Open Access
Add to BibSonomy
Abstract
The indoor user localization function is in high demand for high-speed wireless communications, navigations and smart-home applications. The optical wireless technology has been used to localize end users in indoor environments. However, its accuracy is typically very limited, due to the ambient light, which is relatively strong. In this paper, a novel high-localization-accuracy optical wireless based indoor localization system, based on the use of the mechanism that estimates background light intensity, is proposed. Both theoretical studies and demonstration experiments are carried out. Experimental results show that the accuracy of the proposed optical wireless indoor localization system is independent on the localization light strength, and that an average localization error as small as 2.5 cm is attained, which is 80% better than the accuracy of previously reported optical wireless indoor localization systems.
© 2017 Optical Society of America
1. Introduction
User localization is an important function which has attracted considerable attention over the past decade [1–3], because it significantly improves the quality of service of in-building and indoor wireless communication systems. Indoor localization is particularly highly demanded in high-speed wireless communications, including millimeter-wave and optical-wireless based systems, where only limited coverage and mobility can be provided [4, 5]. Indoor localization also enables other location-based services, such as indoor navigation, location-based advertisements, and proximity-based notifications. In addition, the indoor localization function also plays a vital role in smart-home applications, including the detection of emergencies and sudden falls for elderly people, and the improvement of the performance of Ambient Assistant Living (AAL) systems, which provide smart location-based assistance for personalized daily living and entertainments [6].
The Global Positioning System (GPS) has been widely used in outdoor localization and navigations. However, using GPS in indoor environments is challenging, mainly due to the signal blockage and attenuation issues [7]. To achieve indoor localization, a number of potential schemes have been proposed and studied, including schemes based on the use of radio-frequency (RF), image processing and pyro-electric sensors [8–11]. In RF indoor localization systems, both WiFi and Zigbee technologies have been utilized. However, due to the severe multipath dispersion resulting from signal reflections by walls and furniture, in general the achievable accuracy is limited, typically in the order of tens of centimeters to several meters, depending on the localization mechanism and system architecture. The imaging-sensor-based scheme is capable of providing accurate positioning information (centimeter level). Nevertheless, the typical localization speed is limited and personal privacy is a primary concern to most end users. The use of infrared pyro-electric sensors for indoor positioning is based on monitoring changes in heat radiation, and tens of centimeters or even centimeters accuracy can be achieved. However, such sensors are typically only capable of detecting moving items and are vulnerable to environmental changes.
To provide accurate indoor localization, the use of optical wireless technology has also been proposed and experimentally investigated. Both visible light communication (VLC) [2] and infrared optical wireless based solutions have previously been demonstrated [12]. The VLC based solution can realize both illumination and localization simultaneously, whilst the infrared system can provide the localization function without being sensitive to users, thus overcoming the interference issue that users typically experience. In our previous studies, we have proposed and demonstrated an indoor infrared optical wireless based localization system [12], and shown that by using the received signal strength information, an average localization accuracy of better than 20 cm has been experimentally demonstrated. However, achieving better localization accuracy is challenging and the localization performance is typically vulnerable to background light, which is always present in indoor environments due to both the sunlight and illumination lamps [13].
To further improve the localization accuracy and to provide better robustness against background light, we propose, in this paper, an indoor infrared optical wireless localization system with background light power estimation capability [14]. The background light power estimation is achieved by sending out two localization signals with different powers in subsequent time slots. Due to the relatively stable condition and slow dynamics (such as walking speed and temperature change) in indoor environments, the background light power information can be obtained by measuring the received power levels. Both theoretical studies and demonstrated experiments are carried out. Measurement results show that an average localization accuracy of about 2.5 cm can be achieved with a localization beam footprint of 1 m, which represents an improvement of over 80%, in comparison with the accuracy attained with previously reported optical wireless indoor localization systems [12]. Moreover, the dependence of the localization accuracy on the beam footprint and signal transmission power is investigated experimentally, and results show that both the beam footprint and the signal transmission power have negligible impact on the localization performance.
2. Indoor infrared optical wireless localization system with background power estimation capability
2.1 Localization system architecture
The architecture of proposed indoor infrared optical wireless localization system is shown in Fig. 1. It is similar to the system used for high-speed indoor optical wireless communications [15]. The localization function is inherently provided through the infrared optical wireless localization system. Since a module dedicated to localization is not necessary, the complexity and cost of the indoor infrared optical wireless localization system shown in Fig. 1 is significantly reduced. In brief, the Central Office (CO)/Access Points architecture is adopted with multiple rooms being connected to the same CO, for complexity and cost sharing as well as centralized control functions. Inside each room, ceiling mounted fiber transceivers are employed, and each downlink transmitter consists of a fiber end, a lens for beam divergence adjustment, and MEMS based steering mirrors for changing the beam propagation orientation. After free-space propagation, the signal, through a direct line-of-sight link, is collected by the optical receiver. Here, the simplest non-imaging receiver architecture is used, which includes a compound parabolic concentrator (CPC) and a photo-detector. In the uplink direction, VCSEL serves as the light source, and the same lens and MEMS steering mirror structure is utilized. The uplink signal can be either detected directly or coupled back to the CO for centralized signal detection, processing and characterization. For localization, the direct detection scheme is used.
Fig. 1 Architecture of the proposed indoor infrared optical wireless localization system. CPC: compound parabolic concentrator and PD: photo-detector.
2.2 Indoor localization function
We have previously realized indoor localization based on the use of the “search and scan” process in conjunction with the infrared optical wireless technology [16], where the room is typically divided into multiple cells, and the ceiling mounted fiber transceiver is used to scan the entire room while sending one specific code (“cell number” used) to each cell. When receiving the specific code, the user copies it and sends it back to the fiber transceiver. Through the feedback information, a rough estimation of the location of the user is obtained and the localization accuracy is typically equal to the cell size. We have also proposed to use the received signal strength information to improve the localization accuracy [12]. This is based on the approximately Gaussian intensity distribution of the signal beam after free-space propagation. Therefore, the received power can be expressed by
where Sr is the downlink receiver collection area, Pt is the transmitted localization signal power, which is typically included in the “scan and search” message, ω is the localization beam footprint at the user end, r is the distance between the user’s location and the beam center on the reception plane, and Pr is the measured received optical power. Since the light field distribution is approximately Gaussian, the beam footprint ω here is defined as the Gaussian beam waist - the distance from the beam’s center at which the beam intensity is 1/e2 of its value on the center. Here it is assumed that the receiver is located on a reception plane with fixed height, and hence the beam footprint at the user side is fixed and known by setting the beam divergence angle at the transmitter side. By including both the transmitted localization signal power (Pt) at the fiber transceiver end and the beam footprint (ω) at the user end into the “scan and search” process and monitoring the received optical power (Pr), the distance r can be estimated using Eq. (1). This localization process typically results in a circle from the beam center, and the exact coordinates (x and y) of a user can be then obtained by steering the localization beam in both x and y directions by half of the cell size and sending out the “scan and search” message again, which results in a different estimated distance r. By combing the two r values and the two beam centers information, which is always pre-known, exact coordinates of the user can be calculated.However, in practice, the received optical power measured at the user end (Pr) always contains two parts – the signal power, Prs, and the background light power, Prb, induced by both sunlight and indoor illumination lamps. Therefore, the accuracy of the location estimation based on Eq. (1) decreases with increasing the background light.
To mitigate accuracy degradation, we propose a new approach based on transmitting two localization signals of different power levels, namely, Pt1 and Pt2, at different time instances. Due to the slow dynamics feature in typical indoor environments, where the user’s moving speed is typically low and the environmental changes are slow and infrequent, the background light power collected during the two subsequent time instances can be considered almost constant and the user location unchanged. Therefore, the received signal strengths (Prs1 and Prs2) for the two localization signals can be expressed as
where Pr1 and Pr2 are the measured received optical powers. The beam footprint is kept unchanged during these two time instances. Using Eqs. (2) and (3), the received background light power is given byWith the collected background light power being estimated, the localization accuracy can be improved significantly. The distance between the user’s location and the beam center r can now be calculated as follows
Therefore, by steering the localization beam along both x and y directions by half of the beam footprint and performing the same measurements and estimations, the exact location (coordinates information) of the user can be obtained. The CPC based non-imaging receiver is assumed to be used in the proposed localization system, which provides almost constant concentrator gain as long as the signal incident angle is smaller than the field-of-view of the CPC. Therefore, the impact of signal incident angle here is neglected.
When the background light comes from sunlight, skylight and incandescent lamps, it is typically unmodulated with spectral distribution mainly near the DC. For background light from fluorescent lamps, especially those driven by electronic ballasts, the spectral distribution has significant components with frequencies up to the MHz range. Therefore, if baseband localization signal is used, which has significant spectral components at or near DC, the use of DC-blocks to suppress the background light will result in significant localization signal distortion and reduced received localization signal power, thus affecting the system performance with additional localization error. As a result, the proposed localization system with background light estimation capability is needed.
3. Experiments and results
3.1 Experimental setup
The proposed indoor infrared optical wireless localization system with background light power estimation capability was experimentally demonstrated, using the setup shown in Fig. 2. The downlink localization signal was generated in a central office and then transmitted to the room via the in-building fiber distribution network, which was emulated by 5.6 km single-mode fiber. A lens and MEMS-based steering mirrors were used in the fiber transceiver to control the divergence and propagation direction of the localization signal. After free-space propagation, the signal was collected by a non-imaging CPC (with 45° field-of-view) and detected with a photodiode at the subscriber unit. The localization signal was characterized by a bit-error-rate tester (BERT) to obtain the contained information, such as the transmission power, the beam footprint, and the beam center location. To measure the received localization signal power with reduced measurement error, an optical power meter with a photo-sensitive area larger than the exit surface of the CPC was utilized. For the uplink direction, a low-cost VCSEL, served as the light source, and it was directly modulated by the upstream data. Another lens and MEMS-based steering mirrors were also used in the uplink to generate the required localization feedback beam. The signal then propagated via the free-space to the fiber transceiver before being collected and detected.
In the demonstration experiments, strong background light from table lamps and overhead lamps were incorporated. The definition of the coordinates used is shown in Fig. 2, and the subscriber unit moved along the x-direction or the y-direction to emulate user movements. The downlink signal wavelength was 1533.33 nm and the uplink wavelength was in the 850 nm band, due to the availability of mature and low-cost VCSELs. The localization signal bit rate was 50 Mb/s, and the simplest on-off-keying (OOK) modulation format was used.
3.2 Results and discussions
In the experiments, the localization beam footprint after free-space propagation was fixed at 1 m and the beam center was located at (x, y) = (0.5 m, 0.5 m). The two different power levels, Pt1 and Pt2, of the transmitted localization signals were 2 mW and 2.5 mW, respectively. The signal free-space propagation distance in the z-direction was 2 m. When the subscriber unit was moved along the y = 0.3 m line, the localization results with the background light power estimation method proposed here are shown in Fig. 3. The results without background light power estimation are also shown. The average localization error with and without background light power estimation is about 2.4 cm and 15.7 cm, respectively. It is clear that with the proposed background light power estimation method, much better localization accuracy can be achieved. This was mainly because the background light power can be calculated and its impact can be suppressed. The remaining localization error was mainly due to the fact that the beam intensity profile was not exactly Gaussian distribution at the receiver side, which was a systematic error.
Fig. 3 Localization experimental results with and without the proposed background light power estimation method. y = 0.3 m.
More measurements were carried out when the subscriber unit was moved along the y = 0.1 m, y = 0.2 m, y = 0.3 m, and y = 0.4 m lines. The localization error is summarized in Fig. 4. The beam footprint at the user end was still 1 m, the free-space signal propagation distance in the z-direction was fixed at 2 m, and the two transmission powers Pt1 and Pt2 were 2 mW and 2.5 mW, respectively. It can be seen from the results displayed in Fig. 4 that the impact of background light collected at the receiver side was successfully suppressed. The average localization errors were about 2.13 cm, 2.67 cm, 2.42 cm, and 2.56 cm, respectively, and the maximum error was limited to about 5.27 cm. Compared with the localization results obtained using the previously demonstrated method without background light power estimation [12], a localization accuracy improvement exceeding 80% was attained.
In the proposed system, since the collected background light power is estimated from the two different power levels of the transmitted localization signal, the impact of transmission power levels on the localization performance was investigated experimentally, using the same setup as shown in Fig. 2. For Pt1 = 3 mW and Pt2 = 5 mW, the estimated user location results are shown in Fig. 5. The localization beam footprint was still fixed at 1 m and the signal free-space transmission distance in the z-direction was kept at 2 m. The obtained average localization accuracy was about 2.66 cm and the maximum error is about 5.11 cm. Compared with the previous case where Pt1 = 2 mW and Pt2 = 2.5 mW, a comparable localization performance was achieved. This is because the received signal power is proportional to the transmission power, and as can be seen from Eqs. (1)-(5), the proposed indoor localization system performance is independent of the transmission power levels.
Fig. 5 Estimated user location with the proposed background light power estimation method. Transmission powers were 3 mW and 5 mW, respectively.
Another localization scenario was considered, where the difference between the two power levels of the transmitted localization signal is comparatively small. In the characterization experiments, the two power levels were selected at Pt1 = 3 mW and Pt2 = 3.02 mW, and the measurement results are shown in Fig. 6. When the subscriber unit was moved along the y = 0.3 m and y = 0.4 m lines, the average localization error was 5.15 cm and 5.11 cm, respectively, and the maximum error was around 9.5 cm. This increase in error, in comparison with previous scenarios, was mainly because of the similar received localization signal power levels at the subscriber side, which resulted in inaccurate background light power estimation. Therefore, in practice, the difference between two transmitted localization signal power levels needs to be relatively large in order to minimize the localization error.
Fig. 6 Estimated user location with the proposed background light power estimation method. Transmission powers were 3 mW and 3.02 mW, respectively.
The impact of the localization beam footprint after free-space signal transmission on the performance of the proposed indoor localization system was also experimentally investigated, and the measurement results are shown in Fig. 7. Three different localization beam footprints were investigated, by changing the distance between the fiber end and the diffraction lens inside the fiber transceiver. The two transmission powers were set at Pt1 = 2 mW and Pt2 = 2.5 mW, and the beam center was located at (x,y) = (0.5 m, 0.5 m). The average localization errors were about 2.86 cm, 2.42 cm, and 2.36 cm for beam footprints of 0.8 m, 1 m, and 1.2 m, respectively. It is found that the localization error tends to decrease when the localization beam footprint is large, which is mainly because Gaussian intensity distribution approximation was used. For a larger localization beam, the intensity change over the receiver aperture is smaller, resulting in a lower systematic localization error. In addition, the dependence of the localization error on the beam footprint is relatively weak (in the millimeter level), and hence, it can be neglected in typical scenarios.
Fig. 7 User localization error with the proposed background light power estimation method under different beam footprints. Transmitted localization signal power levels were 2 mW and 2.5 mW, respectively.
4. Conclusions
A novel indoor optical wireless localization system capable of estimating the background light power has been proposed and experimentally investigated. Measuring the background light power collected at the subscriber side and sending out the “search and scan” message with two different power levels have been shown to be effective for the estimation of the background light power. Demonstration experiments have been carried out and results have shown that the impact of background light on the localization accuracy can be suppressed and an average localization error of about 2.5 cm can be realized. In addition, the impact of the localization signal transmitted power levels and the localization beam footprint on the localization performance has been studied experimentally. It has been shown that the localization error is independent from the transmission power levels as long as the difference between two transmission powers is comparatively large, and that the impact of localization beam footprint is negligible. Compared with the localization accuracy obtained previously without background light power estimation [14], an improvement of over 80% has been realized.
Funding
Australian Research Council (ARC) Discovery Early Career Researcher Award (DE150100925); ARC Discovery Program (DP170100268).
References and links
1. Z. Farid, R. Nordin, and M. Ismail, “Recent advances in wireless indoor localization techniques and systems,” J. Comput. Netw. Commun. 2013, 185138 (2013).
2. M. Yasir, S.-W. Ho, and B. N. Vellambi, “Indoor positioning system using visible light and accelerometer,” J. Lightwave Technol. 32(19), 3306–3316 (2014). [CrossRef]
3. A. Gomez, K. Shi, C. Quintana, G. Faulkner, B. C. Thomsen, and D. O’Brien, “A 50 Gb/s transparent indoor optical wireless communications link with an integrated localization and tracking system,” J. Lightwave Technol. 34(10), 2510–2517 (2016). [CrossRef]
4. E. Torkildson, U. Madhow, and M. Rodwell, “Indoor millimeter wave MIMO: feasibility and performance,” IEEE Wirel. Commun. 10(12), 4150–4160 (2011). [CrossRef]
5. C. W. Oh, E. Tangdiongga, and A. M. J. Koonen, “Steerable pencil beams for multi-Gbps indoor optical wireless communication,” Opt. Lett. 39(18), 5427–5430 (2014). [CrossRef] [PubMed]
6. S.-C. Kim, Y.-S. Jeong, and S.-O. Park, “RFID-based indoor location tracking to ensure the safety of the elderly in smart home environments,” Pers. Ubiquitous Comput. 17(8), 1699–1707 (2013). [CrossRef]
7. D. A. Maisano, J. Jamshidi, F. Franceschini, P. G. Maropoulos, L. Mastrogiacomo, A. Mileham, and G. Owen, “Indoor GPS: system functionality and initial performance evaluation,” Int. J. Manuf. Res. 3(3), 335–349 (2008). [CrossRef]
8. A. K. M. M. Hossain, H. N. Van, Y. Jin, and W.-S. Soh, “Indoor localization using multiple wireless technologies,” Proc. IEEE Int. Conference on Mobile Adhoc and Sensor Systems (2007), pp. 1–8.
9. R. C. Luo and O. Chen, “Wireless and pyroelectric sensory fusion system for indoor human/robot localization and monitoring,” IEEE/ASME Trans. Mechatron. 18(3), 845–853 (2013). [CrossRef]
10. J. Kemper and H. Linde, “Challenges of passive infrared indoor localization,” in Proc. 5th Workshop on Positioning, Navigation and Communication (2008), pp. 63–70.
11. H. Hile and G. Borriello, “Positioning and orientation in indoor environments using camera phones,” IEEE Comput. Graph. Appl. 28(4), 32–39 (2008). [CrossRef]
12. K. Wang, A. Nirmalathas, C. Lim, and E. Skafidas, “Experimental demonstration of a novel indoor optical wireless localization system for high-speed personal area networks,” Opt. Lett. 40(7), 1246–1249 (2015). [CrossRef] [PubMed]
13. G. Cossu, R. Corsini, and E. Ciaramella, “High-speed bi-directional optical wireless system in non-directed lone-of-sight configuration,” J. Lightwave Technol. 32(10), 2035–2040 (2014). [CrossRef]
14. K. Wang, A. Nirmalathas, C. Lim, and E. Skafidas, “Experimental demonstration of optical wireless indoor localization system with background light power estimation,” in Proc. Optical Fiber Communications Conference (2015), pp. W2A.63. [CrossRef]
15. K. Wang, A. Nirmalathas, C. Lim, and E. Skafidas, “High-speed optical wireless communication system for indoor applications,” IEEE Photonics Technol. Lett. 23(8), 519–521 (2011). [CrossRef]
16. K. Wang, A. Nirmalathas, C. Lim, and E. Skafidas, “4×12.5 Gb/s WDM optical wireless communication system for indoor applications,” J. Lightwave Technol. 29(13), 1988–1996 (2011). [CrossRef]
