Optimal Packet Size

in Optical Wireless Communication

André G. Vieira and Luiz F. M. Vieira and Marcos A. M. Vieira
Computer Science Department - Universidade Federal de Minas Gerais
Email: {andregarcia, lfvieira, mmvieira}@dcc.ufmg.br

Abstract—Optical Wireless Communication (OWC) proved II. R ELATED W ORK

to be one of the complementary technologies to radio fre-
quency, which can help solve the crowded spectrum problem.
Furthermore, new applications such as the Internet of Things Here, we describe the related work to the optimal packet
and Wireless Sensor Networks can benefit from OWC. With size, maximizing the channel use efficiency, and wireless
the increase in connected devices, using the channel efficiently optical communication.
becomes even more necessary. This paper presents a model for Pompili et al. [4] correlate channel usage efficiency with
the OWC channel and develops a method to calculate the OWC
channel’s efficiency. The model was validated experimentally with
the packet size that optimizes this efficiency for underwater
the OpenVLC 1.1 platform. With it, it is possible to determine networks. Soares Junior and Vieira [5] also show a correlation
the ideal packet size to maximize the channel’s efficiency. Using between packet size and channel usage efficiency, optimizing
Pearson’s correlation method between model data and experi- the packet size to maximize channel utilization. Their study
mental values, we obtained up to 99.8% correlation. focused on ZigBee networks [6].
Some studies have discussed optimizing network parameters
using packet size as a decision variable. Sankarasubrama-
I. I NTRODUCTION niam et al. [7] focus on packet size optimization for energy
efficiency in wireless sensor networks [8], [9]. Jun Yin et
Optical Wireless Communication (OWC) has been gaining al. [10] model the optimal packet size to maximize throughput.
prominence in industry and academia. It has shown to be a Vieira et al.[11] propose a protocol that reduces the latency
complementary alternative to radio frequency (RF) communi- of OWC transmissions using network coding. Soares Junior
cation and can solve the spectrum crunch problem [1]. This et al. [12] propose a technique to maximize throughput in
band occupation has attracted the attention of the academy [2]. wireless sensor networks. No studies have been found that
Mobile devices connected to the Internet have increased, describe the efficiency of using the optical wireless channel.
leading to a higher demand for bandwidth and alternative Visible light communication (VLC) and wireless optical
technologies. communication research areas have increased interest, and the
Cisco [3] estimates that by 2030, 500 billion devices will number of publications is growing yearly [13]. Shao et al. [14]
be connected to the Internet. All of these devices will require and Basnayaka and Haas [15] propose using hybrid commu-
bandwidth and generate much traffic. OWC emerges as a nication systems that use RF and VLC in a complementary
promising communication technology, offering the essential way.
features needed for the connectivity demanded by new devices. In ad hoc network scenarios [16], Cen et al. [17] mention
Competition and demand for the use of the channel will be several applications that can use VLC technology for military
high. Therefore, the optical wireless channel must be used as or civilian purposes. Among the points covered by their work,
efficiently as possible. we can highlight the high security of OWC networks and the
The main contributions of this article are: (i) presentation low energy consumption compared to some RF technologies.
of a model for the optical wireless communication channel, In routing, Jagannath and Melodia [18] developed a routing
(ii) development of a model to maximize the efficiency of protocol for OWC. Such a protocol makes use of link layer
the use of the channel in OWC, (iii) real-world experimental information to perform routing. Matheus et al. [19] also
validation of the efficiency model, and (iv) results that show a developed a dynamic routing protocol for OWC, which can
Pearson correlation between experimental data and the model adapt to changes in network topology or even shadow and
of up to 99.8%. obstacle effects.
The rest of the article is organized as follows: in Section II, In summary, several studies in the literature optimize the
we present the related work. Section III details OWC and packet size to benefit some network parameters. However,
features a channel model. Section IV models the problem and none of them optimizes the use of the optical wireless channel.
presents the results obtained with the modeling. Section V Unlike other works, this maximizes the efficiency of the
brings the model experimental evaluation. Finally, Section VI optical communication channel usage through the optimum
concludes the paper and indicates future work. packet size and presents experimental results.
 TX Packet RX
III. O PTICAL W IRELESS C OMMUNICATION
Header Payload
The VLC uses the range of the electromagnetic spectrum
with wavelengths that can vary between 380 nm to 780 nm
and had its standard defined in 2011 [20]. The standard was

Time
revised in 2018 [21], increasing coverage with wavelengths trx-tx
ACK
from 190 nm to 10,000 nm.
Light can be used as a means of communication [22] and
has several research areas, including body area networks,
communication between vehicles, smart cities, etc. Several
applications include a light positioning system [23], indoor Fig. 2: Diagram for one transmission.
location [24], [25], and the LED-LED or LED-photoreceptor
communication itself.
The optical channel modeling considered in this work is the
same as described by Zhong and Wang [26]. Equation 1 gives σshot = 2q[RPr (1 + MI2 ) + Ibg I2 ]B, (5)
the power received in the photoreceptor PR , where H(0) is
the attenuation of the optical channel, and Pt is the power where q is the electron charge, MI is the modulation index,
emitted by the transmitting LED. Ibg is the background current, I2 is the noise factor and B
is the noise bandwidth. Without generality loss, we use the
PR = H(0)Pt (1) OOK modulation standard (On-Off Keying) for this work. The
modulation index for OOK is 0.5.
Fig. 1 shows the geometry involving communication be-
tween an LED transmitting data and a photoreceptor receiving 
I2 2πΓ

data. The vectors D⃗r and D ⃗s are, respectively, normal to the σthermal = 8πkTk ηAB 2 + ηAI3 B , (6)
G gm
surface of the sender and normal to the surface of the receiver.
The D⃗rs vector points from the photoreceptor’s center towards where k is the Boltzmann constant, Tk is the absolute
the transmitting LED’s center. temperature, η is the capacitance, G is the voltage gain, Γ
is the noise factor, Gm is the FET transconductance ,and I3
is the circuit noise [27].
IV. P ROBLEM M ODELING
This work aims to analyze the impact of packet size on the
efficiency of the wireless optical channel. The model proposed
here can be used to determine the optimal packet size for
transmission between two nodes from a distance d between
Fig. 1: Optical geometric scheme.
them and the θ and φ alignment angles (Equation 2 and 3).
Then, let us consider two OWC communication devices,
Assuming that the LED emits obeying the Lambertian emis-
each equipped with an LED for data transmission and a
sion model, we have that the emitted spectral radiation (R(φ) )
photoreceptor for data reception. Let E(d, Tp ) be the equation
can be obtained by Equation 2 where m is the Lambertian
ln0.5 determining the efficiency of using the channel for commu-
order given by m = ln(cosφ , where φ1/2 is the half power
1/2 ) nication. We want to choose the payload size Tp such that
angle and, for the LED, φ1/2 = 60o . it maximizes E for any value d, θ, and φ. The angles do
(m + 1)cosm φ not directly enter the efficiency calculation. However, they are
R(φ) = . (2) used to obtain the signal-to-noise ratio (SNR).
2π
The model considers the communication between two
The attenuation of the channel can then be calculated using
nodes. The transmission of a packet will be regarded as
Equation 3, where A is the photoreceptor area and d is the
correct when there is no wrong bit in the data packet and the
distance between the devices.
acknowledgment packet (ACK). Fig. 2 illustrates the Diagram
A for a complete transmission.
H(0) = R(φ) cosθ. (3) The IEEE 802.15.7 standard specifies several modulation
d2
schemes. We considered the OOK standard for this work, but
Finally, we can calculate the signal-to-noise ratio (SNR)
the same can be applied to the other modulation standards.
using Equation 4
The Bit Error Rate, Be for the OOK modulation, can be
R2 H 2 (0)Pt2 calculated using Equation 7 [28].
SN R = , (4)
σthermal + σshot √
Be = Q( SN R), (7)
where σthermal and σshot are, respectively, shot noise and √
thermal noise obtained in Equation 5 and in Equation 6. where Q( SN R) is the Marcum-Q function.
 With the value obtained from the error rate, it is possible to We chose three header sizes, two of which were specified by
calculate the packet error rate Pe (Equation 8). N is the value the IEEE organization, 37 bytes and 9 bytes (Maximum and
of the total bits transmitted in a data packet plus the total bits minimum IEEE header size). The third header size chosen was
in an acknowledgment packet. 11 bytes. The value was selected to be tested experimentally
using the OpenVLC [29] platform.
Pe = 1 − (1 − Be )N (8)
Feature Symbol Value
With the packet error rate computed, it is possible to Payload TP variable (bits)
Transmission counter Pc variable1
calculate the number of packets that must be sent for error-free Header size Tc variable (296, 88, 72 bits)
transmission Pc . This calculation is based on the assumption ACK size TA 24 bits
that errors in consecutive packets are independent and that the Bandwidth b 11.67 kbps
Distance d variable in meters
number of retransmissions is infinite (geometric distribution). Light speed c 3x108 m/s
This can be obtained from Equation 9.
TABLE II: Parameters for evaluating the channel efficiency.
1
Pc = (9)
1 − Pe The parameters presented in Table II were used in Equa-
Finally, the channel efficiency E(d, Tp ) can be found as a tion 10 to define the optimal packet size that maximizes
function of the distance and the packet size (Equation 10). the channel efficiency. Fig.s 3a, 3b and 3c show the curves
obtained for the maximum channel efficiency for the headers
of 37, 11, and 9 bytes, respectively.
Tp
E(d, Tp ) = d
, All curves grow to a maximum. This maximum value is
Pc ∗ [Tp + Tc + TA + b ∗ (2 ∗ + trx−tx )]
c the efficiency limit obtained for each distance. We obtain the
(10)
payload value at the maximum points for each curve that
Where Tp is the packet size, Tc and TA are the header and maximizes the channel’s efficiency. This payload value plus
ACK sizes, respectively, the bandwidth is given by b, c is the the header size is called the optimal packet size.
speed of light, and trx−tx is the data packet processing time. The results demonstrate that header size directly impacts
Maximizing channel efficiency is crucial to determine the channel efficiency. Even though larger headers might suggest
optimal packet size, as expressed in Equation 10. This analysis bigger ideal packet sizes, the increased overhead from the
is split into two main sections. Firstly, in Section IV-A, header itself may not outweigh this benefit, especially over
we examine the influence of distance on channel efficiency, longer distances. As the payload size shrinks compared to the
keeping transmitter-receiver alignment fixed (φ and θ at zero). larger header, efficiency suffers.
Secondly, in Section IV-B, we explore channel efficiency
across various distances and angles. B. Non-aligned nodes
By defining the distance between the sender and receiver, we
A. Node alignment can analyze how the optimal packet size behaves by varying
The parameter values used to compute the channel attenu- the nodes’ alignment.
ation in Equation 3 and the SNR in Equation 4 are shown First, we set the distance to 0.2 m and the θ angle to 0º,
in Table I. The parameters used were obtained from the and we vary the φ angle. The header size chosen here is 11
datasheets of the components employed. bytes, equivalent to the OpenVLC header.
Fig. 3d shows the efficiency of using the channel as a
Feature Symbol Value
Photoreceptor area A 5.24x10−6 m2
function of the payload for several angles at 0.20 m between
Photoreceptor responsiveness R 0.45 AW the transmitter and the receiver. For angles less than 80º, occur
Transmitted Power Pt 20x10−3 W a small variation in maximum efficiency. However, maximum
Angle of the receiver φ 0o efficiency drops very quickly at angles greater than 80º.
Reception angle θ 0o
Background current Ibg 5100 µA Next, we set the distance to 0.6 m and repeated the process.
Noise factor I2 0.562 Fig. 4a shows the channel efficiency for a distance of 0.6
Absolute temperature Tk 297 K m. The model indicates that channel efficiency happens more
Capacitance η 112x10−6 pF/cm2
Voltage Gain G 10 smoothly for this distance. However, the model predicts that
Circuit noise I3 0.0868 transmissions are no longer possible at 60º due the fact no
Modulation Index OOK MI 0.5 there is a transition that does not contain an error. Such
Transconductance Gm 30−3
predictions occur for the parameters shown in Tables I and II.
TABLE I: Parameter for SNR computation. Finally, we keep our distance and vary both angles, θ and φ.
We did this using distances of 0.2 m and 0.6 m. We observed
The IEEE 802.15.7 standard specifies all header fields for the same behaviors, rapid growth to maximum efficiency and
the OWC. This header can take different sizes depending on slightly smoother decay. Fig.s 4b and 4c show the efficiency
the type of the packet, which can be the type of beacon, data of using the optical channel for several angles fixing distances
type, confirmation type, command type, etc. of 0.2 m and 0.6 m, respectively.
 Channel Efficiency for 37 bytes header Channel Efficiency for 11 bytes header
1
10 cm 10 cm
30 cm 0.9 30 cm
40 cm 40 cm
50 cm 0.8 50 cm
60 cm 60 cm
70 cm 0.7 70 cm
Channel Efficiency

Channel Efficiency
80 cm 80 cm
100 cm 100 cm
0.6

0.5

0.4

0.3

0.2

0.1

0
0 500 1000 1500 2000
Payload (Bytes) Payload (Bytes)

(a) Channel efficiency for the 37-byte header. (b) Channel efficiency for the 11-byte header.
Channel Efficiency for 9 bytes header
1 Channel efficiency header 11 bytes, d = 20 cm, multiple angles
10 cm 1
0.9 30 cm
40 cm 0.9
0.8 50 cm 5 degree
60 cm 0.8
80 degree
0.7 70 cm
Channel Efficiency

85 degree

Channel Efficiency
80 cm 0.7
100 cm 88 degree
0.6
0.6
0.5 0.5
0.4 0.4
0.3 0.3

0.2 0.2

0.1 0.1

0 0
0 500 1000 1500 2000 0 500 1000 1500
Payload (Bytes) Payload (Bytes)

(c) Channel efficiency for the 9-byte header. (d) Channel efficiency for the 11-byte header and multiple angles.
Fig. 3: Channel efficiency in several scenarios.

Channel efficiency header 11 bytes, d = 60 cm, multiple angles

0.9
5 degree
0.8 25 degree
30 degree
0.7 35 degree
40 degree
Channel Efficiency

0.6 45 degree
50 degree
0.5 55 degree
60 degree
0.4

0.3

0.2

0.1

0
0 500 1000 1500
Payload (Bytes)

(a) Optimal payload size and maximal chan- (b) Optimal payload size and maximal chan- (c) Optimal payload size and maximal chan-
nel usage efficiency per θ. Nodes distance nel usage efficiency. Nodes distance 0.2 me- nel usage efficiency. Nodes distance 0.6 me-
is 0.6 m. ters, the nodes varying the θ and φ angles. ters, the nodes varying the θ and φ angles.
Fig. 4: Channel efficiency usage for different angles.

At 0.2 m distance, with equal angle variations, there’s no V. E XPERIMENTAL E VALUATION

efficiency drop up to 65º when using the optical channel. Past
65º, efficiency declines rapidly, ceasing entirely at 75º. For the experimental evaluation, we used the OpenVLC
1.1 platform. OpenVLC is a low-cost, open-source, wireless
visible light communication platform [29]. It works with a
development platform, the BeagleBone Black (BBB).
Now, we set the distance at 0.6 m and repeat the process Fig. 5 shows the experimental environment. The red arrow
we did for 0.2 m. The efficiency curves are more evenly points to the photoreceptor. Two OpenVLC boards coupled to
distributed as we vary the angles. The uniform distribution is BBBs on the top of a table, so the distance between them
expected since the luminous energy is distributed in space. For could vary. The lighting in the room was controlled. This care
the distance of 0.6 m between the nodes, the most significant was taken so that there was no photoreceptor saturation, which
possible angle was 63º, with an efficiency below 10%. did not become a factor influencing the results.
 Fig. 6b shows the experimental results in contrast to the
model’s prediction. The points obtained are represented in
the graph with a 95% confidence interval, and the correlation
between the curves is 0.977.
Fig. 6c shows the channel efficiency usage for a distance
of 60 cm. The model predicts that for this distance, the
payload value that maximizes the channel usage is 225 bytes.
The theoretical curve and the experimental values obtained
Fig. 5: Experimental environment with two BeagleBones and
correlate with 0.943. Correlation values between 0.9 and 1
two OpenVLC platforms.
show a very high correlation between the curves [31].
This high correlation shows that the experimental data are
consistent with the values obtained through the model. In other
With the number of packets sent, the number of packets sent
words, the two curves follow the same trend. When looking at
correctly, and the size of the payload about the number of bits
Fig. 6a, Fig. 6b, and Fig. 6c, the theoretical and experimental
needed to transmit, we can calculate the channel efficiency
curves appear to be separated by a constant.
usage Em via Equation 11.
As the experimental data factors are not precisely the values
Pe c ∗
Tp calculated by the model, we can mention that they are based
Tp +Tc +TA
Em = , (11) on the existence of free space between the transmitter and the
Ps receiver. Therefore, the model does not predict interference
where Pe c is the number of packets sent without errors, and noise due to signal reflection. The experimental apparatus was
Ps is the total number of packets sent. mounted on a table so that the light emitted was reflected by
We conducted experiments at two distances: 0.2 m and the table, generating noise.
0.6 m. At 0.2 m, we initially aligned the devices with θ and LED power and photoreceptor data are based on nominal
φ angles set to 0º. Payload sizes of 75, 500, 1000, and 1500 values from manufacturers, lacking documentation on LED
bytes were chosen to cover a central range and align with the power supply in OpenVLC. Our model assumes ideal emission
theoretical modeling curve. following the Lambertian model, though real LEDs may vary.
We chose five different payload sizes to perform the exper- Due to limitations of our LED, transmission beyond 50º isn’t
iments at a distance of 60 cm. The sizes used were 225 bytes, feasible, as it primarily emits at smaller angles.
the optimal payload size for a 9-byte header, and a distance
of 60 cm. Two sizes below 225 bytes, 75 and 150 bytes, and VI. C ONCLUSION AND F UTURE W ORK
two values above 300 and 375 bytes were also chosen.
Per packet size, we ran 30 trials sending packets continu- Optical Wireless Communication is a viable solution to
ously for 150 seconds, recording sent and error-free counts. complement RF by providing a solution to the scarcity of spec-
trum, offering necessary resources to many devices. Transmit-
Distance Payload size Channel Efficiency
0.2 m 1863 bytes 99.2%
ting data efficiently and saving resources are essential for many
0.4 m 1500 bytes 97.8% applications. This paper presented a model to determine the
0.6 m 225 bytes 87.2% transmitted packet size to maximize channel usage efficiency.
0.8 m 50 bytes 58.0%
The experimental data makes it possible to conclude that
TABLE III: Optimal payload size, maximal channel efficiency. the proposed model to determine the optimal packet size to
maximize channel efficiency is quite helpful. Furthermore,
Table III summarizes optimal payload values from our the model shows a high correlation with experimental data
model alongside the maximum channel usage efficiency for reaching up to 0.998.
various distances. We assume a 9-byte header size for the table. Future work includes evaluating the model against the
For instance, at a 0.4 m node distance, the optimal payload size other modulation standards described by the IEEE 802.15.7
for maximizing efficiency is 1500 bytes, achieving a maximum standard. It is also expected to assess the model in scenarios
efficiency of 97.8%. where several nodes are communicating and in multiple hop
Fig. 6a illustrates channel efficiency versus payload size at a scenarios, thus analyzing the impact this can have on the
0.2m distance. We compare the model curve with experimental channel usage efficiency.
data and apply Pearson’s correlation method [30], yielding a
correlation trend of 0.998 between the two curves. ACKNOWLEDGEMENTS
After doing the experiments at 0.2 m with the nodes aligned,
we only rotated the emitter by 40º. The receiver continues We acknowledge the financial support received from
pointing toward the sender. This setting varies the φ by 40º CNPq processes 309496/2019-5; CAPES Finance Code 001;
and keeps the θ at 0º. We used five sizes of payload: 25 bytes, FAPEMIG procs. 02400-18; and FAPESP procs. 2020/05131-
75 bytes, 125 bytes, 175 bytes, and 350 bytes. 0; and UFMG PPRq.
 1
1 1
Model
0.9 Experiment
0.9 0.9
Model 0.8

Channel Efficiency
Channel Efficiency

0.8 0.8
Experiment Model

Channel Efficiency
0.7 0.7 Experiment
0.7
0.6 0.6
0.6
0.5 0.5
0.5
0.4 0.4
0.4
0.3 0.3
0.3
0.2 0.2
0.2
0.1 0.1
0.1
0 0
00
200 600 1000 1400 1800 0 50 100 150 200 250 300 350 400 450 500 0 100 200 300 400 500 600 700 800
Payload (bytes) Payload (bytes) Payload (bytes)

(a) Channel efficiency usage for a distance of (b) Channel efficiency usage for a distance of (c) Channel efficiency usage for a distance of
20 cm. 20 cm, angle = 40◦ . 60 cm.
Fig. 6: Channel efficiency usage for different distances and angles.

R EFERENCES [15] D. A. Basnayaka and H. Haas, “Hybrid rf and vlc systems: Improving
user data rate performance of vlc systems,” in 2015 IEEE 81st Vehicular
Technology Conference (VTC Spring), May 2015, pp. 1–5.
[1] D. Goldman, “Sorry, America: Your wireless airwaves are full,” https: [16] M. A. M. Vieira, R. Govindan, and G. S. Sukhatme, “An autonomous
//money.cnn.com/2012/02/21/technology/spectrum crunch/index.htm, wireless networked robotics system for backbone deployment in highly-
February 2012, [Online; access July-07-2020]. obstructed environments,” Ad Hoc Networks, vol. 11, no. 7, pp. 1963–
[2] J. P. D. Vries, L. Simić, A. Achtzehn, M. Petrova, and P. Mähönen, 1974, 2013.
“The wi-fi “congestion crisis”: Regulatory criteria for assessing spectrum [17] N. Cen, J. Jagannath, S. Moretti, Z. Guan, and T. Melodia, “Lanet:
congestion claims,” Telecommunications Policy, vol. 38, no. 8, pp. 838 Visible-light ad hoc networks,” Ad Hoc Networks, vol. 84, pp. 107–123,
– 850, 2014. 2019.
[3] CISCO, “Internet of Things,” https://www.cisco.com/c/dam/en/us/ [18] J. Jagannath and T. Melodia, “Vl-route: A cross-layer routing protocol
products/collateral/se/internet-of-things/at-a-glance-c45-731471.pdf, for visible light ad hoc network,” preprint arXiv:1904.05177, 2019.
2019, [Online; access November-27-2019]. [19] L. M. Matheus, A. B. Vieira, M. A. M. Vieira, and L. F. M. Vieira,
[4] D. Pompili, T. Melodia, and I. F. Akyildiz, “Routing algorithms for “Dyrp-vlc: A dynamic routing protocol for wireless ad-hoc visible light
delay-insensitive and delay-sensitive applications in underwater sensor communication networks,” Ad Hoc Networks, vol. 94, p. 101941, 2019.
networks,” in Proceedings of the 12th Annual International Conference [20] IEEE, “Ieee standard for local and metropolitan area networks–part 15.7:
on Mobile Computing and Networking, ser. MobiCom ’06. New York, Short-range wireless optical communication using visible light,” IEEE
NY, USA: ACM, 2006, pp. 298–309. Std 802.15.7-2011, pp. 1–309, Sep. 2011.
[5] W. A. S. Soares Junior and L. F. M. Vieira, “Redes aéreas ieee 802.15.4: [21] ——, “Ieee standard for local and metropolitan area networks–part 15.7:
tamanho ótimo do pacote e coleta de dados,” 2014. Short-range optical wireless communications,” IEEE Std 802.15.7-2018
[6] Z. Alliance, “Zigbee alliance finalizes specification,” http://www.zigbee. (Revision of IEEE Std 802.15.7-2011), pp. 1–407, April 2019.
org, 2004, [Online; acessado 25/03/2019]. [22] L. E. M. Matheus, A. B. Vieira, J. H. F. Freire, L. F. M. Vieira, M. A. M.
[7] Y. Sankarasubramaniam, I. F. Akyildiz, and S. W. McLaughlin, “Energy Vieira, and O. Gnawali, “Comunicação por Luz Visı́vel: Conceitos,
efficiency based packet size optimization in wireless sensor networks,” Aplicações e Desafios. Minicurso do Simpósio Brasileiro de Redes de
in Proceedings of the First IEEE International Workshop on Sensor Computadores e Sistemas Distribuı́dos 2017,” 2017.
Network Protocols and Applications, 2003., May 2003, pp. 1–8. [23] M. Kavehrad and W. Zhang, Light positioning
[8] L. B. Ruiz, L. H. A. Correia, L. F. M. Vieira, D. F. Macedo, E. F. Naka- system (LPS). Cambridge University Press, Feb
mura, C. M. Figueiredo, M. A. M. Vieira, E. H. B. Maia, D. Câmara, 2015. [Online]. Available: https://www.cambridge.org/core/
A. A. Loureiro et al., “Architectures for wireless sensor networks,” in books/visible-light-communication/light-positioning-system-lps/
Proceedings of the 22nd Brazilian Symposium on Computer Networks 3836D6AC658A4A5B4AD4ED3B5EFB3A5B
(SBRC’04), 2004, pp. 167–218. [24] N. U. Hassan, A. Naeem, M. A. Pasha, T. Jadoon, and C. Yuen, “Indoor
positioning using visible led lights: A survey,” ACM Comput. Surv.,
[9] L. F. M. Vieira, M. A. M. Vieira, L. B. Ruiz1, A. A. F. Loureiro,
vol. 48, no. 2, pp. 20:1–20:32, Nov. 2015.
D. C. Silva, and A. O. Fernandes, “Efficient incremental sensor network
[25] V. Kavitha, B. A. Vijayalakshmi, M. VanithaLakshmi et al., “Vlc-based
deployment algorithm,” pp. 3–14, 2004.
dco-ofdm and wsn in hospitals for medical information transmission,”
[10] Jun Yin, Xiaodong Wang, and D. P. Agrawal, “Optimal packet size in Journal of Optics, pp. 1–9, 2024.
error-prone channel for ieee 802.11 distributed coordination function,” [26] W. Zhong and Z. Wang, “Performance enhancement techniques for
in 2004 IEEE Wireless Communications and Networking Conference indoor vlc systems,” in Visible light Communication. Cambridge
(IEEE Cat. No.04TH8733), vol. 3, March 2004, pp. 1654–1659 Vol.3. University Press, 2015.
[11] A. G. Vieira, L. F. M. Vieira, M. A. M. Veira, and A. B. Vieira., [27] H. Kressel, Semiconductor devices for optical communication. Springer
“Comunicação por luz visı́vel: redução da latência via combinação de Science & Business Media, 2006, vol. 39.
pacotes,” in SBESC 2018 (), nov 2018. [Online]. Available: https://sbesc. [28] T. Komine and M. Nakagawa, “Fundamental analysis for visible-light
lisha.ufsc.br/sbesc2018/proceedings2018/S2/Comunicacao187820 1.pdf communication system using led lights,” IEEE Transactions on Con-
[12] N. dos Santos Ribeiro Júnior, R. C. Tavares, M. A. M. Vieira, L. F. M. sumer Electronics, vol. 50, no. 1, pp. 100–107, Feb 2004.
Vieira, and O. Gnawali, “Codedrip: Improving data dissemination for [29] Q. Wang, D. Giustiniano, and D. Puccinelli, “Openvlc: Software-defined
wireless sensor networks with network coding,” Ad Hoc Networks, visible light embedded networks,” in Proceedings of the 1st ACM
vol. 54, pp. 42–52, 2017. MobiCom workshop on Visible light communication systems. ACM,
[13] L. E. M. Matheus, A. B. Vieira, L. F. M. Vieira, M. A. M. Vieira, and 2014, pp. 15–20.
O. Gnawali, “Visible light communication: Concepts, applications and [30] M. Grčar, D. Mladenič, B. Fortuna, and M. Grobelnik, “Data sparsity is-
challenges,” IEEE Communications Surveys & Tutorials, vol. 21, no. 4, sues in the collaborative filtering framework,” in International Workshop
pp. 3204–3237, 2019. on Knowledge Discovery on the Web. Springer, 2005, pp. 58–76.
[14] S. Shao, A. Khreishah, M. B. Rahaim, H. Elgala, M. Ayyash, T. D. C. [31] M. Mukaka, “Statistics corner: a guide to appropriate use of correlation
Little, and J. Wu, “An indoor hybrid wifi-vlc internet access system,” in medical research,” Malawi Medical Journal, vol. 24, no. 3, pp. 69–71,
in 2014 IEEE 11th International Conference on Mobile Ad Hoc and 2012.
Sensor Systems, Oct 2014, pp. 569–574.