Вы находитесь на странице: 1из 6

IQ-WDM for IEEE 802.

11bb-based LiFi
Ardimas Andi Purwita and Harald Haas
School of Engineering, Institute for Digital Communications, LiFi R&D Centre, The University of Edinburgh
E-mail: {a.purwita, h.haas}@ed.ac.uk

Abstract— In July 2018, a new IEEE task group focusing predicted to handle more than half of the total internet protocol
on light communications was formed, namely IEEE 802.11bb (IP) traffic. In order to cater for this need, the IEEE 802.11
(TGbb). The primary motivation of this task group is to standard- working group formed a task group which focused on the
ize mobile, networked light communications, i.e., light fidelity
(LiFi). At the time of writing, discussions in TGbb still focus high frequency spectrum, e.g., WiGig which is also known
on the physical (PHY) layer, and recently a common mode PHY as the 60 GHz WiFi. Moreover, in July 2018, a task group
has been agreed. The common mode PHY is defined based on focusing on light communications (LC), namely the IEEE
the frequency upconversion of the signal outputs of existing 802.11 Light Communications Amendment - Task Group “bb”
WiFi chipsets. In addition, a DC bias is used to enable the (TGbb) was formed [11]. The vision of TGbb is to realize a
intensity-modulation and direct-detection (IM/DD) over a light
emitting diode (LED). In this paper, we compare it with another mobile and networked solution for OWC, i.e., light fidelity
method, namely in-phase and quadrature wavelength division (LiFi) [12]. This is also the main difference between the
multiplexing (IQ-WDM). IQ-WDM refers to a method where IEEE 802.11bb and the existing OWC standards. The scope of
both I and Q baseband signal from existing WiFi chipsets are TGbb is to specify a new physical (PHY) layer and propose
transmitted independently at different wavelengths. A compre- a few modifications to the IEEE 802.11 media access control
hensive comparison is done, and our error performance results
indicate that IQ-WDM significantly outperforms the frequency (MAC) layer. It is worth noting here that it has already been
upconversion method by 6 dB. We also show that IQ-WDM is emphasized at the beginning of the project to support the
robust againsts IQ imbalance. co-existence of WiFi and LiFi by means of adding the Fast
Index Terms—IEEE 802.11bb, LiFi, PHY, WDM. Session Transfer protocol in the MAC layer.
Regarding the PHY layer, there is a general consensus
I. I NTRODUCTION in TGbb to use orthogonal frequency-division multiplexing
The earliest known demonstration of optical wireless com- (OFDM). At the time of writing, the details of PHY are still
munications (OWC) is a photophone that was developed by being discussed. There are two proposals in TGbb. The first
Alexander Graham Bell in 1880 [1]. A global attempt to one is to use a DC bias and an RF mixer, i.e., frequency
standardize commercial OWC technologies was not made until up-and-down conversion with a center frequency [13]. The
the 1990s. Infrared data association (IrDA) was formed in second one is to reuse the PHY layer from G.vlc [8], which is
1993 to develop interoperable infrared (IR)-based OWC [2]. a DC-biased OFDM or asymmetrically clipped optical OFDM
This was then followed by the formation of IEEE 802.11 with adaptive bit loading. For ease of reference, the second
in 1997, which can achieve a maximum speed of 2 Mbps proposal will be referred to as the LC-optimized proposal.
[3]. This legacy standard also covers specifications of data The main merit of the first proposal is that it requires less
transmission over IR spectrum. In 1999, due to its limited effort to modify the existing WiFi silicon. Off-the-shelf WiFi
ability to provide high data rates and wide coverage, IR chipsets can be used in conjunction with a DC bias and mixer
communications were quickly superseded by IEEE 802.11b to implement LC, for example a demonstration mentioned
(also known as WiFi 1), which could achieve a maximum in the meeting minutes in [14]. Meanwhile, the second ap-
speed of 11 Mbps. In 2000, research on OWC shifted to focus proach is very intrusive, which means that the modification
on the visible light spectrum due to the invention of highly- should be done from the digital domain onward. That is, an
efficient blue light emitting diodes (LEDs) [4]. This new in- effort in the digital domain must be made to ensure that the
terest led to standards in visible light communications (VLC), PHY layer of G.vlc is compatible with the Physical Layer
e.g., the VLC Consortium [5], IEEE 802.15.7 [6], IEEE Convergence Procedure Protocol Data Unit (PPDU) of IEEE
802.15.13 [7] and International Telecommunication Union - 802.11. Moreover, part of the data link layer of G.vlc should
Telecommunication Standardization Sector (ITU-T) G.9991 be adapted to the MAC layer of IEEE 802.11. Indeed, this
(also known as G.vlc) [8]. Meanwhile, radio frequency (RF)- implementation complexity also comes with an advantage in
based WiFi technologies have undergone several amendments, terms of error performance due to the adaptive bit loading.
e.g., IEEE 802.11a/g/n/ac (also known as WiFi 2 to 5) and, Our early contribution has been made in TGbb to justify two
more recently, IEEE 802.11ax (also known as WiFi 6) [9]. proposals by measuring the flatness of the wireless optical
In the latest Cisco Visual Networking Index [10], it is channels [15]. That is, the LC-optimized will be inefficient if
predicted that the annual
 global mobile  data traffic will reach the channel is flat. This flatness test led to a task group motion
almost 103 exabytes 109 gigabytes by 2022. This is known to use the first proposal as a common mode PHY [14].
as the mobile zettabyte era. In that year, WiFi alone will be The use of frequency up-conversion for LC has a significant

‹,(((

Authorized licensed use limited to: Macquarie University. Downloaded on June 22,2020 at 12:24:26 UTC from IEEE Xplore. Restrictions apply.
disadvantage, which is the loss of valuable frequency spectrum Even though it is not common, TGbb decides to add additive
near the DC. The frequency band near the DC is desired white Gaussian noise (AWGN) and instantaneous shot noise
because: (i) nowadays, the bandwidth of LEDs is still limited in the optical domain (after WOCIR), see the group discussion
to a few tens to hundreds of MHz, and (ii) due to the summary in [22]. Another bandpass filter is also used at the
low pass filter characteristic of the non-line-of-sight wireless receiver. Its cutoff frequencies are 48 kHz and 258 MHz, and
optical channels [16]. The reason for this loss is due to the the passband gain is 4.6 dB [21]. An automatic gain controller
spectral leakage of a non-ideal (non-square) spectrum of the (AGC) is used to maintain the signal amplitude before the
transmitted OFDM waveform, which causes aliasing near the frequency down-conversion. A direct conversion detection
DC spectrum. However, having a high center frequency will comprising frequency down-conversions, a matched filter and
also affect the error performance of the system caused by downsampling are used. Finally, recovery of the received
higher attenuation due to the low pass filter characteristic. signals is carried out based on IEEE 802.11ax specifications.
Therefore, a question arises: is there another non-intrusive and In this paper, we aim to compare the frequency up-and-
simple approach that does not sacrifice the near DC spectrum? down conversion with the IQ-WDM method. The main design
This paper aims to answer this question. objective of IQ-WDM remains the same as in the frequency
Contributions: This paper studies the potency of separately up-and-down conversion approach. That is, could we leverage
transmitting signals over in-phase and quadrature (IQ) compo- the output signal of off-the-shelf WiFi chipsets so that it can be
nents using two LEDs and wavelength division multiplexing transmitted over the optical spectrum? Instead of converting
(WDM), namely IQ-WDM. By transmitting baseband signals the IQ baseband signal to passband, the IQ-WDM method
from IQ components, the higher attenuation due to the higher processes both I and Q components separately and treats them
frequency band can be avoided. In addition, WDM is applied as two independent real signals as shown in Fig. 2. (An
in order to mitigate the inter-channel interference. We find that example of off-the-shelf boards that provide baseband I and Q
the error performance gain can be achieved while keeping the WiFi signals can be found in [23].) The upsampling and pulse
modifications to the WiFi silicon to a minimum. shaping are still used at each branch. An important assumption
A related work is discussed in [18], which presents non- in this paper is that there is no inter-channel-interference (ICI).
Hermitian symmetry (NHS)-OFDM. Other than the use of This can be achieved if two light sources operate at different
WDM and the implementation of IQ-WDM using signals fol- wavelengths, denoted by λ1 and λ2 , and both wavelenghts are
lowing IEEE 802.11ax, the contribution of our work compared sufficiently distant. In fact, this is feasible due to the operating
to NHS-OFDM empirically shows that the IQ imbalance due wavelength defined as a common mode is 800 nm to 1,000 nm
to differences in distances between each pair of LEDs and pho- [14]. Therefore, we can assume that two light sources operate
todiodes (PDs) is negligible. The IQ imbalance is mentioned in at near 800 nm and 1,000 nm, respectively. In addition, the ICI-
[18], but it is neglected for simplicity. Another related work is free assumption can also be used if narrowband light sources
recently discussed in [19], where Hermitian-based DC-OFDM such as laser diodes are used. WOCIRs for each wavelength
is used. Similar to the LC-optimized approach, modifications will be discussed in the next section.
to the WiFi silicon are required. Regarding the Rx model, we apply different responsivity
In Section II, our system model will be explained. It is (A/W) corresponding to each wavelength λ1 and λ2 . Following
followed by channel models in Section III. Results and dis- characteristics of typical silicon-based PDs based on [24], we
cussions will be provided in Section IV. Section V concludes add different responsivity factors for both Rx models at each
this paper. branch. In this paper, the responsivity factor for the I-branch
Rx model with λ1 is 0.67, and the other is 0.52. Hence, this
II. S YSTEM M ODEL will cause an IQ imbalance due to the difference in PDs’
attenuation. AGC is used as a first-aid compensation for the
Fig. 1 shows the frequency up-and-down conversion ap- IQ-imbalance. The main advantage of this is that it is not
proach which adheres to the baseline methodology docu- intrusive, i.e., no additional signal processing technique is
ment available in [17]. Following the latest work on the needed in the digital domain. Recall that the methodology
frequency up-and-down conversion approach [13], we use document in [17] also specifies the use of an AGC. This means
OFDM waveforms that comply with IEEE 802.11ax [20]. In that the default configuration detailed in the methodology
addition, intensity-modulation and direct-detection modulation document is robust to this type of imbalance. Matched filters
is assumed in this paper. An upsampling with 1 GSa/s and and downsamplers are used at each branch. Finally, both IQ
pulse shaping with the root raised cosine filter is used. Then, components are combined into complex signals before the
a RF mixer with a center frequency denoted by fc is applied. recovery process.
TGbb also defines a Tx model used for simulation purposes,
see [21]. Based on this, a bandpass filter with the cutoff III. C HANNEL M ODEL : WOCIR
frequencies of 260 kHz and 234 MHz and the passband gain TGbb, in fact, has provided publicly-available reference
of -23.17 dB is used to model the transmitter’s front-end. The channel models with different scenarios, see [25]. How-
channel model denoted by the wireless optical channel impulse ever, the dataset only contains single-input-single-output and
response (WOCIR) will be discussed in the next section. multiple-input-single-output cases. Moreover, the dataset only

Authorized licensed use limited to: Macquarie University. Downloaded on June 22,2020 at 12:24:26 UTC from IEEE Xplore. Restrictions apply.
Fig. 1. Frequency up-and-down conversion method [17].

Fig. 2. The proposed IQ-WDM for IEEE 802.11bb.

TABLE I. Reflectivities of materials used in Fig. 3.


materials reflectivities
paint 0.04
cotton 0.64
skin 0.66
plaster 0.83
pinewood 0.92

of the material for the skin of the user model is taken from
[28]. Based on [26]–[28], reflectivities of these materials at the
wavelength range of 800 nm to 1,000 nm can be considered
flat. Hence, we will not differentiate the reflectivities at λ1 and
λ2 . Values of reflectivities of these materials are summarized
Fig. 3. Elements of a simple office environment. in Table I.
Three different scenarios are used in this paper as illustrated
in Fig. 4. Note that as shown in the inset of Fig. 4(a), there
are two LEDs and PDs in each case to accommodate the IQ-
contain samples for the visible light spectrum only. Therefore, WDM. The distance between both LEDs is 5 cm. The same
we need to generate samples that describe sources occupying distance is also used for both PDs. The detection areas and
infrared spectrum and a 2 × 2 multiple-input-multiple-output field-of-views of all PDs are 1 cm2 and 85◦ , respectively. The
configuration. This section aims to explain this in detail. In half-power semiangles of all LEDs are 40◦ .
addition, another type of IQ imbalance will be introduced. Case 1 depicts the scenario where the user is near the LEDs,
Generally, the IQ imbalance is caused by the fact that the and the PDs face the LEDs. Case 2 depicts the scenario where
distances between each pair of LEDs and PDs are different. the user sits in the corner of the room and uses a LiFi-enabled
Consequently, attenuations and delays for both IQ components USB dongle. Line of sight (LOS) links in both cases exist, but
are not the same. Differences in the reflectiveness of materials the LOS link is not dominant in the second case. Meanwhile,
at different wavelengths might also cause unequal attenuations. case 3 is used to represent a scenario where a LOS link does
In this paper, we will focus on a simple office environment not exist. In short, case 1 is used to represent a channel where
comprising a 4 × 3 × 3 m3 room, a user, a desk and a chair the LOS link is dominant. The second case is used to represent
as depicted in Fig. 3. Arrows show the normal vectors or a channel that is dominated by the non-LOS channel while the
directions of corresponding objects. Fig. 3 depicts a scenario LOS still exists. The third case is used to represent a channel
where a user holds a phone next to their ears while facing a without a LOS link.
desk and a chair. Moreover, the PD is located on the top part In this paper, WOCIRs are generated by means of frequency
of the phone. In this paper, we assume that an LED is located domain approach [29]. To the best of our knowledge, it is one
in the center of the ceiling, facing downwards. of the accurate deterministic methods to generate WOCIRs
Based on [26, Figure 2], we model materials of the desk due to its ability to consider infinite number of reflections.
and chair’s surfaces as a black glossy paint. Plasters are used Phase responses can show the transmission delay at both I
to model the materials of the surfaces of the ceiling and the and Q branches. Based on Fig. 5, which shows the phase
walls. Pinewood is used to model the floor. Cotton is used responses of all cases, the delay differences between I and Q
to model the materials of the body and legs [27]. A model branches are negligible. Meanwhile, magnitude responses for

Authorized licensed use limited to: Macquarie University. Downloaded on June 22,2020 at 12:24:26 UTC from IEEE Xplore. Restrictions apply.
(a) Case 1 (b) Case 2 (c) Case 3
Fig. 4. Three tested cases. Note that in the case 2, the PDs are located on top of the table, which depict the use case of a LiFi-enabled USB dongle.

all cases are depicted in Fig. 6. Unlike the delay imbalance, the spectral leakage is worse in the first case than it does in
the gain imbalance between I and Q branches are noticeable, the third case. It is also important to note here that we do not
i.e., the difference is in the range of 1 to 2 dB at the lower use the same values of center frequency in all cases since we
frequency range, and it can be almost 20 dB for case 3 at aim to compare the IQ-WDM and the frequency upconversion
the high frequency range. Recall that we apply AGC for IQ- approach, whose center frequency is optimally chosen.
WDM (as well as the frequency upconversion method). In the Packet error ratio (PER) performances for all cases are
next section, it will be shown that this 2 dB difference in the depicted in Fig. 8. Note that for fairness of comparison, the
magnitude response is negligible. total signal-to-noise ratio (SNR) of both frequency upcon-
version and IQ-WDM is set to be the same. Generally, IQ-
IV. R ESULTS AND D ISCUSSIONS : E RROR P ERFORMANCE WDM outperform the frequency upconversion approach. The
Before discussing the error performance, it is important to main reason for this is due to the upconversion causing the
note here that the performance of the frequency upconversion transmitted signal to suffer from higher attenuation due to
approach highly depends on the upconverted center frequency. the low pass filter characteristic of the channels. This effect
We also notice that there is a trade-off in choosing either a can be seen clearly by the fact that the SNR gain widens
low or a high center frequency. With a low center frequency, as the channel gets worse, i.e., from 0.5 dB difference to 6
the frequency upconversion approach is vulnerable to aliasing dB difference. Comparisons of IQ-WDM with and without
effect near the DC frequency due to spectral leakage as IQ imbalance are also provided. In this paper, IQ-WDM
illustrated in Fig. 7. Meanwhile, with a high center frequency, without IQ imbalance refers to the scenario where channel
the frequency upconversion approach suffers from the low pass impulse responses (CIRs) of both I and Q branches are the
filter characteristic of non-LOS channels [16]. same. It can be seen that the difference in terms of PER is
Determining a good center frequency also depends on signal negligible. We empirically observe that this resilience is due
bandwidth. At the time of writing, the common mode PHY is to the use of AGC, which maintains signal amplitudes at the
defined with the signal bandwidth of 20 MHz, so we will use receiver. Finally, we can conclude that IQ-WDM is robust
that value in this paper. For each case and fixed modulation to IQ imbalance, assuming that the distance of an LED to
and coding scheme (MCS) index value, we numerically find another LED (and similarly a PD to another PD) is not far.
the best center frequency within the 10 MHz to 15 MHz range This assumption is valid due to the small size of LEDs and
of the frequency spectrum. The main reason for choosing this PDs.
range is that we expect the center frequency to be as close as
possible to the DC frequency in practice due to the bandwidth V. C ONCLUSIONS
limitation of LEDs. The problem formulation in searching for
the optimal center frequency for each MCS index value is as In this paper, we compared the agreed common mode
follows: mandatory PHY layer of IEEE 802.11b, i.e., the frequency
upconversion method, and our proposed IQ-WDM method.
fc∗ = arg min SNR to achieve PER =10−3 . Three different channel realizations were used for the compar-
10MHz ≤ fc ≤15MHz
ison. We generated a channel where the LOS link is dominant;
We find that with the MCS index value of 6 (64-QAM a channel where a non-LOS link is dominant, but the LOS link
and coding rate of 3/4) the optimal center frequencies for still exists; and a channel without a LOS link. We showed that
the first, the second and the third case are 15 MHz, 13 MHz IQ-WDM outperformed the frequency upconversion method in
and 12 MHz, respectively. The fact that the optimal center all cases. In addition, the proposed IQ-WDM was also shown
frequencies decrease can be seen from the flatness of the to be resilient to the gain and delay imbalance at both I and
magnitude response at the range of 0 to 20 MHz. That is, Q branches.

Authorized licensed use limited to: Macquarie University. Downloaded on June 22,2020 at 12:24:26 UTC from IEEE Xplore. Restrictions apply.
(a) Case 1 (a) Case 1

(b) Case 2 (b) Case 2

(c) Case 3 (c) Case 3

Fig. 5. Phase responses. Fig. 6. Magnitude responses.

VI. ACKNOWLEGMENTS
The work of A. Purwita was supported by Indonesian
Endowment Fund for Education (LPDP). The work of H. Haas
was supported by the Wolfson Foundation, the Engineering
and Physical Sciences Research Council (EPSRC) under Es-
tablished Career Fellowship grant EP/R007101/1.
Fig. 7. Aliasing near the DC frequency due to the spectral leakage.
R EFERENCES
[1] A. G. Bell, W. G. Adams, Tyndall, and W. H. Preece, “Discussion on
"the photophone and the conversion of radiant energy into sound",” J. [5] Visible Light Communications Consortium (VLCC). [Online].
Soc. Telegraph Eng., vol. 9, no. 34, pp. 375–383, 1880. Available: http://www.vlcc.net/
[2] Infrared Data Association (IrDA). [Online]. Available: https://www. [6] (2019, September) IEEE 802.15 WPAN™ 15.7 maintenance: Short-
irda.org/ range optical wireless communications task group (TG 7m).
[3] “IEEE Standard for Wireless LAN Medium Access Control (MAC) and [Online]. Available: http://www.ieee802.org/15/pub/IEEE%20802_15%
Physical Layer (PHY) specifications,” IEEE Std 802.11-1997, pp. 1–445, 20WPAN%2015_7%20Revision1%20Task%20Group.htm
Nov 1997. [7] (2019, September) IEEE 802.15 WPAN™ task group 13 (TG13)
[4] Y. Tanaka, S. Haruyama, and M. Nakagawa, “Wireless optical transmis- multi-gigabit/s optical wireless communications. [Online]. Available:
sions with white colored LED for wireless home links,” in 11th IEEE http://www.ieee802.org/15/pub/TG13.html
PIMRC, vol. 2, London, UK, United Kingdom, Sep. 2000, pp. 1325– [8] “High-speed indoor visible light communication transceiver – system
1329 vol.2. architecture, physical layer and data link layer specification,” Rec. ITU-

Authorized licensed use limited to: Macquarie University. Downloaded on June 22,2020 at 12:24:26 UTC from IEEE Xplore. Restrictions apply.
test-of-the-tgbb-refrence-channel-models-with-he-waveform.pptx
[16] V. Jungnickel, V. Pohl, S. Nonnig, and C. von Helmolt, “A physical
model of the wireless infrared communication channel,” IEEE IEEE J.
Sel. Areas Commun., vol. 20, no. 3, pp. 631–640, April 2002.
[17] K. L. Bober, V. Jungnickel, M. Hinrichs, and N. Serafimofski. (2019,
May) TGbb: Evaluation methodology for PHY and MAC proposals.
[Online]. Available: https://mentor.ieee.org/802.11/dcn/19/11-19-0187-
04-00bb-evaluation-methodology-for-phy-and-mac-proposals.docx
[18] C. Chen, W.-D. Zhong, and D. Wu, “Non-Hermitian Symmetry Orthogo-
nal Frequency Division Multiplexing for Multiple-Input Multiple-Output
Visible Light Communications,” J. Opt. Commun. Netw., vol. 9, no. 1,
pp. 36–44, Jan 2017.
[19] A. Zeshan and T. Baykas, “Performance analysis of VLC based on
802.11ac frame structure,” IEEE Communications Letters, vol. 23, no. 9,
(a) Case 1 pp. 1560–1563, Sep. 2019.
[20] “IEEE draft standard for information technology – telecommunications
and information exchange between systems local and metropolitan area
networks – specific requirements part 11: Wireless LAN medium access
control (MAC) and physical layer (PHY) specifications amendment
enhancements for high efficiency WLAN,” IEEE P802.11ax/D3.0, June
2018, pp. 1–682, July 2018.
[21] M. Hinrichs, J. Hilt, P. Hellwig, V. Jungnickel, and K. L.
Bober. (2019, January) IEEE 802.11 TGbb Task Group
on Light Communications: Optical Frontend Model. [Online].
Available: https://mentor.ieee.org/802.11/dcn/19/11-19-0087-01-00bb-
optical-frontend-model-for-phy-simulation.docx
[22] K. L. Bober, R. Mennecke, and N. Serafimofski. (2019, March)
IEEE 802.11 TGbb Task Group on Light Communications:
September, 2019 Vancouver Meeting Minutes. [Online]. Avail-
able: https://mentor.ieee.org/802.11/dcn/19/11-19-0523-00-00bb-tgbb-
(b) Case 2 meeting-minutes-for-march-2019.docx
[23] Qualcomm Technologies, Inc. (2016, Sepp) Qualcomm® Snapdragon™
410E(APQ 8016E) processor: Device Specification. [Online]. Avail-
able: https://www.qualcomm.com/media/documents/files/snapdragon-
410e-apq-8016e-data-sheet.pdf
[24] Edmund Optics. Basic Principles of Silicon Detectors. [Online].
Available: https://www.edmundoptics.com/resources/application-notes/
testing-and-detection/basic-principles-of-silicon-detectors/
[25] M. Uysal, F. Miramirkhani, T. Baykas, and K. Qaraqe. (2018, November)
CIRs of IEEE 802.11bb reference channel models for indoor envi-
ronments. [Online]. Available: https://mentor.ieee.org/802.11/dcn/18/11-
18-1603-01-00bb-cirs-of-ieee-802-11bb-reference-channel-models.zip
[26] F. Miramirkhani, M. Uysal, and E. Panayirci, “Novel channel models
for visible light communications,” in Broadband Access Communication
Technologies IX, B. B. Dingel and K. Tsukamoto, Eds., vol. 9387,
International Society for Optics and Photonics. SPIE, 2015, pp. 150 –
(c) Case 3 162.
Fig. 8. PER vs. SNR with 64-QAM and coding rate of 3/4. [27] R. Kokaly et al., “USGS Spectral Library Version 7,” U.S.
Geological Survey Data Series 1035, 2017. [Online]. Available:
https://crustal.usgs.gov/speclab/data/GIFplots/GIFplots_splib07a/
ChapterA_ArtificialMaterials/splib07a_Cotton_Fabric_GDS437_White_
T G.9991, International Telecommunications Union, 2019. ASDFRa_AREF.gif
[9] S. McCann. (2019, Oct) Official IEEE 802.11 Working Group Project [28] C. C. Cooksey and D. W. Allen, “Reflectance measurements of human
Timelines. [Online]. Available: http://grouper.ieee.org/groups/802/11/ skin from the ultraviolet to the shortwave infrared (250 nm to 2500
Reports/802.11_Timelines.htm nm),” in Proceedings of SPIE Defense, Security & Sensing, Baltimore,
[10] “Cisco visual networking index: Global mobile data traffic forecast MD, May 2013.
update, 2017–2022,” White Paper, Cisco, Feb. 2019. [29] H. Schulze, “Frequency-domain simulation of the indoor wireless optical
[11] (2017, May) Status of IEEE 802.11 light communication TG. [Online]. communication channel,” IEEE Trans. Commun., vol. 64, no. 6, pp.
Available: http://www.ieee802.org/11/Reports/tgbb_update.htm 2551–2562, June 2016.
[12] H. Haas, L. Yin, Y. Wang, and C. Chen, “What is LiFi?” J. Lightw.
Technol., vol. 34, no. 6, pp. 1533–1544, March 2016.
[13] N. Serafimovski, C. Han, T. Weszely, S. Videv, A. Purwita, M. Dehghani,
C. Chen, and H. Haas. (2019, September) Proposal for common-mode
mandatory phy. [Online]. Available: https://mentor.ieee.org/802.11/dcn/
19/11-19-1625-04-00bb-proposed-common-mode-mandatory-phy.pptx
[14] S.-K. Lim, M. Wendt, and N. Serafimofski. (2019,
September) IEEE 802.11 TGbb Task Group on Light
Communications: September, 2019 Hanoi Minutes. [Online].
Available: https://mentor.ieee.org/802.11/dcn/19/11-19-1704-00-00bb-
meeting-minutes-for-september-2019-hanoi.docx
[15] A. Purwita, S. Videv, H. Haas, and N. Serafimofski. (2019,
September) Spectral Flatness Test of the TGbb’s Reference
Channel Models with HE Waveform. [Online]. Available: https:
//mentor.ieee.org/802.11/dcn/19/11-19-1639-01-00bb-spectral-flatness-

Authorized licensed use limited to: Macquarie University. Downloaded on June 22,2020 at 12:24:26 UTC from IEEE Xplore. Restrictions apply.

Вам также может понравиться