Professional Documents
Culture Documents
net/publication/319258577
CITATIONS READS
354 4,376
3 authors, including:
All content following this page was uploaded by Junhai Luo on 01 September 2017.
1 Abstract—Advances in visible light communication (VLC) Combining with some other applications (APPs), indoor posi- 31
2 technology and the ubiquity of illumination facility have led to tioning technology can also be used to provide Location-Based 32
3 a growing interest in VLC-based indoor positioning. Numerous Services (LBS) and advertisements to users. Subsequently, 33
4 techniques have been proposed to obtain better system perfor-
many researchers endeavor to achieve the finer accuracy and
of
34
5 mance. In this paper, we survey over 100 papers ranging from
6 pioneering papers to the state-of-the-art in the field to present the precision. 35
7 positioning technology. Not only the light emitting diode technol- With the proliferation of mobile devices, wireless tech- 36
8 ogy, modulation method and types of receivers are compared but nologies are becoming significant parts in embedded systems 37
9 also a novel taxonomy method is proposed. In this paper, VLC- to provide low-cost, anytime, and anywhere connectivity [3]. 38
10 based indoor positioning systems (VLC-based-IPSs) are classified
11 based on the methods used: 1) mathematical method; 2) sensor- However, the wireless signal in the different environments has 39
12 assisted method; and 3) optimization method. Different from different characteristics. In an outdoor environment, the sig- 40
other survey works, we emphasize and analyze the accuracy of nal often propagates in an open and accessible space with a
ro
13 41
14 VLC-based-IPS in the experiment and simulation environments. small signal attenuation. However, in an indoor environment, 42
15 Meanwhile, this paper illustrates challenges, countermeasures, the movements of persons and locations of items may cause 43
16 and lessons learned.
different degrees of influence on the indoor wireless signal. 44
17 Index Terms—Visible light communication, indoor positioning For outdoor scenarios, Global Positioning System (GPS) is 45
18 system, sensor-assisted method, positioning optimization method, used extensively. However, the positioning accuracy of GPS 46
19 survey.
is located in the order of several meters [4] which is unac- 47
HE INDOOR positioning technique that relies on in- used to facilitate Indoor Positioning System (IPS). 51
21
22
23
T building communications infrastructure has drawn a lot
of attention recently. Its applications cover a wide area and
According to a recent survey [5], there are various indoor
positioning technologies proposed such as Radio-Frequency
52
53
24 have high demand values. In addition, the growing commer- Identification (RFID) [6], [7], Ultra-Wideband (UWB) [8], 54
25 cial interest in Indoor Location-Based Services (ILBS) has Wi-Fi [9], [10], and fingerprint [11]. These technologies deter- 55
26 spurred the development of indoor positioning technique [1]. It mine the relative position between them by moving targets and 56
27 is estimated that the indoor positioning market value will be $5 communication of the fixed unit and then obtain the actual 57
28 billion in 2018 [2]. At the great museums and shopping malls, locations of targets. However, there is usually a need for 58
59
29 users can utilize Indoor Positioning Systems (IPSs) to attain
30 location information and quickly find the exit or the elevator. equipment. These lead to a significant increase in the cost of 60
the Civil Aviation Administration of China under Grant U1733110, in part Due to the high transmission capacities, optical wireless 65
by the Overseas Academic Training Funds, University of Electronic Science networks play an important role in our modern information 66
and Technology of China under Grant 201506075013, and in part by the
Program for Science and Technology Support in Sichuan Province under Grant technology [12]. At present, VLC using Light Emitting Diodes 67
2014GZ0100 and Grant 2016GZ0088. (Corresponding author: Junhai Luo.) (LEDs) has become a hot topic in wireless communications. 68
J. Luo is with the School of Electronic Engineering, University of The White-LED lighting technology as a type of green light- 69
Electronic Science and Technology of China, Chengdu 611731, China, and
also with the Department of Electrical Engineering and Computer Science, ing technology is considered as the next generation indoor 70
University of Tennessee Knoxville, Knoxville, TN 37919 USA (e-mail: lighting because of the advantage of high luminous efficiency 71
junhai_luo@uestc.edu.cn). (up to 110lm/W), long service life (up to 50000hours) and 72
L. Fan is with the School of Electronic Engineering, University of
Electronic Science and Technology of China, Chengdu 611731, China (e-mail: fast response speed (up to 10MHZ). Distinct from the older 73
Digital Object Identifier 10.1109/COMST.2017.2743228 ways which mean LEDs can provide illumination as well as 77
1553-877X c 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
2 IEEE COMMUNICATIONS SURVEYS & TUTORIALS
78 communication. An important application of VLC is employed the challenges, countermeasures, and lessons learned are con- 136
79 to indoor positioning. Nowadays, plenty of investigations have tained in Section V. Finally, we conclude in Section VI. 137
84 sources, which can be readily detected by light sensors design indoor positioning systems. These positioning systems 140
85 embedded in Commercial Off-the-Shelf (COTS) devices with their own characteristics have different advantages and 141
86 (e.g., smartphone, smart glass, and smart watch). disadvantages. Compared with RF technology, VLC as a tech- 142
87 • The light strength is stable at different times of a day, nology of wireless access has the unique advantages. Hence, 143
88 avoiding site-survey and database maintenance. in this paper, we mainly discuss the VLC-based-IPSs. 144
of
91 Compared with other systems, the VLC-based-IPS has the To better meet the market demand, indoor positioning tech- 146
92 following advantages: nology is quickly developed. Basing on the requirements of 147
93 • It can be used in RF sensitive areas like hospitals and air
the positioning applications, we need to choose a particular 148
94 crafts because it does not generate RF interference. technology to realize positioning. In an indoor environment, 149
95 • Light emitting diode offers a narrow beamwidth, which
positioning is mainly achieved through the use of techniques 150
96 enables more precise Angle of Arrival (AOA) information such as Frequency Modulation (FM), Cellular Networks, 151
97 at the receiver side [17]. ZigBee, Wi-Fi, Infrared (IR), Ultrasound, Bluetooth, RFID, 152
ro
98 • VLC has little influence from multipath interference
and UWB. All of those technologies have been discussed in 153
99 because the primary energy of indoor visible light con- many other survey papers about IPS. Hence, we will not repeat 154
100 centrates on the line of sight link. them, but provide a more detailed table to show the system 155
101 • VLC-based positioning systems can be installed inexpen-
performance, advantages, constraints of using this technology. 156
102 sively since they utilize existing lighting systems with We summarize all these indoor positioning technologies used 157
103 very few modifications applied. in literature. Different technologies have different transmission 158
104 • The precision of visible light positioning is higher than
ranges, different accuracies, different scalabilities, different 159
105 traditional positioning. For example, the research shows
EP
costs and different power consumptions. We compare these 160
106 that the number of LED luminaires is ten times more than items and outline the advantages and constraints of these 161
107 Wi-Fi in a typical indoor building [18], which contributes technologies in Table I. 162
108 to the higher accuracy.
109 Previous decade observed a significant research effort
110 directed towards VLC-based positioning system. Hundreds B. VLC 163
111 of papers in this field have been issued. However, there Considering the wireless access, RF technology is mature 164
112 are few surveys on VLC-based positioning system [19]–[22]. and widely used. However, now the RF spectrum is becoming 165
113 Moreover, all of them miss some critical information about increasingly crowded. In other words, facing the rapidly grow- 166
114 the VLC-based-IPS or use the different classification method. ing demand for wireless access, the available RF bandwidth 167
115 We will do a thorough comparison among those surveys may fail to meet the requirements. Hence, researchers begin 168
IEE
116 in Section III. In our paper, we survey the related knowl- to look for other solutions. Fortunately, by using the exist- 169
117 edge of VLC-based positioning system and compare some ing illuminating devices, VLC can be utilized as one of the 170
118 pioneering systems in different ways. We utilize two more promising alternative methods for RF-based communication, 171
119 classification methods: sensor-assisted method and position- especially in indoor communications [23]. The rapid increase 172
120 ing optimization method. This type of classification method in the usage of LEDs has provided a unique opportunity for 173
121 can help researchers quickly find the thought of increasing VLC. With the advantages of lower cost, higher lifetime and 174
122 the precision based on the basic mathematical method. It energy savings, LEDs are expected to replace older illumina- 175
123 can help researchers to understand which method can be uti- tion technologies and become the next generation illumination 176
124 lized to achieve localization and what type of effect can be devices. It is anticipated that LEDs will provide nearly 75% 177
125 achieved, and provide the new ideas of the positioning method. of all lighting by the year 2030 [24]. The new LEDs can 178
126 Some challenges regarding VLC-based positioning are further switch to different light intensity levels at a very fast rate 179
127 discussed. which even can not be perceived by human eyes. Hence, the 180
128 This paper presents a comprehensive survey of the VLC- data can be encoded in the emitting light to realize commu- 181
129 based indoor positioning technology and is organized as nication. It makes LEDs be used for both illumination and 182
130 follows. We explain various indoor positioning technologies communication. Using high-speed response characteristics of 183
131 and introduce VLC in Section II. In Section III, we make a White-LED devices, at the same time of lighting, LEDs can 184
132 comparison among the correlated survey papers and conclude send information. 185
133 our contributions. The VLC-based positioning technology and Compared with other existing technologies of wireless com- 186
134 classification are shown in Section IV. We provide a qualita- munications, VLC is an essential communication tool and has 187
135 tive comparison of these algorithms in Section V. Moreover, the following advantages. 188
LUO et al.: IPSs BASED ON VLC: STATE OF THE ART 3
TABLE I
S UMMARY OF VARIOUS W IRELESS T ECHNOLOGIES
TABLE II
C OMPARISON OF S URVEY PAPERS FOR IPS S
of
ro
EP
189 • Cost Efficiency: VLC uses the existed lighting infrastruc- VLC can also be used in outdoor and underwater environ- 218
190 ture to communicate, which means such systems can be ments. In this article, we mainly describe a critical application 219
191 deployed with few additional devices. In other words, the of VLC, namely IPS. 220
195 of VLC are thousands of times larger than the RF portion IPS is undergoing a rapid development in recent years. 222
196 of the spectrum [25]. Considering the increasing crowded A lot of surveys of IPS can be found [27]–[30]. And those 223
197 RF spectrum, many types of research endeavor to use surveys can help general readers quickly find relevant infor- 224
IEE
198 efficient frequency and spatial reuse. However, the current mation and have an obvious contrast with the VLC-based-IPS. 225
199 RF spectrum is still too scarce to meet the increasing Thus we gather some latest surveys of IPS and do a com- 226
200 wireless access demand. VLC with the broad bandwidth parison in Table II. Moreover, information of VLC is very 227
201 is suitable to design high-capacity mobile data networks important in the design of the VLC-based-IPS. The suitable 228
202 instead of RF-based communication systems. technology of VLC can efficiently improve the performance 229
203 • Energy Efficiency: LEDs use at least 75% less energy and of the VLC-based-IPS. Thus we still collect some surveys of 230
204 last 25 times longer than incandescent lighting [26]. Most VLC [31]–[34] and do a detailed comparison in Table III. 231
205 of the energy of LEDs is used for illumination. Therefore Recently, the VLC-based-IPS get a rapid development, and 232
206 the VLC system is energy efficient. numerous algorithms have been proposed. However, surveys 233
207 • Communication Security: Visible light can not penetrate of the VLC-based-IPS are very few. Here we list four survey 234
208 most objects and walls, and links can be kept confiden- papers [19]–[22] of the VLC-based-IPS and make the compari- 235
209 tial, which means an internal wireless communication son in Table IV. With the development of the LED technology, 236
210 security. the VLC-based-IPS can achieve better performance with the 237
211 Ergul et al. [23] have concluded the potential application high accuracy and low cost. Thus, we believe that a detailed 238
212 areas of VLC. Initially, as an indoor communication technol- survey of the VLC-based-IPS and related knowledge can help 239
213 ogy, VLC can be used at homes to provide Internet access to improve the development of IPS. Compared with other sur- 240
214 residents and realize smart homes. It can be used in shopping vey papers, this paper contains the latest papers and numerous 241
215 malls or museums to attain the wanted information quickly. references. Relevant knowledge of the VLC-based-IPS is rel- 242
216 Because of the security of VLC, it is also probably to be atively comprehensive. The new taxonomy is proposed. Our 243
217 employed in military applications. The authors outline that contributions include the following. 244
4 IEEE COMMUNICATIONS SURVEYS & TUTORIALS
TABLE III
C OMPARISON OF S URVEY PAPERS FOR VLC
of
TABLE IV
C OMPARISON OF S URVEY PAPERS FOR THE VLC-BASED -IPS S
ro
EP
IEE
245 • We summarize a lot of related knowledge of the • We compare the survey papers of the VLC-based-IPS and 264
246 VLC-based-IPS which is critical in the design of IPS. list the different taxonomy method of the survey papers 265
247 According to the specific application scenarios, choos- to help the reader quickly find the information of the 266
248 ing the right LED technology, modulation method and VLC-based-IPS. 267
262 give a direction to find out the way to improve the Compared with other systems, the VLC-based-IPS has 278
263 performance of the system. many advantages. According to the characteristic of RF, we 279
LUO et al.: IPSs BASED ON VLC: STATE OF THE ART 5
TABLE V
C OMPARISON B ETWEEN THE VLC-BASED -IPS, THE RF-BASED -IPS AND THE D EVICE -F REE -IPS
280 can see RF-based-IPS is not suitable to be used in the RF White-LEDs can be modulated at frequencies up to 20MHz, 330
281 sensitive area. The bandwidth of the RF-based-IPS is lim- which means a new opportunity to create a positioning sys- 331
282 ited and too scarce to meet the increasing wireless access tem. We can use the particular technology and manufacturing 332
283 demand. The device-free-IPS analyzes how a human body method to change the emission power and bandwidth of LEDs. 333
284 changes the pattern of received signal strength to detect and Under white light, the colors of the object are likely to be 334
of
285 locate human in an indoor environment. It can utilize the the same as the color under the natural light. That is why 335
286 Infrared, UWB Radar, Ultrasonic, Wi-Fi, Wireless Sensors, the white light is the most common light used in our daily 336
287 RFID, and FM to detect and track the human body without life. Conventionally, there are two ways to produce the white 337
288 wearing any device. Compared with device-based-IPS which light [31]: 338
289 need a particular smartphone-based or tag-based device, the • Blue-LED with phosphor: A yellow phosphor is coated 339
290 device-free-IPS is more vulnerable to environmental interfer- on a Blue-LED. The yellow phosphor can be excited by 340
291 ence and dynamics. Because the device-free-IPS relies on the the blue light and emit yellow light. Therefore combining 341
ro
292 human impact on wireless signals to detect and localize peo- the blue light and the yellow light can produce white 342
293 ple, environmental changes may also require recalibration for light. 343
294 the device-free-IPS to maintain satisfactory performance. The • RGB combination: Three LEDs produce red, green, yel- 344
295 detailed comparison of the VLC-based-IPS, the RF-based-IPS, low light respectively. Then mixing the three kinds of 345
296 and the device-free-IPS is shown in Table V. lights at a decent percentage can get white light. 346
297 Many the VLC-based-IPSs with high precision have been The first method is more commonly used because of the easier 347
298 proposed. Those systems in [35]–[39] reported the position- implementation and lower cost. But using the second method 348
299 ing accuracy up to a few centimeters. However, the accuracy can use Color Shift Keying (CSK) to modulate the data, which 349
EP
300 is attained only by computer simulations. In the practical means better performance in communication. Now there is 350
301 environment, the multipath effects will severely reduce the a growing interest in using White-LEDs to transmit high- 351
302 accuracy. Recent years, other practical VLC-based positioning speed data [43] and for localization [13]–[16], [18], [44], [45]. 352
303 systems are proposed: Eplison [18], Luxapose [40], LIPS [41], In [43], a 550 Mbit/s real-time VLC-based positioning sys- 353
304 PIXEL [42] and so on. Eplison presents the first practical tem of White-LEDs using Non-Return-to-Zero On-Off Keying 354
305 visible light localization system which 90% of the accuracy (NRZ-OOK) modulation is presented. By employing the pro- 355
306 can achieve 0.4m, 0.7m, and 0.8m for three different envi- posed analog pre-emphasis circuit based on NPN transistors 356
307 ronments. Luxapose achieves a localization accuracy of 0.1m. and active post-equalization circuit based on an amplifier, 357
308 Experimental results show the average positioning accuracy of the 3dB bandwidth of VLC link could be extended from 358
309 LIPS is about 0.4m. It is great stability against interference. 3 to 233MHz with the blue-filter. However, the modulation 359
IEE
310 PIXEL is a system which wearable can afford. The software bandwidth of White-LEDs is still limited, which restricts the 360
311 program of PIXEL is designed for both Android phones and modulation rate. 361
312 Google Glass. The main principle of these systems will be RGB-LEDs have relatively higher modulation bandwidth, 362
313 presented later. When VLC is used to realize localization, data rates, and lower response time. Compared with 40Mbps 363
314 we must consider issues such as LED technology, modulation data rates of White-LEDs, RGB-LEDs can reach up to 364
315 method, and types of receivers. We will discuss these factors 100Mbps [23]. In [46], CSK is suggested to be used in RGB- 365
316 subsequently. LEDs to meet the high data rate application. This type of 366
317 1) LED Technology: With the advantages of lower power modulation scheme can transmit the data by modulating red, 367
318 consumption and longer lifetime, LEDs are expected to be the green and blue signal separately. Some researchers propose 368
319 next generation illumination device. Because of this, VLC- a method that only modulates a single color of RGB-LEDs, 369
320 based indoor positioning has attracted a lot of attention. As while other colors remain to provide white color for illumi- 370
321 illumination is the primary purpose of the LED luminaire, nation. Employing RGB-LED, some outstanding achievements 371
322 it should not be affected by the communication. How to of efficient modulation are concluded in [47]. Wang et al. [47] 372
323 design the LED luminaire is a crucial part of the VLC-based employ Carrier-less Amplitude and Phase (CAP) modulation 373
324 positioning. The LED luminaire is used for communica- and Recursive Least Square (LS) based adaptive equalization 374
325 tion, and the driver circuit is modified to modulate the data. in the high-speed RGB-LED VLC system and achieve a very 375
326 Both White-LEDs and RGB-LEDs can be utilized to trans- high data rate of 4.5Gb/s at an average indoor distance. The 376
327 mit information in the VLC-based-IPS. Next, we will list the RGB-LEDs employed to transmit data can provide higher data 377
328 strength and weakness of the White-LEDs and RGB-LEDs rate over White-LEDs at the cost of greater complexity. The 378
329 respectively. comparison of the two technologies is presented in Table VI. 379
6 IEEE COMMUNICATIONS SURVEYS & TUTORIALS
380 Some researchers have proposed the VLC-based-IPS using the which transmit parallel data streams simultaneously [31]. 438
381 RGB-LEDs [36], [48]. However, it is sensitive to the frequency deviation. OFDM 439
382 2) Modulation Method: We have mentioned that LEDs needs some modifications because the conventional OFDM 440
383 should be taken into account for both illumination and com- method can use the complex and bipolar signals which do 441
384 munication purposes. Hence, it is important that a proper not directly suit VLC whose signals are unipolar and real. 442
385 modulation method is used to modulate the signal to switch We can use the Hermitian symmetry to generate real-time sig- 443
386 LEDs at desired frequencies which contain the information. It nals for implementation at the cost of doubling the required 444
387 means that the modulation method for VLC should achieve not bandwidth. Some articles have presented some schemes 445
388 only the data rate but also to meet the demand of illumination. such as Asymmetrically Clipped Optical OFDM (ACO- 446
389 In other words, there are two questions to be considered: dim- OFDM) [56], DC-biased Optical OFDM (DCO-OFDM) [57] 447
390 ming control and flicker control. Sometimes the illumination and Asymmetrically clipped DC-biased Optical OFDM (ADO- 448
391 is low in a room, while we still need to achieve the data rate OFDM) [58]. These schemes are described in detail in above 449
of
393 harmful to our eyes when we stay long in the environment of Color Shift Keying (CSK) is designed for VLC to solve the 451
394 fluctuations of the light intensity, which means the modulation problem of lower data rate and support the dimming control. 452
395 method should support the flicker control. CSK uses the intensity of the red, green, and blue colors in the 453
396 As an alternative of RF, VLC has some difference from it. LED source to modulate the signal. This type of the scheme 454
397 Hence, the phase or amplitude modulation method is not suit- relies on the color space chromaticity diagram. Recently CSK 455
398 able for VLC. In VLC, the modulation method encodes the has attracted significant attention, and some research results 456
399 information in the intensity of the light wave. There are lots of have been presented in [59]–[61]. 457
ro
400 modulation methods proposed for VLC. Various modulation Carrier-less Amplitude and Phase (CAP) is efficient to 458
401 methods have been used in the VLC-based positioning. We increase the VLC link capacity under limited bandwidth and 459
402 first introduce some types of modulation methods for VLC, high spectral efficiency. The CAP scheme is composed of 460
403 and then we will present some articles which employ the two orthogonal signals but has no need for overhead and car- 461
404 modulation method in VLC-based positioning. rier [62]. Commonly, CAP is much simpler than OFDM and 462
405 On-Off Keying (OOK) is a type of the simplest technique can be implemented without FFT and IFFT blocks. However, it 463
406 modulating data through turning the LED on and off to trans- is less flexible than OFDM. In [63]–[65], some characteristics 464
407 mit data bits 1 and 0. It is easy to be implemented. Therefore are studied. 465
EP
408 it is widely used in wireless communications. However, this We compare these modulation methods in spectral effi- 466
409 modulation method neither supports dimming control nor pro- ciency, system complexity, power efficiency, data rate, dim- 467
410 vides a high data rate. In NRZ-OOK, the positive and negative ming control and flicker effect in Table VII. 468
411 signs are used to express the 0s and 1s respectively. It also can 3) Types of Receivers: As we know, a receiver is an impor- 469
412 carry more information. In [43], based on the NRZ-OOK mod- tant part of VLC-based positioning because a large amount 470
413 ulation method, authors design optimal analog pre-equalizer of information needs to be obtained by it, such as IDs, the 471
414 and post-equalizer and employ them in the white phospho- base station position, signal strength, RRS, TOA, or geometric 472
415 rescent light LED for practical high-speed, low complexity information and so on. According to the positioning algorithm, 473
416 application, and finally, realize a 550Mbit/s real-time VLC different information is required. Otherwise, various types of 474
417 link. receivers can affect the data rate which is important in VLC. 475
IEE
418 Pulse Position Modulation (PPM) is a type of mod- We divide the receivers into two classes: Photodiode (PD) and 476
419 ulation method based on pulse position. The pulse posi- image sensors. Both of them receive the modulated signals 477
420 tion identifies the transmitted symbol. The symbol duration and decode the data. Next, we will present the strength and 478
421 is divided into t slots of equal duration, and a pulse is weakness of the two classes. 479
422 transmitted in one of the slots. The position of the pulse PD: Using the property of PD that it is highly sensitive to 480
423 identifies the transmitted symbol [31]. The PPM is sim- the light, many PD-based VLC positioning systems are pro- 481
424 pler but with lower spectral efficiency and a lower data posed. This type of technique is easy to be implemented and 482
425 rate. Therefore lots of other PPMs have been proposed can achieve a relatively high data rate. The response time is 483
426 such as Variable PPM (VPPM) [49], Overlapping PPM a lack, which can contribute to measuring TOA of the signal 484
427 (OPPM) [50], Multi-Pulse PPM (MPPM) [51], Overlapping at high accuracy. In other words, the PD-based VLC position- 485
428 MPPM (OMPPM) [52], Expurgated PPM (EPPM) [53], ing system is the most likely to choose the method of RSS, 486
429 Multilevel EPPM (MEPPM) [54], and Differential PPM TOA, and TDOA algorithms. However, image sensors based 487
430 (DPPM) [55]. These modulation methods have their advan- approach can get better results when using the vision analysis 488
431 tages and weaknesses. When we introduce them to VLC-based algorithms. Otherwise, the multiplexing technique is needed 489
432 positioning, we should take these pros and cons into consid- in the PD-based positioning system because it cannot sepa- 490
433 eration and choose the most suitable one. rate light source. In [13] and [66], the PD-based-IPS has been 491
435 can be used in VLC to reduce the inter-symbol interfer- Image Sensors: Compared with Photodiodes (PDs), it is 493
436 ence and increase the utilization of bandwidth. In OFDM, less expensive. The camera is equipped with most hand-held 494
437 the channel is divided into multiple orthogonal subcarriers devices while PDs are not, which means utilizing PDs as the 495
LUO et al.: IPSs BASED ON VLC: STATE OF THE ART 7
TABLE VI
S UMMARY OF LED T ECHNOLOGIES
TABLE VII
S UMMARY OF M ODULATION M ETHODS
of
TABLE VIII
TAXONOMY OF M ATHEMATICAL M ETHODS
496
497
ro
receiver needs additional cost. Using image sensors, the multi-
plexing techniques are not necessary because the image sensor
optimization method concentrates on how to improve the
accuracy and track-ability after the positioning algorithm. We
527
528
EP
498 can spatially separate light sources. Moreover, the positioning take the filtering technique, spring model and normalizing 529
499 accuracy is not subject to the multipath effect. Therefore, the method as the positioning optimization method. Taxonomy of 530
500 image sensor techniques can be developed quickly. Some stud- VLC-based positioning systems is illustrated in Figure 1. 531
501 ies can be found in [39], [40], [67], and [68]. Huynh et al. [67]
502 use image sensors to capture LEDs and infer the position of an C. Mathematical Method 532
503 image sensor according to the relationship between the known
504 three-dimensional (3D) coordinates of LEDs in the real world In this section, we will present the mathematical method 533
505 and two-dimensional (2D) coordinates of LEDs in the image. which includes the proximity, triangulation, and fingerprint. 534
506 Utilizing the available data, Huynh et al. [67] can determine Some survey papers have concluded these positioning algo- 535
507 the position using different methods. rithms. Therefore, we simply describe the principle of these 536
509 With the development of wireless technology, research 1) Proximity: As the simplest positioning method, prox- 540
510 results have been applied into IPS with various wireless tech- imity can provide characteristic relative location information 541
511 nologies. Different VLC-based positioning systems have been based on the range of a known station or a near AP. In detail, 542
512 presented in these papers. To evaluate the state-of-the-art the if there is just one beacon detecting the mobile device, it will 543
513 VLC-based-IPSs, a taxonomy should be defined, and various be considered as being collocated with a mobile device. When 544
514 algorithms should be compared and analyzed within the groups more than one beacon can identify the mobile device, we can 545
515 with common characteristics. In this paper, the VLC-based- derive the position of the mobile device according to the bea- 546
516 IPSs are classified based on three key features: mathematical con whose signal is the strongest, because we think the signal 547
517 method, sensor-assisted method, and positioning optimization strength is positively related to the distance. From this, we 548
518 method. can also conclude that the number of beacons determines the 549
519 We put the general three categories: proximity, triangulation, positioning accuracy. However, the variance of proximity is 550
520 and fingerprint, into a class called the mathematical method. high such that sometimes it cannot meet the demand of local- 551
521 We add two more categories: sensor-assisted method and ization. In [69]–[74], some results that VLC-based positioning 552
522 positioning optimization method. The sensor-assisted method uses the proximity to realize the localization are presented. 553
523 mainly focuses on how the additional sensor can help the In [69], an indoor navigation system is proposed, which 554
524 VLC-based-IPSs to achieve better accuracy. In this approach, employs LED lights and a geomagnetic sensor. It can provide 555
525 we discuss four types of sensors: image sensor, accelerator, the accurate position and travel direction. The geomagnetic 556
526 light sensor and multiple optical receivers. The positioning sensor is used to obtain the route information. This system 557
8 IEEE COMMUNICATIONS SURVEYS & TUTORIALS
of
Fig. 1. Taxonomy of positioning technologies. ro
EP
558 can provide voice navigation and can be naturally divided In [74], a system that can achieve long range indoor posi- 588
559 into three steps. The accuracy of the scheme is approximately tioning is presented. The system is a hybrid system where the 589
560 1-2m. To optimize the cost and implementation, VLC-based five-hop ZigBee wireless network is employed to unite with 590
561 positioning using LED lamps as beacons is shown in [70]. The VLC to positioning. The advantages of the system are low 591
562 system consists of mobile phone and LED lights, where the power, high security, and extensive area. 592
563 lamps repeatedly send ID or coordinate, and then the mobile 2) Triangulation: Triangulation employs the geometric 593
564 phone can derive the exact position using the received data. properties of triangles to estimate the location of the mobile 594
565 Two types of beacons: passive beacon and active beacon are device. Through measuring the distance to the fixed known 595
566 proposed in this article. Error free range for communication measurement points, or the received signal angle, and using 596
567 is up to 4.5m. the geometric properties of triangles, we can derive the 597
IEE
568 A hybrid system with VLC and ad-hoc wireless network location. According to the different methods using the posi- 598
569 infrastructure can provide accurate and convenient position tion estimate, we divide the triangulation into three classes: 599
570 information in [71]. Two designs of non-carrier and 4MHz AOA, TOA, and RSS. We will introduce them in detail 600
571 carrier VLC-based hybrid positioning technology have been subsequently. 601
572 shown. For the initial design, the error-free communication a) AOA: AOA realizes locating according to the angle 602
573 range is about 0.33-0.403m which is suitable for the low between the base station and the mobile device. The principle 603
574 data rate scene. However, for the second one, it is about of positioning is that using the intersection of multiple direc- 604
575 0.0057-0.479m, which can be used in the high data rate and tions of the wireless signal estimate the location of the mobile 605
576 wide-range reception scene. device. The AOA method requires at least two base stations. If 606
577 A system that uses VLC and 6-axes sensor can provide we can attain two or more angles of incidence, it is possible 607
578 higher accuracy than the simple VLC-based positioning sys- to calculate the intersection of multiple lines using geomet- 608
579 tem [72]. A Switching Estimated Receiver Position (SERP) ric relationships while this intersection is the location of the 609
580 scheme can decrease the estimated error distance in the paper. mobile device, as shown in Figure 2. 610
581 Sertthin et al. [73] employ the rotation matrix and Support Supposing the angles of incidence between the azimuth line 611
582 Vector Machines (SVMs) to calculate precise Field of View of wireless signal and the base station, namely BS1 and BS2, 612
583 (FOV) limit, and the possible azimuth and tilt angulations are are denoted by θ1 and θ2 respectively. The coordinate of the 613
584 also defined. The paper simulates the model which uses the 6- mobile device is (x0 , y0 ), and the coordinate of the base station 614
585 axis sensor to support the azimuth and tilt information and then is (xi , yi ). Then they satisfy the following formula: 615
586 proves that this model can reduce at least 80% of computations
587 during the Geometric Optics (GO) calculation. tan θi = (yi − y0 )/(xi − x0 ), i = 1, 2 (1) 616
LUO et al.: IPSs BASED ON VLC: STATE OF THE ART 9
of
Fig. 3. Positioning based on TOA.
617 AOA is very susceptible to the external environment, and son experiment shows that the accuracy of the RSS method is 658
618 its positioning accuracy is lower than TOA. Otherwise, the four times higher than AOA which can reach 5cm. The authors 659
ro
619 positioning accuracy of AOA will decrease when the dis- also report that the three factors, which can affect the accu- 660
620 tance between the mobile device and base station increases. racy of AOA-based positioning, are systematic errors, angular 661
621 Meanwhile, AOA needs the additional hardware which precision, and geometry on position uncertainty. 662
622 increases the cost. b) TOA: In an actual environment, the distance between 663
623 Next, we will introduce the AOA method mentioned in some the mobile device and each base station is not the same. Hence, 664
624 papers. A high accuracy system using AOA is shown in [40], the distance between the mobile device and nearby base station 665
625 where the accuracy is about 0.1m. Based on AOA, a system is calculated according to the time which the signal uses to 666
626 which wearables can afford is presented in [42]. In this paper, arrive at the base station. In a 2D-plane, when the distance
EP
667
627 a polarization-based modulation is suggested, which is flicker- between an object and three non-collinear reference points is 668
628 free and can provide a low pulse rate. To handle the uneven known, the position of the target can be achieved. 669
629 sampling problem, an adaptive down sampling algorithm is Suppose that the propagation time between the mobile 670
630 employed. The authors consider the numerical optimization device and the ith base station is ti , and the propagation speed 671
631 of the positioning algorithm and this system can be used in is v. Then the distance si between the mobile device and 672
632 smartphone and Google Glass. The accuracy of the system is base station can be estimated as v ∗ ti . We suppose that the 673
633 about 0.25m in [75]. The light sensor and accelerometer are coordinates of the mobile device and base station are (x0 , y0 ) 674
634 used to estimate the position of the mobile device. and (xi , yi ) respectively. Then, we can derive the following 675
635 The number of PDs in a circular-PD-array which can affect formula: 676
637 fied [76]. Then the truncated-weighting method is employed to (xi − xo )2 + (yi − yo )2 = si , i = 1, 2, 3 . . . n (2) 677
640 than the average and Max method. Liu et al. [77] consider that three circles whose centers are different base stations and the 679
641 it is harder to determine the precise position information when radius is the distance si . In the ideal case, the three circles will 680
642 the straightforward information provides tracking for the blind intersect at one point, and this point is the spatial position of 681
643 user. Hence, the authors utilize the sensor angles to calculate the target point in Figure 3. 682
644 the user’s exact location. For testing the system in the real This method requires that transmitter and receiver have 683
645 experiment, the accuracy is as small as 10cm. accurate time synchronization. The positioning result is heav- 684
646 A LOSOWC system for tracking is proposed in [43]. In this ily dependent on the hardware which may significantly 685
647 article, the reference LED corresponds to the unique identi- increase the cost. 686
648 fier which is represented by a single tone. Then the receiver Because the TOA method needs strict time synchronization, 687
649 detects the number of the reference LEDs and computes their there is relatively small amount of research. In [16], a system 688
650 positions. based on the TOA method has been presented. It assumed 689
651 In [78], the AOA-based VLC positioning system whose that the synchronization between the transmitter and receiver 690
652 accuracy can reach 8cm in the simulation is shown. The is perfect. In this system, white-LEDs are used for illumina- 691
653 authors derive the expression of the Cramer-Rao bound (CRB) tion and signal transmission. The device (receiver) receives 692
654 in detail. The system [79] uses photoreceiver which consists the signals and estimates the distance of each LED. The range 693
655 of three PDs to facilitate differential photocurrent sensing of can be obtained by measuring the TOA signal from LED. The 694
656 the incident light AOA. AOA is not sensitive to the power and CRB of the TOA-based-IPS is discussed in this paper. 695
10 IEEE COMMUNICATIONS SURVEYS & TUTORIALS
the optical signal. To estimate the position of the device, the 735
As for the noise error, an average filter is used to mitigate it. 743
of
been conducted, while the accuracy is not clearly shown in 747
Fig. 4. Positioning based on TDOA.
this paper. 748
the distance between the mobile device and the base sta- 750
696 TDOA is a method to improve TOA, and this method only
tion using the attenuation property of emitted signal strength. 751
697 requires time synchronization between the base stations that
With increasing distance, the signal strength decreases which 752
698 participate in positioning without the need for time-critical
means that RSS can obtain the distance. Some researchers 753
699 synchronization between the base stations and target. TDOA
ro
try to utilize the signal attenuation-based methods to calcu- 754
700 method measures the propagation time difference between the
late the signal path-loss due to propagation. However, the 755
701 mobile device and base stations, then converts it into the
path-loss models may fail because of the severe multipath 756
702 distance by multiplying the speed of propagation.
fading. Of course, there are many improved methods to 757
703 Then we can attain the location of the mobile client using
achieve better positioning accuracy. Some systems presented 758
704 the hyperbolic positioning method. As Figure 4 shows, the
previously also use RSS to positioning such as the systems 759
705 distance difference between the base station BS1 and BS2
of [13], [15], [18], [35], and [66]. OFDM applied in VLC- 760
706 is s12 = v ∗ (t1 − t2 ) . It is sure that the mobile device
based positioning to mitigate the multipath effect is introduced 761
EP
707 must lie in the hyperbola whose focuses are the two base
in [66]. The performance of OFDM as a multi-carrier modula- 762
708 stations and the focal length is the distance difference s12 (see
tion scheme and IM/DD as a single-carrier modulation scheme 763
709 the solid line of Figure 4). Meanwhile, another hyperbola can
is compared, and the results show that the positioning accuracy 764
710 be drawn whose focuses are the BS1 and BS3, and the focal
of OFDM is much better. The effect of subcarrier numbers is 765
711 length is their distance difference s13 (see the dotted line of
discussed. 766
712 Figure 4). Then the intersection of the two hyperbolas is the
In [44], CRB is used to analyze the theoretical accuracy 767
713 location of the mobile device. We suppose that the coordi-
of the VLC-based-IPS. It shows that the limited accuracy of 768
714 nates of the mobile device and the base station are (x0 , y0 )
4.78cm can be obtained. The influences to CRB of differ- 769
715 and (x1 , y1 ) respectively. Then we can derive the following
ent parameters are discussed in detail. Otherwise, the results 770
716 formula
indicate that the positioning accuracy with the square LEDs 771
IEE
is lower than that of the triangle LEDs. The state estima- 772
717 s1i = (x1 − xo )2 + (y1 − yo )2
tion model is employed in the system of [45] for positioning. 773
2 User mobility and dynamic device orientation are taken into 774
718 − (xi − xo )2 + (yi − yo )2 , i = 2, 3 (3) account. The Kalman Filter (KF) recursive estimation is 775
employed in this paper. The results show that the location 776
719 Equation (3) represents two solutions corresponding to the two error is about 5cm. Gu et al. [84] propose a Gaussian Mixture 777
720 intersections of the hyperbola. When putting it into the appli- Sigma-point Particle Filter (GM-SPPF) in the RSS-based 778
721 cation, we take advantage of prior knowledge to distinguish system to reduce the positioning error. 779
722 which is the real-world coordinates. Compared with TOA, Gu et al. [85], [86] mainly concentrate on the impact of mul- 780
723 it is less affected by the environment. Therefore, a higher tipath reflections on the RSS-based system performance. The 781
724 positioning accuracy can be provided by TDOA. Combined Deterministic and Modified Monte Carlo method 782
725 We have previously introduced the system in [14] which (CDMMC) is employed to calculate the impulse response [85]. 783
726 provides a very high accuracy within 1cm. The positioning Hence, calibration methods are employed to reduce the posi- 784
727 algorithm used in this paper is the TDOA method. Using the tioning error that the accuracy of this system can reach 785
728 TDOA method, the synchronization between the LED panel 0.558m. In [86], a system using the OOK modulation method 786
729 and receiver is not necessary but requires the synchronization and the CDMMC method is suggested. The accuracy of this 787
730 between transmitters [80]. The system of [80] utilizes TDOA scheme can be 0.8064m. 788
731 of the pilot signals to estimate the position of the device. The The effect of the diffuse light reflection on the performance 789
732 average of the location error of 0.14m is proposed in [81]. of the RSSI-based system is discussed in [87]. This paper 790
733 In the paper, the authors use the Intensity Modulation/Direct shows that the diffuse reflection has a significant effect on 791
LUO et al.: IPSs BASED ON VLC: STATE OF THE ART 11
792 the positioning error so that the error can increase consider- the modulation scheme is OOK. Hence the system is easy to 850
793 ably. The ICE effect of the VLC-based-IPS using the RSS deploy, and the accuracy is about 5cm. 851
794 technology is presented in [88]. Later we will discuss ICI in 3) Fingerprint: Fingerprint means the feature or character- 852
795 detail. istic of signals that are location dependent. In most times, RSS 853
796 Jeong et al. [89] focus on the problem of the effect of the is considered as a type of fingerprint. It is assumed that in dif- 854
797 tilted receiver angle, and a compensated scheme is designed in ferent areas the features of signals are different. The fingerprint 855
798 this paper. It shows that the estimated location changes in the uses the difference to correspond to the particular position. 856
799 same direction with the receiver tilted. In this paper, the error There are two types of fingerprint: radio-map based on fin- 857
800 compensated scheme utilizes multiplying the weighting factor gerprint localization and map-free fingerprint localization. The 858
801 to correct the shape of the estimated position and shifting the map based fingerprint localization contains two stages: offline 859
802 estimated position to minimize the error using a gyroscope. stage and online stage (or run-time stage). In the offline stage, 860
803 The simulation and experiment show that the error is 0.7424cm the site survey is performed in an indoor environment and 861
804 and 1.6226cm respectively. a radio-map is generated. The location positioning technique 862
of
805 The Optical Orthogonal Codes (OOCs) are applied to dis- uses the currently observed signal strengths and radio-map to 863
806 tinguish different signals at the receiver in [90]. Then the estimate the position of the target. However, the diffraction 864
807 RSS technology is used to estimate the position of the and reflection can affect RSS, which leads to poor accuracy. 865
808 mobile device. Otherwise, the reusing/combining codeword is Map-free fingerprint methods as another fingerprint can reduce 866
809 employed in this system to expand the system and use the the complexity of mapping maintaining. Next, we will intro- 867
810 code resource efficiently. The simulation results show that duce some papers that use fingerprint technology to achieve 868
811 the system can achieve an excellent accuracy of 0.08m and positioning. 869
ro
812 the receiver structure is simple. In [96], LED lamps with a unique address transmit the sig- 870
813 In [91], the Dual-Tone Multi-Frequency (DTMF) technique nal. The signal is modulated using the NRZ-OOK modulation 871
814 is employed in the RSSI-based positioning system. As we pre- method. The mobile device receives the signal, then the coor- 872
815 sented before, LED has its ID encoded into DTMF signal in dinates of the device can be estimated by the correlation of 873
816 this system. The LED lamp, as a transmitter, sends the DTMF the received signal and predetermined addresses. Further, the 874
817 signal periodically. Then the receiver can utilize the received authors define a Combination of Correlation Values (CCV) 875
818 DTMF signal to calculate the channel gain of the different to obtain the position of the device. The accuracy is about 876
819 LED lamp. Finally, solving the Lambertian equation group 4.38cm. 877
EP
820 can obtain the position of the mobile phone. The simulation The iterative Maximum Likelihood (ML) approach is 878
821 results show that the positioning error is about 18mm. employed in the positioning algorithm in [97]. The ML solu- 879
822 The work in [92] shows a positioning algorithm which tion cannot be obtained in the closed form of the regular 880
823 contacts the received light intensity at the receiver with the model. However, the iterative method does not require the vari- 881
824 pre-calculated light intensity at a given position. The LED AP ance of the measurement noise distribution. The LS method 882
825 sends a location tracking request message, while the mobile can obtain the solution quickly but not always optimal. Of 883
826 device measures the received light intensity and sends the course, the computational complexity will increase because of 884
827 response message back to the LED AP for position estima- the iterative process. Some papers have shown that the accu- 885
828 tion. The process is called the message exchange between racy of the system using the iterative approach highly relies 886
829 the LED AP and a mobile receiver. The simulation can on the initial guess. Fortunately, the iterative ML method has 887
IEE
830 show the performance of the system, i.e., the accuracy is a complete initial guess as for the LS solution. The simula- 888
831 about 30mm. tion shows that a positioning error lower than 10cm can be 889
832 An accuracy of 0.5mm of IPS is shown in [93]. In this obtained when the SNR is 15dB. 890
833 system, the authors use visible light to recover the VLC chan- A self-localization method and high robustness system are 891
834 nel feature and estimate the position of the mobile device by shown in [98]. The system uses the Bayesian signal model 892
835 solving the Lambertian transmission equation group. The sys- and considers the inherent sparsity to detect the LED lamps, 893
836 tem can be considered as a cost-efficient and higher resolution which means little complexity. Because the system does not 894
837 system. The paper shows that increasing transmitting power need synchronization among the beacons, it allows LEDs to 895
838 improves noise tolerance and the receiver height cannot affect be fully autonomous. The authors use a suitable probabilistic 896
839 the performance of the algorithm. model to achieve positioning whose accuracy is about 0.81cm. 897
840 A Received Signal Strength Ratio (RSSR) is applied in the The system in [99] designs the transmitter and receiver to 898
841 system of [94]. As usual, LED lamps are used to transmit the enable them to utilize the visible light for localization. In the 899
842 information. Using the received signal contained in each time transmitter, before being sent to LEDs, the voltage or cur- 900
843 slot, RSSR is easy to be obtained. According to RSSR, we rent is modulated with the continuous waves. LEDs transmit 901
844 can derive the distance ratio and then at least three equations visible light where a sinusoidal wave modulates the intensity. 902
845 could be built. Otherwise, modulating frequencies of different LEDs are dif- 903
846 Zhang and Kavehrad [95] use the adopting framed slotted ferent. Then the receiver calculates the received power spectral 904
847 ALOHA protocol in the RSS-based positioning system and density, extracts the power of each modulating frequency as 905
848 state that the synchronization is not required between LEDs. the metric, and connects with the parameters of the fingerprint 906
849 In this system, four LEDs can transmit the own code, and map which is corrected in the offline stage. The simulation 907
12 IEEE COMMUNICATIONS SURVEYS & TUTORIALS
TABLE IX
S UMMARY OF M ATHEMATICAL M ETHODS
of
ro
EP
908 and experiment show that the accuracy of the system reaches The VLC-based positioning using fingerprint technology and 935
910 Biagi et al. [100] propose a solution of Localization, Access, we will introduce these articles in detail. 937
911 and Transmission (LAT). It can be explained as follows: i) Based on the mathematical method of positioning systems, 938
912 the indoor localization system uses a power measurement a lot of papers are contained and presented in detail. Further, 939
913 database, ii) MIMO architecture supports VLC and provides we summarize these articles that use the numerical method to 940
914 high data rate by PPM, iii) Time Division Multiple Access achieve the positioning in Table IX. We mainly compare the 941
IEE
915 (TDMA) is used in VLC connectivity links. The numerical system cost, complexity and the accuracy of these systems. 942
916 experiment has verified the performance. Otherwise, some individual points will be presented in the 943
917 The uniform deployment of infrared LED devices is also note column in Table IX. 944
921 namely the power samples of the transmitters and time sam- As an IPS, the ability to provide high accuracy is necessary. 946
922 ples for the impulse responses in the offline stage. In the online To achieve high precision, many sensors are used to assist 947
923 stage, compared with real-time impulse response peaks in the positioning, which has been proved to be efficient. The sen- 948
924 map, the position of the mobile device can be obtained. sor is embedded in the mobile device such as a smartphone. 949
925 An IPS based on White-LEDs which uses the corre- Of course, the additional sensor can be applied in positioning. 950
926 lation method is proposed in [102], and the accuracy is Lots of sensors can be used in VLC-based positioning. In this 951
927 about 0.651cm. In [103], to mitigate interference, Time- subsection, we just consider four types of conventional sen- 952
928 Division-Multiplexing (TDM) is employed. Otherwise, the sors: image positioning/image sensor, accelerator, light sensor 953
929 TDM scheme can selectively receive signal. As a result, the and multiple optical receivers. Then we will describe these 954
930 average error distance is 1.689cm. The system in [104] also articles which use the sensor-assisted method to obtain better 955
931 employs TDM to remove ICI. However, the error range is positioning performance. 956
932 1.5787cm which is lower than that of [103]. In [104], the 1) Image Sensor: Commonly, image positioning uses the 957
933 Extinction Ratio (ER) distribution is employed, and NRZ- image sensor or camera to capture images, and then detect 958
934 OOK modulates the location information of the transmitter. the presence of the luminaires in the image, decoding their 959
LUO et al.: IPSs BASED ON VLC: STATE OF THE ART 13
960 transmitted identifiers for and/or locations. The location and signal through the lenses and demodulate the position informa- 1018
961 orientation of the luminaires can be derived in [40]. In this tion of LEDs. Finally, utilizing the geometrical relations can 1019
962 positioning technology, the number of image sensors and how derive the unknown position. In this system, two image sensors 1020
963 to obtain the position of the mobile devices may be different. are used and installed in the same plane, and the lenses need 1021
964 In [40], Luxapose consists of the visible light beacon, identical properties. Therefore, the positioning error distance 1022
965 smartphone, and a cloud server. The smartphone is unmod- is about 10cm. 1023
966 ified, and LEDs are slightly modified. The system requires In [108], four single LEDs with different colors are used 1024
967 overhead lights with known positions and nearby beacons to as transmitters because of the installation easiness and lower 1025
968 have relative positions. The AOA positioning process contains cost. Different colors can identify the position information 1026
969 two steps: using the image sensor to capture a picture, and of four LEDs. Two PC cameras receive the signal of LEDs 1027
970 detecting at least three beacons with known 3-D coordinates. and demodulate the position information. From calculating the 1028
971 Assuming that the camera geometry is known and it can be geometric relationship of the images on the two image sensors, 1029
972 determined by the pixels of the beacons projected. Then uti- the position can be obtained. In this system, a vector estimation 1030
of
973 lizing the properties of similar triangles, we can derive the algorithm is employed to calculate the four quadratic equations 1031
974 position of the mobile device respect to the beacon’s coordi- derived from the geometric relationship. 1032
975 nate system. The experiments show that the accuracy is about Wang et al. [109] use the low-resolution photodiode array 1033
976 0.1m. and the front camera of the mobile phone to achieve position- 1034
977 Huynh et al. [67] use a single image sensor to estimate ing. By matching the spot information with visual information 1035
978 distance and use the intervals to build a set of equations to and position information, the position of the mobile device 1036
979 obtain the position of the mobile device. The image sensor can be estimated. Meanwhile, a collinear-equation model is 1037
ro
980 noise which is ignored in most of the research can reduce the derived in this paper. A detailed positioning process is pre- 1038
981 accuracy of the positioning. An image sensor noise model and sented, and the simulation results show that the accuracy is in 1039
982 noise reduction mechanism are proposed in [67]. The results the order of centimeters. 1040
983 show that the positioning error can almost keep stable at all In [110], the LED-aided and encoder-based navigation sys- 1041
984 levels of noise added to the image. The accelerometer sensor is tem is proposed, which can estimate the position and pose 1042
985 also employed in this system. As a result, mobile devices can of a vehicle moving in a 2D-plane. The authors use LEDs, 1043
986 have arbitrary orientation and the positioning error is about a high-speed camera, and two encoders to enable the naviga- 1044
987 10cm. tion. The high-speed camera and encoders are equipped with 1045
EP
988 Two image sensors used in positioning are shown in [68] the front wheels. The camera can capture the image which may 1046
989 where the two image sensors are placed in the same plane. contain LEDs to measure LED coordinates and recover ID of 1047
990 Through the lens, the two image sensors receive the signal LED. Encoders can gauge the rotation of the front wheels. 1048
991 from all LEDs and utilize the geometric relationship to cal- Finally, the authors use an Extended Kalman Filter (EKF) to 1049
992 culate the distance between LEDs and the unknown position. fuse measurements of the encoder and camera to obtain a state 1050
993 Then the authors employ the LS estimation and vector esti- estimate. 1051
994 mation method to determine the position respectively. A few 2) Accelerometer: As we know, the received power can be 1052
995 centimeters of accuracy can be reached. used to derive the distance between the receiver and LEDs. 1053
996 In [105], a system that can measure the position and direc- However, the received power can be affected by the dis- 1054
997 tion of the mobile device is proposed. It is pointed out tance, the irradiance, and incidence angles. In spite of the 1055
IEE
998 that the position can be measured outside in this paper. An fixed position of the receiver, the received power is differ- 1056
999 image sensor receives the signals and demodulates the coor- ent when the incident angle of the receiver is changed. To 1057
1000 dinate information, and then the fingerprint method is used determine the orientation of the receiver, the incident and 1058
1001 to positioning. The accuracy is about 7cm in the simulation the irradiation angles, the accelerometer which is embedded 1059
1002 environment. in most smartphones are employed in some papers. We will 1060
1003 A system using the image sensor and accelerometer sen- introduce several papers using the accelerometer for position- 1061
1004 sor to estimate the location of the mobile device is proposed ing and explain the goodness of the accelerometer in these 1062
1005 in [106]. The experimental results reveal that the positioning systems. 1063
1006 error is about 5cm. The positioning process can be described In [38], the accelerometer is used in this positioning system 1064
1007 in three steps: finding color LEDs, obtaining camera incli- to provide the measurements of the acceleration of a receiver, 1065
1008 nation constraint and estimating position. The image sensor and then these measurements can be used to determine the 1066
1009 captures images with two LEDs lights and inputs the world receiver’s orientation. The received signal power and the mea- 1067
1010 positions of LEDs to the position estimate program. Then the surements of the accelerometer are used to form a closed-form 1068
1011 position estimation can determine the location by finding the expression which can estimate the position of the receiver. 1069
1012 colors and utilizing the relation between light color and light Meanwhile, the accelerometer measurements can be utilized 1070
1013 position. Moreover, using the acceleration sensor can obtain to remove some assumptions that the angle of a transmitter’s 1071
1014 the camera inclination. irradiance equals the incidence angle at the receiver. Hence 1072
1015 In the system in [107], the positioning algorithm needs at the height of the receiver from the ground is not needed. 1073
1016 least three LEDs to transmit their 3D coordinate information. The experiments show that the accuracy is about 0.5m in the 1074
1017 Then the two image sensors with unknown position receive the simulation environment. 1075
14 IEEE COMMUNICATIONS SURVEYS & TUTORIALS
1076 Yasir et al. [38] continue to study IPS by using visible light, is proposed in this paper. This method needs at least three 1134
1077 accelerometer, and the results shown in [75]. The accuracy is properly oriented sensors to receive the signal strengths and 1135
1078 increased to about 0.25m in the experimental environment. the sensors’ orientation. Combined with the pre-defined RSS 1136
1079 Similarly, the height of the receiver from the ground and the model, the measured data provides sufficient spatial constraints 1137
1080 alignment of the receiver’s normal with LED’s normal are to estimate the position of the mobile phone. 1138
1081 not required. LEDs broadcast their known coordinates to the In [111], considering the low-power applications, 1139
1082 receiver. By using TDMA approaches, the receiver can esti- Rajagopal et al. show that how the tags can demodu- 1140
1083 mate the received power. The irradiance angles of transmitters late the same signals using a light sensor instead of a camera. 1141
1084 can be obtained from the measurements of the accelerometer. By employing the Binary Frequency Shift Keying (BFSK) 1142
1085 Then by using the triangulation technology, the distance can be modulation in the system, the data rate has been increased 1143
1086 derived. It can be used to estimate the position of the mobile substantially. A fixed differential photosensor is employed in 1144
1087 device. In [39], the accelerometer sensor and multiple optical IPS [112]. Incident light of the photosensor can generate the 1145
1088 receivers are employed and then the triangulation technology photocurrents which are functions of the incident angle of 1146
of
1089 is used to estimate the position of the mobile device. the light. It means that by utilizing differential photocurrents, 1147
1090 The accelerometer sensor is used in [67], which the differential photosensors can estimate the AOA of signals 1148
1091 allows the mobile device to have an arbitrary orientation. from two LEDs. Then the triangulation technology can be 1149
1092 Huynh et al. [67] state that the accelerometer allows low-noise used to determine the position of the mobile device. The 1150
1093 measurement of acceleration in three perpendicular axes, thus accuracy is about 4cm. 1151
1094 it can sense tilt, motion, shock, and vibration in a smartphone. 4) Multiple Optical Receivers: the VLC-based-IPS using 1152
1095 In this system, the accelerometer sensor is used to measure the RSS technology requires multiple transmitters or receivers. 1153
ro
1096 the tilt forward and backward on the x-axis and the tilt side We have presented numerous multiple transmitter systems 1154
1097 to side on the y-axis. Employing the measurements of the based on the RSS technology which needs to know the 1155
1098 accelerometer sensor, the positioning can succeed in determin- transmission distance between the receiver and at least three 1156
1099 ing the position of the mobile device at arbitrary orientations. transmitters. However, interference such as ICI technology 1157
1100 Similarly, in [106], both an image sensor and an accelerome- needs to be employed in this kind of system. Another choice 1158
1101 ter are used in the positioning system. The detailed calibration is that a single transmitter and multiple receivers are used 1159
1102 procedure of the position detector using the accelerometer is in IPS. However, additional problems need to be solved effi- 1160
1103 shown in this paper. Using accelerometer, the camera inclina- ciently. Next, we will introduce some papers to describe this 1161
EP
1104 tion constraint can be obtained. By utilizing the results, the kind of approach. 1162
1105 direction of the camera can be estimated. In [15], the system with the single transmitter and multiple 1163
1106 3) Light Sensor: It is found that the light sensor has a deter- receivers using the RSSI technology obtains the position of the 1164
1107 ministic sensitivity to both distance and incident angle of the mobile device. The obvious advantages of this kind of system 1165
1108 light signal. Therefore, it is suitable for using the light sensor are not only removing ICI but also extending the positioning 1166
1109 as the receiver of VLC-based positioning. The light sensor can range to the single Rx. The system is shown in Figure 5. The 1167
1110 retrieve the beacon information, measure RSS and then esti- radius of the device is 0.2m. When one receiver is selected as 1168
1111 mate the position of the mobile device. The light sensor can the reference assuming Rx1 , the relative positions of Rx2 and 1169
1112 be found on the mainstream mobile devices. Meanwhile, to Rx3 can be expressed as 1170
1125 positioning system is presented in [41]. Firstly, the authors The system of [37] uses a single transmitter and multiple 1178
1126 conduct an experiment to demonstrate that the light sensor titled optical receivers to estimate the position information of 1179
1127 can infer the distance and angular information from the light the mobile device. Because a single transmitter is used, ICI can 1180
1128 signal. The information received by the light sensor may mix be ignored. However, considering that the RSS difference is 1181
1129 the background light signals such as daylight and artificial induced by the incidence angle difference, the problem that the 1182
1130 lights. Hence, a low-end microcontroller is employed to switch system needs a minimum separation distance can be solved. 1183
1131 on and off at a specified frequency. Then FFT and inverse In this system, the 2D-positioning is based on AOA, and the 1184
1132 FFT are used to extract the signal strength at a particular 3D-positioning is using RSS. The positioning error is less 1185
1133 frequency. A Multi-Face Light Positioning (MFLP) algorithm than 6cm. 1186
LUO et al.: IPSs BASED ON VLC: STATE OF THE ART 15
TABLE X
S UMMARY OF S ENSOR -A SSISTED M ETHODS
of
receiving node with the known coordinate. To get better perfor- 1207
mance, the iterative mechanism is suggested to be employed. 1208
The simulation results show that the positioning error can be 1209
ro
less than 2mm, while the accuracy of 4cm can be achieved in 1210
the experimental environment. 1211
1187 Xie et al. [41] state that their system has the following
1188 advantages. For example, the receiver can be mobile, and its trackability. 1226
1189 heights can be unknown, and its normal does not require to be 1) Filtering Technique: To achieve better performance in 1227
1190 aligned with those of transmitters and so on. In this system, the positioning system, some filter techniques are employed. 1228
1191 at least three tilted optical receivers and one accelerometer are We can divide these into two classes: KF and Particle Filter 1229
1192 required. To receive different light intensities from the same (PF). Some papers are collected to show the advantages of the 1230
1193 LED, optical receivers should be tilted at different angles. filtering technique in IPS. 1231
1194 According to the received light intensities and the receivers’ a) KF: Lots of estimators are proposed to be used in 1232
1195 orientations, by employing the triangulation technology, the mobile positioning and tracking as a location estimators such 1233
1196 position of the mobile device can be obtained. The accuracy as KF and EKF. KF requires that the system is dynamic and the 1234
1197 of less than 6cm can be achieved when the average receiver measurement model is linear. However, EKF can linearize the 1235
1198 speed is 1.3m/s by experiments with four optical receivers. nonlinear model. Hence it is suitable for use in these nonlinear 1236
1199 In [113] a fixed single optical transmitter and tilted multiple systems. 1237
1200 optical receivers consist of the positioning system. Utilizing In [45], a state estimation model is used to achieve local- 1238
1201 the coordinates of LED and the received signal power dif- ization and mobile tracking. The KF recursive estimation 1239
1202 ference between PDs can obtain an accuracy of about 4cm. technology is employed in this system. In this system, the 1240
1203 Xu et al. [114] focus on the scenario where the multi-PDs ter- authors describe the process of operation of KF in four steps: 1241
1204 minal device is located. Due to the multiple PDs in a device initialization, prediction, measurement, and update. However, 1242
1205 where their relative positions are known, positioning is dif- considering the KF linear requirement, other methods are 1243
1206 ferent from these positioning scenarios which have the fixed presented in this paper: EKF and Unscented Transformation 1244
16 IEEE COMMUNICATIONS SURVEYS & TUTORIALS
1245 (UT) which propagate system state and covariance informa- weights to particles, continuous walking and achieving conver- 1303
1246 tion through nonlinear transformations rather than analytically gence. The detailed positioning process is shown in this paper, 1304
1247 linearizing the transformation. Experiments are conducted to and a realistic light strength model is proposed. Otherwise, 1305
1248 compare the two technologies of EKF and UT. the authors show how the system can coexist with Wi-Fi. 1306
1249 Before describing the principle of the paper [110], we focus Experiments in typical office and library environments confirm 1307
1250 on the EKF technology. In [110], both the motion and mea- the effectiveness and the robustness of Lightitude. 1308
1251 surement models are nonlinear, which means that KF is not In [119], both EKF and PF technologies are employed 1309
1252 suitable. However, EKF is employed to fuse the measure- in IPS which uses the AOA-RSS positioning algorithm. 1310
1253 ments of the encoder and camera to produce a state estimate. Compared with EKF linearizing the system model to adapt 1311
1254 Accurate initialization plays a significant role in the perfor- the nonlinear system, PF is inherently suitable for the non- 1312
1255 mance of EKF. Therefore this article mainly concentrates on linear system, which means that the linearize process can be 1313
1256 the initialization algorithm which is formulated as a Bayesian removed and implemented. When the number of particles is 1314
1257 estimation problem. Otherwise, various aspects of the initial- small, the positioning error of PF is larger than that of EKF. 1315
of
1258 ization and EKF aiding using LED features are discussed, and However, with an increasing number of particles, the posi- 1316
1259 the experiments are conducted to verify the performance. tioning error of PF is less than EKF. Of course, both the two 1317
1260 Using KF to detect the direction of a robot is presented technologies can improve the system accuracy. 1318
1261 in [115]. In this paper, the accurate positioning comes from Considering the non-Gaussian measurement noise and 1319
1262 two reasons that the dead reckoning and KF using the output of nonlinear measurement model in [120], PF technology is 1320
1263 a geomagnetic sensor and gyro sensor. The experiments show employed to obtain the position of the robot. The motion 1321
1264 the lowest error comes from KF compared to the geomagnetic model, initialization, and localization method are described 1322
ro
1265 sensor and gyro sensor. The maximum directional error of KF in detail by a mathematical formula. The experiment shows 1323
1266 is kept to less than 6 degrees. Meanwhile, the positioning error that a better trackability can be obtained by using PF. 1324
1267 is 10.5cm. Ganti et al. [121] study the tracking capability using KF and 1325
1268 The EKF technology is used in [116], and the detailed PF. In KF technology, the Gaussian distribution assumption of 1326
1269 mathematical model is shown in this article. The offline cal- the state variables is made. Hence, in the scene which does 1327
1270 ibration method is proposed, and state initialization method not follow the Gaussian distribution, KF may have poor per- 1328
1271 is analyzed. In [117], an integrated AOA-RSS localization formance. However, PF does not have the limitation. Hence, 1329
1272 method is used in positioning, and the EKF estimator is the experiment where a wild value is added shows that both 1330
EP
1273 employed to achieve better performance. The EKF estimator is KF and PF can estimate the deviation of the real path, while 1331
1274 a compensator to reduce the accumulative errors through con- PF is faster than KF in determining the aggregation of the 1332
1275 trolling the mobile robot moving on the specified path. The correct path. 1333
1276 simulation results prove that EKF can achieve a smaller posi- 2) Spring Model: A Localization Algorithm based on a 1334
1277 tioning error and reduce the accumulative errors of the mobile Spring Model (LASM) method is employed in the system 1335
1278 robot. so that the computational complexity may be reduced with 1336
1279 b) PF: The basic idea of the PF algorithm is adopted by the same accuracy [122]. The principle of the algorithm is 1337
1280 a set of independent random samples from the proposal dis- to simulate the dynamics of the physical spring system and 1338
1281 tribution called particles to represent a posteriori probability. then obtain the position of the node. In this algorithm, the 1339
1282 Then obtaining new observations is to update the posteriori sensor nodes are set as particles and connected with neighbor 1340
IEE
1283 probability. Particles pass through a certain way, and Bayes nodes by virtual springs. As we know, the spring will force the 1341
1284 rule updates the weight. The advantage of this scheme is the nodes to go back to the stable positions from the other random 1342
1285 adaptation to the nonlinear model of Gaussian noise without positions. Hence, the blind nodes can be located by calculating 1343
1286 the assumption of the linear and Gaussian hypothesis of the the related forces with the neighbor nodes. The LASM system 1344
1287 system. is simple to be implemented. In [123], Wi-Fi and Bluetooth 1345
1288 GM-SPPF technology which combines the Gaussian environment IPSs have used the spring model to collaborative 1346
1289 Mixture Model (GMM) and Sigma approach is employed for positioning. In this system, the position and distance informa- 1347
1290 higher accuracy [84]. GMM is a Probability Density Function tion are first estimated by Wi-Fi and Bluetooth sensor. Then 1348
1291 (PDF) composed of a weighted sum of Gaussian components the confidence scores of the estimated position and distance 1349
1292 which is used to represent a set of particles. Sigma-Point are measured. Finally, the modified spring model utilizes that 1350
1293 Approach (SPA) is applied to sample the posterior distribu- information to correct the positioning error. 1351
1294 tion of a set of points. The GM-SPPF approach can reduce In [124], an improved spring model is used for collaborative 1352
1295 the computational cost. Otherwise, the system in [84] uses VLC-based indoor positioning. The basic principle is similar 1353
1296 the OOK modulation scheme and trilateration technique to to that of the paper [123] where the coordinate position is 1354
1297 estimate the position. The experiments verify that employing first estimated. Then the spring model is employed to cor- 1355
1298 GM-SPPF can obtain a better performance and the accuracy rect positioning errors. A full scheme of the improved spring 1356
1299 is about 9cm. model in the VLC-based-IPS is shown in Figure 6. In this sys- 1357
1300 The system in [118] called Lightitude employs PF for its tem, no additional sensors are needed, except for two types of 1358
1301 localizing scheme, and then the scheme can be described nodes. One is the unknown nodes which may require position- 1359
1302 as the following four parts: initialization stage, attributing ing services, while the other is nodes used to assist positioning. 1360
LUO et al.: IPSs BASED ON VLC: STATE OF THE ART 17
TABLE XI
S UMMARY OF P OSITIONING O PTIMIZATION M ETHODS
of
the VLC-based-IPSs that we have discussed in this paper. 1389
The challenges and the corresponding countermeasures of the 1390
VLC-based-IPSs are analyzed. Moreover, we give a detailed 1391
description of lessons learned in this section. 1392
1370
simulation results. Note that because the testing environment 1410
1371 ters which can affect the received optical power. The authors is different, a significant difference exists in the measurement 1411
1372 pointed out that three techniques can be used to determine the range. We should not just care about statistical accuracy. 1412
1373 optimal value of n, namely mathematical analysis, calibration,
1374 and fixed n for transmitter diversity.
1375 The positioning optimization methods mainly focus on the B. Challenges 1413
1376 filtering technology, while other methods have attained little VLC-based positioning has also attracted much attention. 1414
1377 attention. However, it is a good idea to employ technology Many technical issues have been solved, and various algo- 1415
1378 from other subjects for better performance. Through the sim- rithms have been proposed to obtain better performance. 1416
1379 ple description of the spring model and normalizing method, However, there are still some challenges which need the 1417
1380 the researchers can extend the optimization method to other researchers to overcome with great efforts. 1418
1381 aspects. That is why we will take the positioning optimization Inter-Cell Interference (ICI) in the VLC-based-IPS means 1419
1382 method as a class alone. Then we still compare these papers LEDs in the same room can cause severe interference to each 1420
1383 in Table XI. other [31]. As we know, the smaller cell radius can provide 1421
the higher capacity, which means LEDs should be deployed 1422
1384 V. D ISCUSSIONS , C HALLENGES , C OUNTERMEASURES , densely. Each cell modulates different information to provide 1423
1385 AND L ESSONS L EARNED different VLC applications. If the adjacent lighting cells use 1424
1386 We have introduced and discussed various VLC-based-IPSs the same frequency band, ICI problem will be produced. Thus 1425
1387 in Section IV. In addition, a more comprehensive compari- in the VLC-based-IPS which has multiple transmitters (LEDs), 1426
1388 son will be presented in this section which contains all of ICI should be taken into account in the VLC-based positioning. 1427
18 IEEE COMMUNICATIONS SURVEYS & TUTORIALS
TABLE XII
S UMMARY AND C OMPARISON OF VCL-BASED -IPS S
of
ro
EP
IEE
1428 Multipath reflection is a major factor which may decrease Lack of fingerprint and effect of the orientation of 1447
1429 the indoor positioning accuracy in the complex environ- receiver: The visible light requires fingerprints to identify 1448
1430 ment. In a room, the corner area suffers severely from LEDs uniquely. The similar RLS may be captured at different 1449
1431 the reflections, while the impact of reflections decreases in positions, which means an ambiguity problem that the same 1450
1432 the edge areas. The authors have reported their research RLS may imply multiple possible locations. It leads to wrong 1451
1433 results in [86], [125], and [126]. Gu et al. [86] and estimated position of the person. As we mentioned before, 1452
1434 Kashani et al. [126] have analyzed multipath reflection, which receiving device’s orientation and altitude may affect RLS. 1453
1435 shows that the multipath reflection will decrease the position- Some papers need to know the height of the receiver from the 1454
1436 ing accuracy, and the edge and corner areas have the relatively ground, while others may fix the receiver at a special orienta- 1455
1437 higher positioning error than other areas. Gu et al. [125] tion. Assuming that the receiver is a smartphone, its orientation 1456
1438 use the Combined Deterministic and Modified Monte Carlo and altitude are not fixed during the process of human move- 1457
1439 (CDMMC) approach to estimate the channel impulse response ment. Therefore, this method will limit the applicability of the 1458
1440 considering multipath reflections. The results show that local- VLC-based-IPS. 1459
1441 ization error increases from 1.2272m to 2.0989m at the Long delay is one of the limitations of the positioning 1460
1442 different location of the room when the multipath reflections system. Nowadays, a lot of applications need the real-time 1461
1443 are taken into account. Note that the center of the room service, especially for the navigation system. If the system can- 1462
1444 where the multipath reflection is weak has the accuracy with not provide the real-time positioning information, the moving 1463
1445 0.0019m. From the data, we can see multipath reflection will users may get lost. For the VLC-based-IPS, the data collec- 1464
1446 severely decrease the accuracy of IPS. tion, running the positioning algorithm and other measures 1465
LUO et al.: IPSs BASED ON VLC: STATE OF THE ART 19
1466 to decrease the positioning error will cost some time. The presented in [130]. A multiple access scheme based on block 1522
1467 positioning system exists for the time delay. Hence, different encoding TDM verifies that it can shorten the delay time effi- 1523
1468 system models have the different delays. ciently. In the environment with multiple lamps, it may take a 1524
1469 LED to Internet connectivity will become a trend and long time to receive signals because every light needs a time 1525
1470 challenge when lots of LEDs are deployed indoors. It is nec- slot to transmit the data in the conventional TDM scheme. 1526
1471 essary to connect LEDs to the Internet to create a VLC-based In [131], nine LEDs compose one block to reduce the system 1527
1472 broadband access network [31]. However, as we mentioned delay and ICI. The simplified localization algorithm, design- 1528
1473 before, the interference which may reduce the data rate should ing new modulation method and suitable design of the system 1529
1474 be taken into account. The cost is an important factor which can decrease the delay. 1530
1475 may affect the use of the system. LED to Internet connectivity need to design a novel 1531
1476 Some other challenges such as ambient light noise, syn- technology providing high-speed Internet connectivity and 1532
1477 chronization, flickering, receiver design and energy efficiency, reducing the cost. The transmitted data can be dynamic. 1533
1478 uplink and RF augmentation, mobility and coverage need to Through an advanced computer or a microcontroller, the differ- 1534
of
1479 be addressed shortly. Solving these challenges is essential so ent messages can be sent to different locations to meet various 1535
1480 that VLC-based positioning systems can be deployed in prac- demands. It means the network can control the information of 1536
1481 tice. Researchers try to find new and efficient solutions to these the LED lamp. 1537
ro
1484
tors into account. When taking the environmental effect into 1540
1485 ICI mitigation is critical in the VLC-based-IPS. Some account, the system is more complex. Hence many additional 1541
1486 researchers have paid a lot of attention to the ICI reduction in measures should be employed to solve the problem. Here we 1542
1487 VLC [127]–[128]. In [127], the Filter Bank-based Multicarrier list some lessons learned. 1543
1488 (FBMC) is employed to mitigate ICI generated in OFDMA- The multiple transmitters and a single receiver in IPS or 1544
1489 based cell partitioning. A carrier allocation VLC system which single transmitter and multiple receivers in IPS should be con- 1545
1490 adopts different carriers among the system’s respective cells sidered. The VLC-based-IPS using RSS technology requires 1546
1491 is proposed [128]. Then Kim et al. [128] employ carrier dis- multiple transmitters or receivers. Thus how to choose the 1547
EP
1492 tribution technology to the VLC-based-IPS to mitigate ICI, two different systems and which kind of result will bring are 1548
1493 and the research results are shown in [13], [88], and [129]. important. In this first kind of system, the multipath reflection 1549
1494 The frequency band allocation can be used to mitigate ICI. interference such as ICI technology needs to be solved. Hence, 1550
1495 Another way is to design the VLC-based-IPS using multiple the detailed method to decrease the multipath reflection and 1551
1496 optical receivers. We have presented that the VLC-based-IPS ICI is presented in Section VI Countermeasures. As for the 1552
1497 using the RSS technology requires multiple transmitters or second type of IPS, we have collected this kind of system 1553
1498 receivers. Thus if we use a single transmitter and multiple in the class of sensor-assisted method. Researchers can study 1554
1499 optical receivers in IPS, the interference causing LEDs will these references to find some ideas. 1555
1500 disappear. However, the other problems caused by the multiple Choosing the suitable LED technology and modulation 1556
1501 optical receivers need to be solved. method can improve the system performance. The RGB-LEDs 1557
IEE
1502 Mitigating the effect of multipath reflection is necessary can provide higher data rate over White-LEDs at the cost of 1558
1503 for the VLC-based-IPS. A calibration approach is proposed complexity. Moreover, the suitable design of the modulation 1559
1504 in [85] to reduce the effect of the multipath reflections. An method can further improve the data rate of the system using 1560
1505 OFDM-the VLC-based-IPS is suggested to mitigate the mul- the RGB-LEDs. The efficient modulation method can also 1561
1506 tipath reflections and provides a high data rate transmission decrease the effect of the multipath reflection. We have listed 1562
1507 in [66] and [126]. Designing new calibration methods to a lot of modulation methods in Section IV. Researchers can 1563
1508 increase the positioning accuracy in the corner and edge areas choose or redesign the suitable modulation method accord- 1564
1509 may be a good way. Deploying LEDs at a suitable layout ing to the actual application scenarios. Besides, an excellent 1565
1510 design to increase the coverage area of LEDs can decrease new modulation method can avoid the flickering and improve 1566
1511 the average positioning error. the detection performance. Otherwise, a new multiplexing 1567
1512 Solution to lack of fingerprint and effect of the orienta- technique can shorten the positioning latency. 1568
1513 tion of the receiver: Hu et al. [118] utilize the user’s mobility Using some sensors reasonably can obtain good position- 1569
1514 to obtain a RLS set which is spatially related and combines it ing precision in the VLC-based-IPS. Using image sensors, the 1570
1515 with inertial sensor data into a PF for positioning. To address multiplexing techniques are not required and the positioning 1571
1516 the effect of the orientation, in [118], a light strength model accuracy is not affected by the multipath effect. Moreover, the 1572
1517 is employed to connect every receiving device’s status with system can place two image sensors and use the geometric 1573
1518 RLS. We have presented that the sensor-assisted method can relationship to determine the position. The accelerometer can 1574
1519 help to remove the limitation such as using the accelerometer. be used to determine the orientation of the receiver, the inci- 1575
1520 Decreasing delay is one of the important aspects of the dent, and the irradiation angles. Because the received power 1576
1521 positioning system. In the VLC-based-IPS, a solution is can be affected by the distance, the irradiance, and incidence 1577
20 IEEE COMMUNICATIONS SURVEYS & TUTORIALS
1578 angles, thus using the accelerometer to obtain the orientation [2] P. Connolly and D. Boone, “Indoor location in retail: Where is the 1633
1579 of the receiver can improve the estimated accuracy of the dis- money?” ABI Res., San Francisco, CA, USA, Tech. Rep. 1148, 1634
Mar. 2013. 1635
1580 tance. The light sensor has a deterministic sensitivity to both [3] Q. Xu, R. Zheng, W. Saad, and Z. Han, “Device fingerprinting in wire- 1636
1581 distance and incident angle of the light signal. Thus it can less networks: Challenges and opportunities,” IEEE Commun. Surveys 1637
1582 infer the distance and angular information from the light sig- Tuts., vol. 18, no. 1, pp. 94–104, 1st Quart., 2016. 1638
[4] D. Dardari, P. Closas, and P. M. Djurić, “Indoor tracking: Theory, 1639
1583 nal. For the low-power applications, the researchers can use a methods, and technologies,” IEEE Trans. Veh. Technol., vol. 64, no. 4, 1640
1584 light sensor instead of a camera. Moreover, the geomagnetic pp. 1263–1278, Apr. 2015. 1641
1585 sensor and gyro sensor may be helpful. Also, the researcher [5] Y. Gu, A. Lo, and I. Niemegeers, “A survey of indoor positioning 1642
systems for wireless personal networks,” IEEE Commun. Surveys Tuts., 1643
1586 can attempt to employ other sensors into the VLC-based-IPS. vol. 11, no. 1, pp. 13–32, 1st Quart., 2009. 1644
1587 Designing suitable calibration methods and using correct [6] D. Hahnel, W. Burgard, D. Fox, K. Fishkin, and M. Philipose, 1645
1588 optimization method (such as filtering technique) can reduce “Mapping and localization with RFID technology,” in Proc. IEEE Int. 1646
Conf. Robot. Autom. (ICRA), vol. 1. New Orleans, LA, USA, 2004, 1647
1589 the positioning error. An efficient calibration method can be pp. 1015–1020. 1648
1590 designed to adjust the positioning results especially in the cor- [7] B. S. Çiftler, A. Kadri, and İ. Güvenç, “Fundamental bounds on 1649
of
1591 ner position which the positioning error is often larger. And the RSS-based wireless localization in passive UHF RFID systems,” in 1650
Proc. IEEE Wireless Commun. Netw. Conf., New Orleans, LA, USA, 1651
1592 filtering technique is helpful to improve the positioning preci- Mar. 2015, pp. 1356–1361. 1652
1593 sion. A running average filter may be efficient to decrease the [8] S. Gezici et al., “Localization via ultra-wideband radios: A look at 1653
1594 noise error. positioning aspects for future sensor networks,” IEEE Signal Process. 1654
Mag., vol. 22, no. 4, pp. 70–84, Jul. 2005. 1655
1595 The new mathematical model is a good choice to real- [9] A. H. Sayed, A. Tarighat, and N. Khajehnouri, “Network-based wire- 1656
1596 ize positioning. The TOA-RSS[132], AOA-RSS-based algo- less location: Challenges faced in developing techniques for accurate 1657
1597 rithms [117], [119], [133], [134] and triangulation-fingerprint wireless location information,” IEEE Signal Process. Mag., vol. 22, 1658
no. 4, pp. 24–40, Jul. 2005. 1659
ro
1598 method [135] have been proposed. Comparing with the com- [10] I. Guvenc and C.-C. Chong, “A survey on TOA based wireless localiza- 1660
1599 mon mathematical methods(such as TOA, TDOA, AOA, RSS), tion and NLOS mitigation techniques,” IEEE Commun. Surveys Tuts., 1661
1600 those methods can combine the respective advantages of the vol. 11, no. 3, pp. 107–124, 3rd Quart., 2009. 1662
[11] K. Kaemarungsi, “Efficient design of indoor positioning systems based 1663
1601 common mathematical methods and provide better perfor- on location fingerprinting,” in Proc. IEEE Int. Conf. Wireless Netw. 1664
1602 mance. Moreover, the dead reckoning and the spring model Commun. Mobile Comput., vol. 1. 2005, pp. 181–186. 1665
1603 can be employed in the VLC-based-IPS. Even the range-free [12] A. S. Thyagaturu, A. Mercian, M. P. McGarry, M. Reisslein, and 1666
W. Kellerer, “Software defined optical networks (SDONs): A com- 1667
1604 positioning method can be combined with the range-based prehensive survey,” IEEE Commun. Surveys Tuts., vol. 18, no. 4, 1668
1605 positioning method to realize positioning. pp. 2738–2786, 4th Quart., 2016. 1669
EP
1606 Combining VLC with other wireless technologies such as [13] H.-S. Kim, D.-R. Kim, S.-H. Yang, Y.-H. Son, and S.-K. Han, “An 1670
indoor visible light communication positioning system using a RF 1671
1607 Wi-Fi, Bluetooth or ZigBee, may be a good choice to realize carrier allocation technique,” J. Lightw. Technol., vol. 31, no. 1, 1672
1608 hybrid positioning. And this kind of strategy may integrate the pp. 134–144, Jan. 1, 2013. 1673
1609 advantages of different wireless technologies. [14] S.-Y. Jung, S. Hann, and C.-S. Park, “TDOA-based optical wireless 1674
indoor localization using LED ceiling lamps,” IEEE Trans. Consum. 1675
Electron., vol. 57, no. 4, pp. 1592–1597, Nov. 2011. 1676
[15] S.-H. Yang, E.-M. Jung, and S.-K. Han, “Indoor location estimation 1677
1610 VI. C ONCLUSION based on LED visible light communication using multiple optical 1678
1611 With the large requests of LBS, the indoor positioning has receivers,” IEEE Commun. Lett., vol. 17, no. 9, pp. 1834–1837, 1679
Sep. 2013. 1680
1612 attracted an enormous amount of interest in both academic [16] T. Q. Wang, Y. A. Sekercioglu, A. Neild, and J. Armstrong, “Position 1681
1613 and industrial communities. Numerous research results have accuracy of time-of-arrival based ranging using visible light with appli- 1682
been reported. Based on the investigation and universality of cation in indoor localization systems,” J. Lightw. Technol., vol. 31, 1683
IEE
1614
no. 20, pp. 3302–3308, Oct. 15, 2013. 1684
1615 VLC, the indoor positioning as an important application of [17] M. Bilgi, A. Sevincer, M. Yuksel, and N. Pala, “Optical wireless 1685
1616 VLC has experienced a fast development. In this paper, we localization,” Wireless Netw., vol. 18, no. 2, pp. 215–226, 2012. 1686
1617 make an in-depth survey of the VLC-based-IPSs. VLC-based [18] L. Li, P. Hu, C. Peng, G. Shen, and F. Zhao, “Epsilon: A visible light 1687
based positioning system,” in Proc. 11th USENIX Symp. NSDI, Seattle, 1688
1618 indoor positioning technologies such as LED technology, mod- WA, USA, 2014, pp. 331–343. 1689
1619 ulation method, and types of receivers are compared. In this [19] W. Chunyue et al., “The research of indoor positioning based on visi- 1690
1620 paper, we propose a new taxonomy that two new classes, ble light communication,” China Commun., vol. 12, no. 8, pp. 85–92, 1691
Aug. 2015. 1692
1621 namely, sensor-assisted method and positioning optimization [20] T.-H. Do, M. Yoo, and Y. Nemirovsky, “An in-depth survey of visi- 1693
1622 method are addressed. It is convenient for researchers to find ble light communication based positioning systems,” Sensors (Basel), 1694
1623 a method to improve system performance, to capture the opti- vol. 16, no. 5, p. 678, 2016. 1695
[21] R. Mautz and S. Tilch, “Survey of optical indoor positioning systems,” 1696
1624 mized research direction and to understand the degree of in Proc. Int. Conf. Indoor Position. Indoor Navig., Guimarães, Portugal, 1697
1625 the studies in the field. Further challenges are shown in this 2011, pp. 1–7. 1698
1626 paper. Tables are listed to compare the positioning systems and [22] N. Ul Hassan, A. Naeem, M. A. Pasha, T. Jadoon, and C. Yuen, “Indoor 1699
positioning using visible LED lights: A survey,” ACM Comput. Surveys, 1700
1627 the performances of the systems proposed in the pioneering vol. 48, no. 2, pp. 1–32, Nov. 2015. 1701
1628 papers. [23] O. Ergul, E. Dinc, and O. B. Akan, “Communicate to illuminate: State- 1702
of-the-art and research challenges for visible light communications,” 1703
Phys. Commun., vol. 17, pp. 72–85, Dec. 2015. 1704
1629 R EFERENCES [24] “Energy savings forecast of solid-state lighting in general illu- 1705
mination applications,” United States Dept. Energy, Washington, 1706
1630 [1] S. He and S.-H. G. Chan, “Wi-Fi fingerprint-based indoor position- DC, USA, Tech. Rep. 68, Aug. 2014. [Online]. Available: 1707
1631 ing: Recent advances and comparisons,” IEEE Commun. Surveys Tuts., http://apps1.eere.energy.gov/buildings/publications/pdfs/ssl/energy 1708
1632 vol. 18, no. 1, pp. 466–490, 1st Quart., 2016. savingsforecast14.pdf 1709
LUO et al.: IPSs BASED ON VLC: STATE OF THE ART 21
1710 [25] T. Gilling, The STREAM TONE: The Future of Personal Computing. [48] A. Sewaiwar, P. P. Han, Y. H. Chung, and Y. Ho, “3-Gbit/s indoor 1786
1711 Leicester, U.K.: Troubador, 2015. visible light communications using optical diversity schemes,” IEEE 1787
1712 [26] LED Lighting. (Mar. 2016). [Online]. Available: http://energy.gov/ Photon. J., vol. 7, no. 6, pp. 1–9, Dec. 2015. 1788
1713 energysaver/articles/led-lighting [49] IEEE Standard for Local and Metropolitan Area Network—Part 1789
1714 [27] L. Zhu, A. Yang, D. Wu, and L. Liu, “Survey of indoor positioning 16: Air Interface for Fixed Broadband Wireless Access Systems— 1790
1715 technologies and systems,” in Proc. Int. Conf. Life Syst. Model. Simul. Amendment 2: Medium Access Control Modifications and Additional 1791
1716 Intell. Comput. Sustain. Energy Environ. (LSMS ICSEE), Shanghai, Physical Layer Specifications for 2-11 GHz, IEEE Standard 802.16a- 1792
1717 China, 2014, pp. 400–409. 2003, pp. 1–292, 2003. 1793
1718 [28] R. Harle, “A survey of indoor inertial positioning systems for pedes- [50] D.-S. Shiu and J. M. Kahn, “Differential pulse-position modulation 1794
1719 trians,” IEEE Commun. Surveys Tuts., vol. 15, no. 3, pp. 1281–1293, for power-efficient optical communication,” IEEE Trans. Commun., 1795
1720 3rd Quart., 2013. vol. 47, no. 8, pp. 1201–1210, Aug. 1999. 1796
1721 [29] L. Mainetti, L. Patrono, and I. Sergi, “A survey on indoor position- [51] M. Noshad and M. Brandt-Pearce. (Aug. 2013). Can Visible 1797
1722 ing systems,” in Proc. 22nd Int. Conf. Softw. Telecommun. Comput. Light Communications Provide Gb/s Service. [Online]. Available: 1798
1723 Netw. (SoftCOM), Split, Croatia, 2014, pp. 111–120. https://arxiv.org/pdf/1308.3217.pdf 1799
1724 [30] A. Yassin et al., “Recent advances in indoor localization: A survey [52] T. Ohtsuki, I. Sasase, and S. Mori, “Overlapping multi-pulse pulse 1800
1725 on theoretical approaches and applications,” IEEE Commun. Surveys position modulation in optical direct detection channel,” in Proc. Tech. 1801
1726 Tuts., vol. 19, no. 2, pp. 1327–1346, 2nd Quart., 2017. Program Conf. Rec. IEEE ICC, vol. 2. Geneva, Switzerland, May 1993, 1802
of
1727 [31] P. H. Pathak, X. Feng, P. Hu, and P. Mohapatra, “Visible light pp. 1123–1127. 1803
1728 communication, networking, and sensing: A survey, potential and chal- [53] M. Noshad and M. Brandt-Pearce, “Expurgated PPM using balanced 1804
1729 lenges,” IEEE Commun. Surveys Tuts., vol. 17, no. 4, pp. 2047–2077, incomplete block designs,” IEEE Commun. Lett., vol. 16, no. 7, 1805
1731 [32] C. Rohner, S. Raza, D. Puccinelli, and T. Voigt, “Security in visi- [54] M. Noshad and M. Brandt-Pearce, “Multilevel pulse-position mod- 1807
1732 ble light communication: Novel challenges and opportunities,” Sensors ulation based on balanced incomplete block designs,” in Proc. 1808
1733 Transducers J., vol. 192, no. 9, pp. 9–15, 2015. Glob. Commun. Conf. GLOBECOM, Anaheim, CA, USA, 2012, 1809
1734 [33] L. U. Khan, “Visible light communication: Applications, architecture, pp. 2930–2935. 1810
1735 standardization and research challenges,” Digit. Commun. Netw., vol. 3, [55] D. Zwillinger, “Differential PPM has a higher throughput than PPM 1811
no. 2, pp. 78–88, 2017. for the band-limited and average-power-limited optical channel,” IEEE 1812
ro
1736
1737 [34] C. Medina, M. Zambrano, and K. Navarro, “LED based visible light Trans. Inf. Theory, vol. IT-34, no. 5, pp. 1269–1273, Sep. 1988. 1813
1738 communication: Technology, applications and challenges—A survey,” [56] J. Armstrong and A. J. Lowery, “Power efficient optical OFDM,” 1814
1739 Int. J. Adv. Eng. Technol., vol. 8, no. 4, pp. 482–495, 2015. Electron. Lett., vol. 42, no. 6, pp. 370–372, Mar. 2006. 1815
[57] J. Armstrong and B. J. C. Schmidt, “Comparison of asymmetrically 1816
1740 [35] W. Zhang, M. I. S. Chowdhury, and M. Kavehrad, “Asynchronous
clipped optical OFDM and DC-biased optical OFDM in AWGN,” IEEE 1817
1741 indoor positioning system based on visible light communications,”
Commun. Lett., vol. 12, no. 5, pp. 343–345, May 2008. 1818
1742 J. Opt. Eng., vol. 53, no. 4, pp. 1–9, 2014.
[58] S. D. Dissanayake and J. Armstrong, “Comparison of ACO-OFDM, 1819
1743 [36] G. Cossu et al., “A visible light localization aided optical wireless
DCO-OFDM and ADO-OFDM in IM/DD systems,” J. Lightw. 1820
1744 system,” in Proc. IEEE GLOBECOM Workshops, Houston, TX, USA,
Technol., vol. 31, no. 7, pp. 1063–1072, Apr. 1, 2013. 1821
1745 2011, pp. 802–807.
[59] E. Monteiro and S. Hranilovic, “Constellation design for color-shift 1822
EP
1746 [37] S.-H. Yang, H.-S. Kim, Y.-H. Son, and S.-K. Han, “Three-dimensional
keying using interior point methods,” in Proc. IEEE GC Wkshps, 1823
1747 visible light indoor localization using AOA and RSS with multiple
Anaheim, CA, USA, 2012, pp. 1224–1228. 1824
1748 optical receivers,” J. Lightw. Technol., vol. 32, no. 14, pp. 2480–2485,
[60] E. Monteiro and S. Hranilovic, “Design and implementation of color- 1825
1749 Jul. 15, 2014.
shift keying for visible light communications,” J. Lightw. Technol., 1826
1750 [38] M. Yasir, S.-W. Ho, and B. N. Vellambi, “Indoor positioning system vol. 32, no. 10, pp. 2053–2060, May 15, 2014. 1827
1751 using visible light and accelerometer,” IEEE/OSA J. Lightw. Technol., [61] R. Singh, T. O’Farrell, and J. P. R. David, “An enhanced color shift 1828
1752 vol. 32, no. 19, pp. 3306–3316, Oct. 1, 2014. keying modulation scheme for high-speed wireless visible light com- 1829
1753 [39] M. Yasir, S.-W. Ho, and B. N. Vellambi, “Indoor position tracking munications,” J. Lightw. Technol., vol. 32, no. 14, pp. 2582–2592, 1830
1754 using multiple optical receivers,” J. Lightw. Technol., vol. 34, no. 4, Jul. 15, 2014. 1831
1755 pp. 1166–1176, Feb. 15, 2016. [62] D. Karunatilaka, F. Zafar, V. Kalavally, and R. Parthiban, “LED based 1832
1756 [40] Y.-S. Kuo, P. Pannuto, K.-J. Hsiao, and P. L. Dutta, “Indoor positioning indoor visible light communications: State of the art,” IEEE Commun. 1833
1757 with mobile phones and visible light,” in Proc. 20th Annu. Int. Conf. Surveys Tuts., vol. 17, no. 3, pp. 1649–1678, 3rd Quart., 2015. 1834
1758 Mobile Comput. Netw., Maui, HI, USA, 2014, pp. 447–458. [63] F. M. Wu et al., “Performance comparison of OFDM signal and 1835
1759 [41] B. Xie et al., “LIPS: A light intensity-based positioning system for cap signal over high capacity RGB-LED-based WDM visible light 1836
IEE
1760 indoor environments,” ACM Trans. Sensor Netw., vol. 12, no. 4, communication,” IEEE Photon. J., vol. 5, no. 4, p. 7901507, Aug. 2013. 1837
1761 Nov. 2016, Art. no. 28. [64] R. Rodes et al., “Carrierless amplitude phase modulation of VCSEL 1838
1762 [42] Z. Yang, Z. Wang, J. Zhang, C. Huang, and Q. Zhang, “Wearables can with 4 bit/s/Hz spectral efficiency for use in WDM-PON,” Opt. Exp., 1839
1763 afford: Light-weight indoor positioning with visible light,” in Proc. 13th vol. 19, no. 27, pp. 26551–26556, 2011. 1840
1764 Annu. Int. Conf. Mobile Syst. Appl. Services, Florence, Italy, 2015, [65] F. Sjoberg, “A VDSL tutorial,” Luleå Tekniska Univ., Luleå, 1841
1765 pp. 317–330. Sweden, Tech. Rep. 2000, Sep. 2016. [Online]. Available: 1842
1766 [43] H. L. Li, X. B. Chen, J. Q. Guo, and H. D. Chen, “A 550 Mbit/s http://epubl.ltu.se/1402-1528/2000/02/LTU-FR-0002-SE.pdf 1843
1767 real-time visible light communication system based on phosphorescent [66] M. Aminikashani, W. J. Gu, and M. Kavehrad. Indoor Location 1844
1768 white light LED for practical high-speed low-complexity application,” Estimation With Optical-Based OFDM Communications. Accessed on 1845
1769 Opt. Exp., vol. 22, no. 22, pp. 27203–27213, 2014. Jun. 2015. [Online]. Available: https://arxiv.org/pdf/1506.07571v1.pdf 1846
1770 [44] X. L. Zhang, J. Y. Duan, Y. G. Fu, and A. C. Shi, “Theoretical accu- [67] P. Huynh et al., “VLC-Based positioning system for an indoor envi- 1847
1771 racy analysis of indoor visible light communication positioning system ronment using an image sensor and an accelerometer sensor,” Sensors 1848
1772 based on received signal strength indicator,” J. Lightw. Technol., vol. 32, (Basel), vol. 16, no. 6, p. 783, 2016. 1849
1773 no. 21, pp. 4180–4186, Nov. 1, 2014. [68] M. S. Rahman, M. M. Haque, and K.-D. Kim, “Indoor positioning 1850
1774 [45] M. Rahaim, G. B. Prince, and T. D. C. Little, “State estimation by LED visible light communication and image sensors,” Int. J. Elect. 1851
1775 and motion tracking for spatially diverse VLC networks,” in Proc. Comput. Eng., vol. 1, no. 2, pp. 161–170, 2011. 1852
1776 IEEE Globecom Workshops (GC Wkshps), Anaheim, CA, USA, 2012, [69] M. Nakajima and S. Haruyama, “New indoor navigation system for 1853
1777 pp. 1249–1253. visually impaired people using visible light communication,” EURASIP 1854
1778 [46] J.-W. Shi, J.-K. Sheu, C.-H. Chen, G.-R. Lin, and W.-C. Lai, “High- J. Wireless Commun. Netw., vol. 2013, no. 1, pp. 524–529, 2013. 1855
1779 speed GaN-based green light-emitting diodes with partially N-doped [70] G. del Campo-Jimenez, J. M. Perandones, and F. J. Lopez-Hernandez, 1856
1780 active layers and current-confined apertures,” IEEE Electron Device “A VLC-based beacon location system for mobile applications,” in 1857
1781 Lett., vol. 29, no. 2, pp. 158–160, Feb. 2009. Proc. IEEE Int. Conf. Localization GNSS (ICL GNSS), Turin, Italy, 1858
1782 [47] Y. G. Wang, X. X. Huang, L. Tao, J. Y. Shi, and N. Chi, “4.5-Gb/s 2013, pp. 1–4. 1859
1783 RGB-LED based WDM visible light communication system employing [71] Y. U. Lee and M. Kavehrad, “Two hybrid positioning system design 1860
1784 CAP modulation and RLS based adaptive equalization,” Opt. Exp., techniques with lighting LEDs and ad-hoc wireless network,” IEEE 1861
1785 vol. 23, no. 10, pp. 13626–13633, 2015. Trans. Consum. Electron., vol. 58, no. 4, pp. 1176–1184, Nov. 2012. 1862
22 IEEE COMMUNICATIONS SURVEYS & TUTORIALS
1863 [72] C. Sertthin, E. Tsuji, M. Nakagawa, S. Kuwano, and K. Watanabe, [93] Z. Zhou, M. Kavehrad, and P. Deng, “Indoor positioning algorithm 1940
1864 “A switching estimated receiver position scheme for visible light based using light-emitting diode visible light communications,” Opt. Eng., 1941
1865 indoor positioning system,” in Proc. 4th Int. Symp. Wireless Pervasive vol. 51, no. 8, pp. 1–6, 2012. 1942
1866 Comput. (ISWPC), Melbourne, VIC, Australia, 2009, pp. 1–5. [94] S.-Y. Jung, C.-K. Choi, S.-H. Heo, S. R. Lee, and C.-S. Park, “Received 1943
1867 [73] C. Sertthin, T. Fujii, O. Takyu, Y. Umeda, and T. Ohtsuki, “On physical signal strength ratio based optical wireless indoor localization using 1944
1868 layer simulation model for 6-axis sensor assisted VLC based position- light emitting diodes for illumination,” in Proc. IEEE Int. Conf. 1945
1869 ing system,” in Proc. Glob. Telecommun. Conf., Houston, TX, USA, Consum. Electron. (ICCE), Las Vegas, NV, USA, 2013, pp. 63–64. 1946
1870 2011, pp. 1–5. [95] W. Zhang and M. Kavehrad, “A 2-D indoor localization system based 1947
1871 [74] Y. U. Lee and M. Kavehrad, “Long-range indoor hybrid localization on visible light LED,” in Proc. IEEE Photon. Soc. Summer Topical 1948
1872 system design with visible light communications and wireless net- Meeting Series, Seattle, WA, USA, 2012, pp. 80–81. 1949
1873 work,” in Proc. IEEE Photon. Soc. Summer Topical Meeting Series, [96] S.-Y. Jung, S. Hann, S. Park, and C.-S. Park, “Optical wireless indoor 1950
1874 Seattle, WA, USA, Jul. 2012, pp. 82–83. positioning system using light emitting diode ceiling lights,” Microw. 1951
1875 [75] M. Yasir, S.-W. Ho, and B. N. Vellambi, “Indoor localization Opt. Technol. Lett., vol. 54, no. 7, pp. 1622–1626, 2012. 1952
1876 using visible light and accelerometer,” in Proc. Glob. Commun. [97] J. Lim, “Ubiquitous 3D positioning systems by LED-based visible light 1953
1877 Conf. (GLOBECOM), Atlanta, GA, USA, Dec. 2013, pp. 3341–3346. communications,” IEEE Wireless Commun., vol. 22, no. 2, pp. 80–85, 1954
1878 [76] S. Lee and S.-Y. Jung, “Location awareness using angle-of-arrival Apr. 2015. 1955
1879 based circular-PD-array for visible light communication,” in Proc. 18th [98] G. Kail, P. Maechler, N. Preyss, and A. Burg, “Robust asynchronous 1956
of
1880 Asia–Pac. Conf. Commun. (APCC), Jeju-do, South Korea, 2012, indoor localization using LED lighting,” in Proc. IEEE Int. Conf. 1957
1881 pp. 480–485. Acoust. Speech Signal Process. (ICASSP), Florence, Italy, 2014, 1958
1882 [77] X. Liu, H. Makino, and Y. Maeda, “Basic study on indoor location pp. 1866–1870. 1959
1883 estimation using visible light communication platform,” in Proc. 30th [99] J. Vongkulbhisal, B. Chantaramolee, Y. Zhao, and W. S. Mohammed, 1960
1884 Annu. Int. Conf. IEEE Eng. Med. Biol. Soc., Vancouver, BC, Canada, “A fingerprinting-based indoor localization system using intensity mod- 1961
1885 2008, pp. 2377–2380. ulation of light emitting diodes,” Microw. Opt. Technol. Lett., vol. 54, 1962
1886 [78] X. Q. Sun, Y. G. Zou, J. Y. Duan, and A. C. Shi, “The positioning no. 5, pp. 1218–1227, 2012. 1963
1887 accuracy analysis of AOA-based indoor visible light communication [100] M. Biagi, A. M. Vegni, and T. D. C. Little, “LAT indoor MIMO- 1964
1888 system,” in Proc. Int. Conf. Optoelectron. Microelectron., Changchun, VLC—Localize, access and transmit,” in Proc. IEEE Int. Workshop 1965
1889 China, 2015, pp. 186–190. Opt. Wireless Commun. (IWOW), Pisa, Italy, 2012, pp. 1–3.
ro
1966
1890 [79] A. Arafa, S. Dalmiya, R. Klukas, and J. F. Holzman, “Angle-of-arrival [101] A. M. Vegni and M. Biagi, “An indoor localization algorithm in a 1967
1891 reception for optical wireless location technology,” Opt. Exp., vol. 23, small-cell LED-based lighting system,” in Proc. IEEE Int. Conf. Indoor 1968
1892 no. 6, pp. 7755–7766, 2015. Position. Indoor Navig. (IPIN), Sydney, NSW, Australia, 2012, pp. 1–7. 1969
1893 [80] T.-H. Do and M. Yoo, “TDOA-based indoor positioning using visible [102] S. Hann, J.-H. Kim, S.-Y. Jung, and C.-S. Park, “White LED ceiling 1970
1894 light,” Photon. Netw. Commun., vol. 27, no. 2, pp. 80–88, 2014. lights positioning systems for optical wireless indoor applications,” in 1971
1895 [81] Y. H. Choi, I. H. Park, Y. H. Kim, and J. Y. Kim, “Novel LBS Proc. 36th Eur. Conf. Exhib. Opt. Commun., Turin, Italy, Sep. 2010, 1972
1896 technique based on visible light communications,” in Proc. IEEE pp. 1–3. 1973
1897 Int. Conf. Consum. Electron. (ICCE), Las Vegas, NV, USA, 2012,
[103] S.-H. Yang, D.-R. Kim, H.-S. Kim, Y.-H. Son, and S.-K. Han, “Indoor 1974
1898 pp. 576–577.
positioning system based on visible light using location code,” in 1975
1899 [82] J. H. Y. Nah, R. Parthiban, and M. H. Jaward, “Visible light communi-
EP
Proc. 4th Int. Conf. Commun. Electron. (ICCE), Hue, Vietnam, 2012, 1976
1900 cations localization using TDOA-based coherent heterodyne detection,”
pp. 360–363. 1977
1901 in Proc. IEEE 4th Int. Conf. Photon. (ICP), 2013, pp. 247–249.
[104] S.-H. Yang, D.-R. Kim, H.-S. Kim, Y.-H. Son, and S.-K. Han, “Visible 1978
1902 [83] K. Panta and J. Armstrong, “Indoor localization using white LEDs,”
light based high accuracy indoor localization using the extinction ratio 1979
1903 Electron. Lett., vol. 48, no. 4, pp. 228–230, 2012.
distributions of light signals,” Microw. Opt. Technol. Lett., vol. 55, 1980
1904 [84] W. Gu, W. Zhang, J. Wang, M. A. Kashani, and M. Kavehrad, “Three
no. 6, pp. 1385–1389, 2013. 1981
1905 dimensional indoor positioning based on visible light with Gaussian
[105] M. Yoshino, S. Haruyama, and M. Nakagawa, “High-accuracy posi- 1982
1906 mixture sigma-point particle filter technique,” in Proc. Broadband
tioning system using visible LED lights and image sensor,” in Proc. 1983
1907 Access Commun. Technol., 2015, p. 7.
Radio Wireless Symp., Orlando, FL, USA, 2008, pp. 439–442. 1984
1908 [85] W. J. Gu, M. Aminikashani, P. Deng, and M. Kavehrad, “Impact of
1909 multipath reflections on the performance of indoor visible light posi- [106] T. Tanaka and S. Haruyama, “New position detection method using 1985
1910 tioning systems,” J. Lightw. Technol., vol. 34, no. 10, pp. 2578–2587, image sensor and visible light LEDs,” in Proc. 2nd Int. Conf. Mach. 1986
1911 May 15, 2016. Vis., Dubai, UAE, 2009, pp. 150–153. 1987
1912 [86] W. J. Gu, M. Aminikashani, and M. Kavehrad, “Indoor visible [107] M. S. Rahman, M. M. Haque, and K.-D. Kim, “High precision indoor 1988
1913 light positioning system with multipath reflection analysis,” in Proc. positioning using lighting LED and image sensor,” in Proc. 14th 1989
IEE
1914 IEEE Int. Conf. Consum. Electron., Las Vegas, NV, USA, 2016, Int. Conf. Comput. Inf. Technol. (ICCIT), Dhaka, Bangladesh, 2011, 1990
1916 [87] N. A. Mohammed and M. A. Elkarim, “Exploring the effect of diffuse [108] B. Y. Kim, J.-S. Cho, Y. Park, and K.-D. Kim, “Implementation of 1992
1917 reflection on indoor localization systems based on RSSI-VLC,” Opt. indoor positioning using LED and dual PC cameras,” in Proc. 4th Int. 1993
1918 Exp., vol. 23, no. 16, pp. 20297–20313, 2015. Conf. Ubiquitous Future Netw. (ICUFN), 2012, pp. 476–477. 1994
1919 [88] H.-S. Kim, D.-R. Kim, S.-H. Yang, Y.-H. Son, and S.-K. Han, [109] Y. Q. Wang, Y. K. Gong, and Z. F. Shi, “Research on the collinear equa- 1995
1920 “Inter-cell interference mitigation and indoor positioning system based tion model of visual positioning based on visible light communication,” 1996
1921 on carrier allocation visible light communication,” in Proc. 5th IEEE in Proc. MATEC Web Conf., vol. 22. 2015, Art. no. 02003. 1997
1922 Int. Conf. Signal Process. Commun. Syst. (ICSPCS), Honolulu, HI, [110] D. Zheng, R. Vanitsthian, G. Chen, and J. A. Farrell, “LED-based 1998
1923 USA, 2011, pp. 1–7. initialization and navigation,” in Proc. Amer. Control Conf. (ACC), 1999
1924 [89] E.-M. Jeong, S.-H. Yang, H.-S. Kim, and S.-K. Han, “Tilted receiver Washington, DC, USA, 2013, pp. 6199–6205. 2000
1925 angle error compensated indoor positioning system based on visible [111] N. Rajagopal, P. Lazik, and A. Rowe, “Visual light landmarks for 2001
1926 light communication,” Electron. Lett., vol. 49, no. 14, pp. 890–892, mobile devices,” in Proc. IPSN, Berlin, Germany, 2014, pp. 249–260. 2002
1927 Jul. 2013. [112] A. Arafa, X. Jin, and R. Klukas, “Wireless indoor optical positioning 2003
1928 [90] S. Yamaguchi, V. V. Mai, T. C. Thang, and A. T. Pham, “Design with a differential photosensor,” IEEE Photon. Technol. Lett., vol. 24, 2004
1929 and performance evaluation of VLC indoor positioning system using no. 12, pp. 1027–1029, Jun. 15, 2012. 2005
1930 optical orthogonal codes,” in Proc. IEEE 5th Int. Conf. Commun. [113] S.-H. Yang, E.-M. Jeong, and S.-K. Han, “Indoor positioning based on 2006
1931 Electron. (ICCE), Da Nang, Vietnam, 2014, pp. 54–59. received optical power difference by angle of arrival,” IEEE Electron. 2007
1932 [91] P. Luo et al., “An indoor visible light communication positioning Lett., vol. 50, no. 1, pp. 49–51, Jan. 2014. 2008
1933 system using dual-tone multi-frequency technique,” in Proc. 2nd Int. [114] W. Xu, J. Wang, H. Shen, H. Zhang, and X. H. You, “Indoor positioning 2009
1934 Workshop Opt. Wireless Commun. (IWOW), Newcastle upon Tyne, for multiphotodiode device using visible-light communications,” IEEE 2010
1935 U.K., 2013, pp. 25–29. Photon. J., vol. 8, no. 1, pp. 1–11, Feb. 2016. 2011
1936 [92] Y. Kim, J. Hwang, J. Lee, and M. Yoo, “Position estimation algo- [115] H. Nishikata et al. Basis Research of Indoor Positioning 2012
1937 rithm based on tracking of received light intensity for indoor visible Method Using Visible Light Communication and Dead 2013
1938 light communication systems,” in Proc. IEEE 3rd Int. Conf. Ubiquitous Reckoning. Accessed on Sep. 2011. [Online]. Available: 2014
1939 Future Netw. (ICUFN), Dalian, China, 2011, pp. 131–134. http://ipin2011.dsi.uminho.pt/PDFs/Poster/47_Poster.pdf 2015
LUO et al.: IPSs BASED ON VLC: STATE OF THE ART 23
2016 [116] D. Zheng, G. Chen, and J. A. Farrell, “Navigation using linear Junhai Luo received the B.S. degree in com- 2083
2017 photo detector arrays,” in Proc. IEEE Int. Conf. Control Appl. (CCA), puter science and appliance from the University 2084
2018 Hyderabad, India, 2013, pp. 533–538. of Electronic Science and Technology of China in 2085
2019 [117] N.-T. Nguyen, N.-H. Nguyen, V.-H. Nguyen, K. Sripimanwat, and 2003, the M.S. degree in computer appliance tech- 2086
2020 A. Suebsomran, “Improvement of the VLC localization method using nology from the Chengdu University of Technology, 2087
2021 the extended Kalman filter,” in Proc. IEEE Region 10 Conf. (TENCON), Chengdu, China, in 2006, and the Ph.D. degree in 2088
2022 Bangkok, Thailand, 2014, pp. 1–6. information and communication engineering from 2089
2023 [118] Y. Q. Hu et al. A Visible Light-Based Indoor Positioning the University of Electronic Science and Technology 2090
2024 System. Accessed on Jun. 2015. [Online]. Available: of China. He was a Visiting Scholar with McGill 2091
2025 https://arxiv.org/pdf/1505.05977v3.pdf University, Canada and the University of Tennessee, 2092
2026 [119] N.-T. Nguyen, A. Suebsomran, K. Sripimanwat, and N.-H. Nguyen, Knoxville, TN, USA. He was promoted to an 2093
2027 “Design and simulation of a novel indoor mobile robot localization Associate Professor in 2011. His research interests and papers are mostly in 2094
2028 method using a light-emitting diode positioning system,” Trans. Inst. the areas of wireless communication, target detection, and information fusion. 2095
2029 Meas. Control, vol. 38, no. 3, pp. 305–314, 2016. He is a member of the IEICE and the CCF. 2096
2030 [120] D. Zheng, K. Cui, B. Bai, G. Chen, and J. A. Farrell, “Indoor localiza-
2031 tion based on LEDs,” in Proc. IEEE Int. Conf. Control Appl. (CCA),
2032 Denver, CO, USA, 2011, pp. 573–578.
of
2033 [121] D. Ganti, W. Zhang, and M. Kavehrad, “VLC-based indoor positioning
2034 system with tracking capability using Kalman and particle filters,” in
2035 Proc. IEEE Int. Conf. Consum. Electron. (ICCE), Las Vegas, NV, USA,
2036 2014, pp. 476–477.
2037 [122] W. Chen et al., “Localization algorithm based on a spring
2038 model (LASM) for large scale wireless sensor networks,” Sensors,
2039 vol. 8, no. 3, pp. 1797–1818, 2008.
2040 [123] D. Taniuchi, X. Liu, D. Nakai, and T. Maekawa, “Spring model based
2041 collaborative indoor position estimation with neighbor mobile devices,”
2042 IEEE J. Sel. Topics Signal Process., vol. 9, no. 2, pp. 268–277,
ro
Liying Fan received the B.S. degree in electronic 2097
2043 Mar. 2015.
information engineering from Southwest University, 2098
2044 [124] Z. J. Luo, W. N. Zhang, and G. F. Zhou, “Improved spring model-based
in 2015. She is currently pursuing the M.S. degree 2099
2045 collaborative indoor visible light positioning,” Opt. Rev., vol. 23, no. 3,
in information and communication engineering with 2100
2046 pp. 479–486, 2016.
the University of Electronic Science and Technology 2101
2047 [125] W. J. Gu, M. A. Kashani, and M. Kavehrad. Multipath Reflections
of China. Her research interests and papers are 2102
2048 Analysis on Indoor Visible Light Positioning System. Accessed on
mostly in the areas of wireless communication and 2103
2049 Apr. 2015. [Online]. Available: https://arxiv.org/pdf/1504.01192v1.pdf
information fusion. 2104
2050 [126] M. A. Kashani, W. J. Gu, and M. Kavehrad. Indoor Positioning in High
2051 Speed OFDM Visible Light Communications. Accessed on May 2015.
2052 [Online]. Available: https://arxiv.org/pdf/1505.01811v1.pdf
EP
2053 [127] S. Y. Jung, D. H. Kwon, S. H. Yang, and S. K. Han, “Inter-cell
2054 interference mitigation in multi-cellular visible light communications,”
2055 Opt. Exp., vol. 24, no. 8, pp. 8512–8526, 2016.
2056 [128] H.-S. Kim, D.-R. Kim, S.-H. Yang, Y.-H. Son, and S.-K. Han,
2057 “Mitigation of inter-Cell interference utilizing carrier allocation in vis-
2058 ible light communication system,” IEEE Commun. Lett., vol. 16, no. 4,
2059 pp. 526–529, Apr. 2012.
2060 [129] H.-S. Kim, D.-H. Kwon, S.-H. Yang, Y.-H. Son, and S.-K. Han,
2061 “Channel assignment technique for RF frequency reuse in CA-VLC-
2062 based accurate optical indoor localization,” J. Lightw. Technol., vol. 32,
2063 no. 14, pp. 2544–2555, Jul. 15, 2014.
2064 [130] Y. N. Hou, S. L. Xiao, H. F. Zheng, and W. S. Hu, “Multiple
2065 access scheme based on block encoding time division multiplexing
2066 in an indoor positioning system using visible light,” IEEE/OSA J. Opt. Husheng Li received the B.S. and M.S. degrees
IEE
2105
2067 Commun. Netw., vol. 7, no. 5, pp. 489–495, May 2015. in electronic engineering from Tsinghua University, 2106
2068 [131] K. Yan, H. Y. Zhou, H. L. Xiao, and X. L. Zhang, “Current status of Beijing, China, in 1998 and 2000, respectively, 2107
2069 indoor positioning system based on visible light,” in Proc. 15th Int. and the Ph.D. degree in electrical engineering 2108
2070 Conf. Control Autom. Syst., Busan, South Korea, 2015, pp. 565–569. from Princeton University, Princeton, NJ, USA, 2109
2071 [132] M. F. Keskin and S. Gezici, “Comparative theoretical analysis of in 2005. From 2005 to 2007, he was a Senior 2110
2072 distance estimation in visible light positioning systems,” J. Lightw. Engineer with Qualcomm Inc., San Diego, CA, 2111
2073 Technol., vol. 34, no. 3, pp. 854–865, Feb. 1, 2016. USA. In 2007, he joined the EECS Department, 2112
2074 [133] A. Şahin, Y. S. Eroǧlu, İ. Güvenc, N. Pala, and M. Yüksel. Hybrid 3D University of Tennessee, Knoxville, TN, USA, as 2113
2075 Localization for Visible Light Communication Systems. Accessed on an Assistant Professor. He was promoted to an 2114
2076 Aug. 2015. [Online]. Available: https://arxiv.org/pdf/1508.05776v1.pdf Associate Professor in 2013. His research is mainly 2115
2077 [134] G. B. Prince and T. D. C. Little, “A two phase hybrid RSS/AoA algo- focused on statistical signal processing, wireless communications, networking, 2116
2078 rithm for indoor device localization using visible light,” in Proc. IEEE smart grid, and game theory. He was a recipient of the Best Paper Awards of 2117
2079 Glob. Commun. Conf., Anaheim, CA, USA, 2012, pp. 3347–3352. the EURASIP Journal of Wireless Communications and Networks 2005, the 2118
2080 [135] S. Feng, X. Li, R. Zhang, M. Jiang, and L. Hanzo, “Hybrid position- EURASIP Journal of Advances in Signal Processing 2015, IEEE ICC 2011, 2119
2081 ing aided amorphous-cell assisted user-centric visible light downlink IEEE SmartGridComm 2012, and the Best Demo Award of IEEE Globecom, 2120
2082 techniques,” IEEE Access, vol. 4, pp. 2705–2713, 2016. 2010. 2121