies to predict the position of an object in the building. Research [5] using the GSM signal strength to estimate the position of an object in the building. Study using WiFi signal strength ( IEEE 802.11g ) as a reference for the location estimation. Signal strength was collected using a laptop as measured using NetSurveyor application[6]. Recording the signal strength is done manually using paper records and ms. excel. Location estimation done using Naïve Bayes classification algorithm[7]. Another study is using mobile devices to collect data. The process of fingerprint every point made by measuring the signal strength captured by mobile devices. This study compares three classification algorithms, namely MLP NN, GR NN and k NN. This study resulted in the conclusion that the k - NN algorithm is the best classification algorithm[8]. This research using WIFI signal strength to estimate the location. The results of the fingerprint data is raw data which will be processed again. Data processing results in fingerprint filter and selected to determine the values of measurement results that are considered irrelevant[9]. Stage location estimation done using clustering[10] and classification techniques. Clustering is used to select and filter all data of fingerprinting. The aim is that the data would be much less tested. So that, the computation times becomes faster. After the
978-1-4799-7447-4/14/$31.00 ©2014 IEEE
clustering process is done, then the location estimates continued using classification algorithms. The use of RFID technology for building systems in the area of indoor location detection utilizing the communication between RFID tags with RFID reader. This system can be used to monitor assets and navigation as practiced by Ekahau. Determination of the position of an object is done by detecting the connection between the RFID tag and the RFID reader[11]. This system has drawbacks in terms of implementation costs. To be able to detect the position of an object in an area, the RFID reader must be installed on the entire area to be detected. Furthermore, the wireless technology that can be used to establish the location detection system in indoor areas is the Global System for Mobile communication or known as GSM. GSM is a technology that can be found anywhere. GSM signal is able to serve the area or outdoor space. Therefore GSM utilized to establish the location detection system in indoor areas. Resolution which can be detected by using GSM is approximately > 10 meters[12]. This led to GSM is not suitable to be applied in the indoor area. Furthermore, to measure the GSM signal takes the measured tool or special applications. Not all mobile devices have the ability to measure signals from GSM. One of the applications that can be used to measure the GSM signal is TEMS software. WLAN is one technology that is widely used in building location detection system in indoor areas. WLAN is a wireless technology that is often found in the campus area, offices, and shopping centers. Basically WLAN use is not specific to location detection system. Rather to for data communication. WLAN is widely used by researchers to establish the location detection system in indoor areas because it has several advantages. First, the WLAN works on 2.4 GHz frequency that can penetrate walls. Secondly, no need for line of sight conditions between the transmitter and receiver. This advantage makes the WLAN is very appropriate for use in building an indoor positioning system applications. RADAR uses WLAN to build a location detection system in indoor areas[13]. The study was conducted by measuring the signal level at some reference point to three transmitters. For the estimation process were used nearest neighbour classification algorithm. These results explain some of the things that affect the results of the estimation in an indoor location detection system in indoor areas, namely: user orientation, the amount of training data , and the amount of test data[14] . Once the technology, the next process is to determine the measurement technique. Several measurement techniques have been used to establish the location detection system in indoor areas. among others, proximity sensing, lateration, angulation, dread reckoning[15], and pattern recognition[16].
III. RESEARCH METHODOLOGY This research was conducted through the stages several stages as shown in the picture below
A. Preparation of systems and locations Prepare all the needs of software and hardware used. The location chosen was 6 block area 2nd floor Polytechnic Caltex Riau campus. In the area of the room used 4 of 6 existing space. This area is then performed on the manufacturing grid. Fig 1 Research methodology
B. Fingerprinting application[17] This study made use of fingerprinting applications on Androidbased smartphones. This application will connect to a server. Use of this application is intended to speed up the process of fingerprinting
Fig 2 fingerprinting application
A. Making grid Making grid aims to create a reference point where the level of any signal from the access points will be measured. Grid has made 2 x 2 m. Making the grid is done by measuring the entire area to determine the area to be used. Furthermore, Figure III.2 showed a grid implementation on the Politeknik Caltex Riau campus area. Altogether there are 104 points that are used as reference points
B. Training phase After grid creation process is completed, the next process is to measure the signal level on each point of reference points. This process is known as the training phase. Measurements were performed done using fingerprinting applications. At any point of reference measurements performed 10 times.
Fig 4 fingerprinting process
AP 2
RSS2
AP 1
AP 3 RSS3
RSS1
Fig 3 grid
C. Location estimation At this stage, 80 tests data will be processed to obtain estimates of the location for each of the test data. Classification algorithms k – NN and naive bayes will be use to make location estimation. K-NN classification algorithm is measured by the distance from the nearest neighbor value. The value of k is the number of neighbours specified. . The equation for the k-NN algorithm can be seen in the formula 1[18] 2
d= ∑
(1)
RSSIci = RSS value on training phase RSSIpi = RSS value on positioning phase
Server
Fig 5 communication system beetwen server and application
This research will be given a margin of error of 1 meter. This means that estimates of the position will remain accurate to say if the results produced by the system estimates were within the threshold of real coordinates. The next test is done by calculating the distance error calculation the value of the difference between the coordinates of the estimated results with actual coordinates. In this study, the distance error is calculated using the formula 2.
IV. ANALYSIS Accuracy analysis is performed in the final stage of this research. At this stage it will be done by comparing the estimation system with the real results are taken while collecting the test data. Figure III.4 show the fingerprinting process.
-
(2)
A. Signal pattern. The next process is seen in the pattern of distribution of the signal on each - each of the three reference points of access points that have been determined. This pattern indicates the received signal level on each - each reference point. Figure IV.1, VI.2, and VI.3 shows the signal distribution patterns of RSS1, RSS2, and RSS3.
Fig 8 RSS3
Fig 6 RSS1
B. Data analysis In this study, the signal level measured at the reference point 104. At any point of reference measurements performed 10 times. Table IV.1 show measurement data RSS at a predetermined reference point. From the measurement results obtained 1040 as the data. Distribution of signals obtained can be seen in Figure TABLE IV.1 Database fingerprinting RSS1 (dBm)
RSS2 (dBm)
RSS3 (dBm)
Koordinat x|y
Kelas
2
-89 -91
-87 -82
-73 -69
5 | 37 8 | 38
234 234
…
…
…
…
…
…
1024
-93
-94
-82
21 | 36
No. 1
240
Fig 7 RSS2
signal strength (dBm)
-120 -100 -80 -60 -40 -20 1
101 201 301 401 501 601 701 801 901 1001
0 cacah data RSS1
RSS2
RSS3
Fig 9 Distribution of received signal strength
9th Work. Positioning, Navig. Commun., no. 3, pp. 187–192, Mar. 2012.
75 [7]
S. S. K. Chairani, Widyawan, “Indoor Localization berbasis RSS Fingerprint Menggunakan IEEE 802.11n,” Conf. Inf. Technol. SESI Indones. Dep. Electr. Eng., no. July, 2012.
[8]
C. Beder, A. McGibney, and M. Klepal, “Predicting the expected accuracy for fingerprinting based WiFi localisation systems,” 2011 Int. Conf. Indoor Position. Indoor Navig., pp. 1–6, Sep. 2011.
[9]
X. Liang, X. Gou, and Y. Liu, “FINGERPRINT-BASED LOCATION POSITONING.”
[10]
H. Cheng and F. Wang, “Clustering algorithms research for device-clustering localization,” no. November, 2012.
[11]
M. Bouet and A. L. dos Santos, “RFID tags: Positioning principles and localization techniques,” 2008 1st IFIP Wirel. Days, pp. 1–5, Nov. 2008.
[12]
K. Heurtefeux and F. Valois, “Is RSSI a Good Choice for Localization in Wireless Sensor Network?,” 2012 IEEE 26th Int. Conf. Adv. Inf. Netw. Appl., pp. 732–739, Mar. 2012.
[13]
P. Bahl and V. N. Padmanabhan, “RADAR: an in-building RF-based user location and tracking system,” Proc. IEEE INFOCOM 2000. Conf. Comput. Commun. Ninet. Annu. Jt. Conf. IEEE Comput. Commun. Soc. (Cat. No.00CH37064), vol. 2, pp. 775–784.
[14]
H. Liu, S. Member, H. Darabi, P. Banerjee, and J. Liu, “Survey of Wireless Indoor Positioning Techniques and Systems,” vol. 37, no. 6, pp. 1067–1080, 2007.
I. Kim, E. Choi, and H. Oh, “Observation and motion models for indoor pedestrian tracking,” 2012 Second Int. Conf. Digit. Inf. Commun. Technol. it’s Appl., pp. 482–485, May 2012.
[15]
Y. Xu, Y. Sun, and S. Member, “Neural Network-Based Accuracy Enhancement Method for WLAN Indoor Positioning,” pp. 1–5, 2012.
Y. Kim, H. Shin, and H. Cha, “Smartphone-based Wi-Fi Pedestrian-Tracking System Tolerating the RSS Variance Problem,” no. March, pp. 11–19, 2012.
[16]
I. Bisio, F. Lavagetto, M. Marchese, M. Pastorino, and A. Randazzo, “Trainingless fingerprinting-based indoor positioning algorithms with Smartphones using electromagnetic propagation models,” 2012 IEEE Int. Conf. Imaging Syst. Tech. Proc., pp. 190–194, Jul. 2012.
70 Accuracy(%)
65 60 55 50 45 40 1
2 toleransi 1 m
3 2x2
I. CONCLUSION
Fig 10 accuracy improvement
From the analysis conducted found that the algorithm V. CONCLUSION K nearest neighboor provide results that are more accurate location estimates than naive Bayes algorithm. K nearest neighbor gives 50% accuracy rate of 37.5% and Naive Bayes. Grid size used determines the accuracy of the estimation. The expansion of the grid is done to raise the level of accuracy of 50% to 72.5%. To obtain the ideal signal distribution pattern should use a wireless device that position can be determined. To build an indoor detection systems should use a wider combination with a clustering algorithm. The goal is to filter the data to be estimated VI. REFERENCES [1]
[2]
[3]
J. Machaj, P. Brida, and R. Piche, “Rank based fingerprinting algorithm for indoor positioning,” 2011 Int. Conf. Indoor Position. Indoor Navig., pp. 1–6, Sep. 2011.
[4]
C. Feng, S. Member, W. Sy, A. Au, S. Valaee, S. Member, and Z. Tan, “Received-Signal-Strength-Based Indoor Positioning Using Compressive Sensing,” vol. 11, no. 12, pp. 1983–1993, 2012.
[17]
Muhammad Ihsan Zul, Mochammad Susantok, Muhammad Diono, Hendra, Ari Kurniawan, “RSS Fingerprint Berbasis Mobile untuk Estimasi Lokasi Menggunakan Wireless LAN (4),” Appl. Bus. Eng. Conf., 2013.
[5]
L. E. N. Hani Rubiani, Widyawan, “GSM Fingerprint Untuk Deteksi Lokasi Dalam Gedung Dengan Menggunakan Algoritma Naive Bayes,” Conf. Inf. Technol. SESI Indones. Dep. Electr. Eng., no. July, 2012.
[18]
S. Gansemer, U. Großmann, and S. Hakobyan, “RSSI-based Euclidean Distance Algorithm for Indoor Positioning adapted for the use in dynamically changing WLAN environments and multi-level buildings,” no. September, pp. 15–17, 2010.
[6]
S. Shrestha, E. Laitinen, J. Talvitie, and E. S. Lohan, “RSSI channel effects in cellular and WLAN positioning,” 2012
