Conference PaperPDF Available

Analysis and Optimization of the Network Throughput in IEEE 802.15.13 based Visible Light Communication Networks

Authors:

Abstract

In line with the growing interest on visible light communication (VLC), IEEE has initiated standardization efforts on this emerging technology. In this work, we consider IEEE 802.15.13 Optical Wireless Personal Area Networks (OWPAN) standard draft. The underlying MAC protocol uses contention free and contention access periods. For a standard-compliant VLC network, we analyze the network load and propose an algorithm to improve the network throughput by proper selection of period lengths. Our suggested algorithm improves the network performance by at least 5% in the case of variable network traffic up to 15 active users.
Analysis and Optimization of the Network Throughput in
IEEE 802.15.13 based Visible Light Communication Networks
Yusuf Bulbul,, Mohammed Elamassie, Tuncer Baykas, Murat Uysal,
Department of Electrical and Electronics Engineering, ¨
Ozyeˇ
gin University, Istanbul, Turkey.
Hyperion Technologies, Istanbul, Turkey.
Department of Electrical and Electronics Engineering, Kadir Has University, Istanbul, Turkey.
Abstract—In line with the growing interest on visible light
communication (VLC), IEEE has initiated standardization efforts
on this emerging technology. In this work, we consider IEEE
802.15.13 Optical Wireless Personal Area Networks (OWPAN)
standard draft. The underlying MAC protocol uses contention
free and contention access periods. For a standard-compliant
VLC network, we analyze the network load and propose an
algorithm to improve the network throughput by proper selection
of period lengths. Our suggested algorithm improves the network
performance by at least 5% in the case of variable network traffic
up to 15 active users.
Index Terms—802.15.13, MAC, OWPAN, slotted Aloha, GTS
Reservation, Network Throughput Optimization
I. INTRODUCTION
The spectrum crunch in traditional indoor radio networks
has motivated the introduction of visible light communi-
cation (VLC) as an alternative wireless access solution.
Consequently, VLC equipment vendors and research institu-
tions have initiated relevant standardization efforts. The IEEE
802.15.13 Multi-Gigabit/s Optical Wireless Communications
Task Group (TG13) is working on Physical (PHY) and Media
Access Control (MAC) layer designs, which will deliver data
rates up to 10 Gbit/s. The first draft of TG13 was finalized
in 2019. As of May 2021, the group is in the final phase of
standard preparation.
802.15.13 MAC protocol uses ”slotted Aloha” and ”reserved
dynamic time slots” protocols simultaneously to provide the
required quality-of-service (QoS) by resource reservation and
achieve high resource utilization with multiplexing gain during
the contention periods. However, reservation operation can
significantly affect the behavior of the network. How to split
time between reservation periods with reserved time slots and
how to determine the number of reservation slots are crucial
tasks of 802.15.13 MAC Network Coordinator.
The IEEE 802.15.13 MAC superframe consists of a con-
tention access period (CAP) and a contention free period
(CFP) [1]. In the CAP, each node transmits its packets to the
coordinator using slotted Aloha protocol through contention
with the other nodes. On the other hand, in the CFP, each
node transmits its packets to any device in the network by
using the guaranteed time slots (GTSs) dedicated to itself. In
order to transmit packets in the CFP, a node transmits a request
packet to the coordinator during the CAP. If the coordinator
successfully receives the request packet, it allocates the GTS
to the node according to its buffer or queue. Therefore, the
performance of the CAP and that of the CFP are related to
each other.
For the network coordinator, defining CAP duration is vital
for the network data rate optimization. The CAP duration
affects package drop rate and total throughput. The coordinator
has to define the total number of CAP slots according to user
numbers in the network. And then, it can decide the maximum
CFP slots which can be assigned for each user according to
the queue length of devices. High CFP slot numbers also
cause more buffer and latency but more throughput. Thus,
CFP duration is also vital for overall performance. In this
paper, we investigate how those period lengths and time slot
numbers should be determined and optimized by the coordi-
nator. According to a series of simulations, we further propose
a heuristic algorithm to improve total network throughput
efficiency.
In the literature, various techniques have been used to
analyze and improve the performance of contention-based
reservation protocols. D. Lee, et. al. proposed the use of
Markov chains to model the distributed coordination func-
tion (DCF) for saturated stations. [2]. Another method depends
on the users in the network repeatedly updating information
in order to reflect any changes in their data rates [3]. A
similar approach is the modified packet transmission in the
TDMA period to reduce transmission overhead [4]. Mean
value analysis, which evaluates the average value of system
variables, such as station transmission probability, collision
probability, frame service time, can be used as well [5], [6],
[7], [8]. The method in [5] evaluates the energy efficiency
through computer-based simulations. In [6], the goodput1and
delay is improved for high priority data by dividing the GTS
length from historical packet size and adjusting values. In
[7] and [8], the average frame service time and throughput
is investigated to improve network efficiency based on Mean
Value Analysis. Authors in [9] and [10], studied the delay
performance of DRP (Distributed Reservation Protocol) with
arbitrary reservation patterns in superframes. In [11], different
transmission timing with a priority of packets is proposed
to improve throughput. Most of the existing works however
build upon simplified assumptions and do not include crucial
parameters in practical deployment such as package drop,
buffer, latency, and desired data rate for devices in the network
1Goodput is the number of useful information bits that is delivered by the
network to a particular destination per unit of time.
Beacon Contention Free Period (CFP)Contantion Access Period (CAP)
1 ... 100 102 103 104 105 106 107 ... 1000 1001 1002 1003 1004 1005 1006 ...
Number of Superframe Slots
Superframe Slot Duration
Fig. 1: Superframe Structure.
which motivates our study.
The remainder of this paper is organized as follows. In
Section II, we provide the overview of IEEE 802.15.13 MAC
Layer. In Section III, we describe the simulation scenarios
and simulator software setup . In Section IV, we present the
numerical results. In Section V, we introduce the proposed
algorithm and simulation results. Finally, We conclude this
paper in Section VI.
II. TH E OVERVIE W OF IEEE 802.15.13 MAC LAYER
In the 802.15.13 MAC Layer, two channel access algorithms
exist in the standard, namely the ”beacon-enabled channel
access” and ”non-beacon enabled channel access”. If OWPAN
runs in beacon-enabled channel access mode, channel time
is subdivided into subsequent superframes. Each superframe
comprises three major parts: a beacon transmission period, an
optional CAP, and the CFP. In the CAP, devices may access
the channel randomly utilizing slotted Aloha. Random channel
access in the CAP is only allowed for specific procedures and
frame types. All other frame transmissions take place within
the CFP. The CFP consists of reserved resources, called GTSs,
which are assigned to each device for a given superframe. The
coordinator coordinates and announces GTS allocations.
A superframe consists of superframe slots. The total number
of slots is a variable determined by the OWPAN coordinator
and announced to the devices in the beacon frame. The
maximum number of slots within a superframe is 65535. Each
superframe slot has a specific duration. Hence, superframe
slots and their respective durations determine the duration of
each superframe. The superframe structure is illustrated in
Fig. 1. The MAC protocol makes use of integer numbers of
superframe slots to specify duration within the superframe.
That can be the duration of the CAP, CAP slots, GTS, and
other sub-parts of the superframe. Each OWPAN coordinator
defines the superframe structure for its coordinated OWPAN.
Consecutive superframes of an OWPAN do not necessarily
have to be adjacent but may have channel time between them
that is not used by the OWPAN.
In a superframe, three consecutive slot groups are used for
the beacon transmission, the CAP and CFP respectively. The
number of superframe slots reserved for the beacon transmis-
sion depends on the length of the beacon frame. The length of
the CAP is determined by the OWPAN coordinator, and may
change from superframe to superframe. The remaining slots
in the superframes are used for the CFP, and can be used for
frame transmissions between the devices and the coordinator.
When a device does not have sufficient GTS time for its
transmissions, it may perform the resource request procedure.
For example, this may be the case after the device connectivity
was interrupted and the coordinator stopped allocating GTSs.
In that case, the device may transmit a feedback control frame
in the CAP to signal the requirement for additional GTS time.
Thus, the network can be maintained by the coordinator.
III. IEEE 802.15.13 DATA RATE S
In this section, we present the 802.15.13 MAC and physical
(PHY) layer data rates and underlying modulation and coding
schemes.
A. Physical Layer Data Rate
To calculate the data rates at MAC layer, we need to start our
analysis from the PHY layer. In 802.15.13 standard, Physical
Layer data rates depends on the OCR (Optical Clock Rate),
and modulation order according to the suitable SNR value
of the user. Optical Clock Rate is the frequency at which the
data is clocked out to the optical source. The standard supports
multiple Optical Clock Rates (OCRs) to accommodate a wide
variety of optical sources and receivers. In Table 1, OCR20 (20
MHz Optical Clock Rate) data rates are shown. Modulations
of BPSK, QPSK, 16-QAM and 64-QAM are supported along
with 1/2 and 3/4-rate convolutional coding.
Let NOF DMidenote the number of OFDM symbols for
the user ith, with Idenoting the total number of users.
Furthermore, let Ndenote the number of sub carrier in each
OFDM symbol, and Ncp denote the length of cyclic prefix.
The length of overall packet excluding both synchronization
header (SHR) and Physical Layer header (PHR) for the ith
user is given as NDOW Ni=NOF DMi(Ncp +N). After pulse
shaping, the total number of samples for the user in the frame
is given by NUPi=NDO W NiUwhere NUPidenotes the up-
sampling factor.
TABLE I: Supported OCR20 Physical Layer Data Rate and
Throughput in Mb/s.
Mod. Symbol OCR20 Bits Throughput
BPSK CC(1/2) 2046 6 Mb/s 24480 4,34 Mb/s
BPSK CC(3/4) 2046 9 Mb/s 36756 6,66 Mb/s
QPSK CC(1/2) 2046 12 Mb/s 49032 8,89 Mb/s
QPSK CC(3/4) 2046 18 Mb/s 73584 13,34 Mb/s
16-QAM CC(1/2) 2046 24 Mb/s 98136 17,78 Mb/s
16-QAM CC(3/4) 2046 36 Mb/s 147240 26,68 Mb/s
64-QAM CC(1/2) 2046 48 Mb/s 196344 35,58 Mb/s
64-QAM CC(3/4) 2046 54 Mb/s 220896 40,02 Mb/s
Out of Nsub carriers, NDdata sub carriers are loaded with
data. Each data sub carrier for the ith user is loaded with ki=
log2(Mi)with Midenoting the selected modulation order for
the ith user. Let BTi denote the total number of transmitted
data bits in a frame for the ith user. The number of actual
transmitted bits due to the punctured convolutional coding for
the ith user is then given as Bi=BTi CCiwhere CCiis the
code rate. The required number of OFDM symbols for the ith
user is therefore obtained as NOF DMi=dBi/NDlog2(Mi)e.
Devices may require different number of symbols (NUPi)
which depends on the selected modulation order of the ith
user and the selected code rate. This in fact requires assigning
different time slots for different users. Let TSdenote the
sampling time. Then the required time for transmitting one
frame for the ith user is given as
Ti= (dBi/(NDlog2(Mi))e) (N+Ncp)U Ts·(1)
Let DRiand FSidenote, respectively, the number of data
bits per second (i.e., data rate) and number of frames per
second for the ith user. Noting DR=BTFS, the required
number of frames per second for the ith user can be calculated
as
FSi=DRi/BT i·(2)
Further noting FSi= 1/Tiand utilizing (1) and (2), we
have TS= min(TS1,TS2, ...TSi)where TSi,i= 1,2,3...I.
TSi=1
FSidBi(NDlog2Mi)e(N+Ncp)U·(3)
B. MAC Layer Data Rates
The MAC Layer will receive samples each TSsecond.
Based on quantization level of L(i.e., fixed point conversion),
MAC layer will handle each Tssecond, log2(L)bits. The
resulting MAC layer data rates depend on channel access
mechanism and software implementation. There are three time
periods as Beacon Period, CAP and CFP. The users only can
send data in the assigned part of CFP for them. The total
superframe duration is given by TTotal =TB+TCAP +TCF P
where TBdenotes beacon period duration, TCAP denotes CAP
duration and TCF P stands for CFP duration.
Every superframe is divided into time periods and
also time periods are divided into slots. Every period
duration can be therefore described in terms of time
slots. Let SB, SCAP , SC F P denote the slot numbers and
TSB, TSCAP , TSCF P denote the slot lengths. Beacon period
duration, CAP duration and CFP duration are respectively
given by
TB=SBTSB,(4)
TCAP =SC AP TSCAP ,(5)
TCF P =
N
X
1
SCF P TSCF Pi·(6)
The slot lengths are defined by the network coordinator
and it may be redefined in every superframe. The Beacon
Period, CFP and CAP slot lengths are calculated according
to transmission delay of the users. At least one frame must
be able to send in the one slot length by any user in the
network. Therefore, coordinator should choose the unit slot
length of periods, according to the user’s Physical Layer
data rate and throughput in the network. Slot Length of the
periods depend on the transmission delays . Total delay of
Beacon Period, CAP and CFP slots are the sum of MAC and
PHY Layer delay. MAC Layer delay is caused by software
implementation delay and PHY Layer delay is caused by
the configuration and channel. Beacon Period, CAP and CFP
delays are defined as follows: TSB=TMB+TPBwhere
TMB, TPBare the MAC and PHY Beacon Package Transmis-
sion Delays, TSCAP =TMCAP +TPC AP where TMCAP , TPC AP
are the MAC and PHY CAP Package Transmission Delays.
TSCF P =TMCF P +TPCF P where TMCF P , TPCFP are the MAC
and PHY CFP Data Package Transmission Delays.
The MAC protocol data unit (MPDU) at the output of the
MAC sublayer passes through the PHY Layer and becomes
the PHY service data unit (PSDU) at the output of the
PHY Layer. The PSDU is prefixed with a synchronization
header (SHR), containing the preamble sequence field; and
a PHY header (PHR), which, among other things, contains
the length of the PSDU in octets. The preamble sequence
enables the receiver to achieve synchronization. The SHR,
PHR, and PSDU together form the PHY frame or PHY
Layer data unit (PPDU). PHY Layer delay includes data
transmission PPDU header delay which includes SHR and
PHR parts, data transmission delay (PSDU) and short inter
frame space time duration, because of different package size
and supported data rates. The Beacon Period, CAP and CFP
have different transmission delays. These transmission delays
can be described as TDelay =Mi/DRiwhere Miis frame
size and DRiis the throughput of supported PHY Data Rate
of the ith user. The Physical Layer throughputs are presented
in Table 1.
The CFP Physical Layer delay includes TH,TSI F S ,TG,
which respectively denote header transmission delay, short
inter frame space duration and guard time. TR,TD,TMD ,C,
TSup are respectively CAP Response-Request Package Delay,
CFP Data Package Delay, Max Drift Time, Clock Accuracy
and Superframe Duration. Mathematically speaking, they are
defined by
TSB=TH+TB+TSI F S ,(7)
TPCAP = 2 (TH+TR+TSI F S ),(8)
TPCF P =TH+TD+TG,(9)
TG=2(TMD ),(10)
TMD =C/TSup ,(11)
TMD >=TS I F S ·(12)
Beacon Period, CAP and CFP delays could be different for
users in the network, because of different supported Physical
Layer data rates and throughput. Coordinator should consider
the Beacon Period, CAP and CFP delays of users for defining
the slot lengths. Any user in the network can send at least one
data package within one period slot. Therefore, Beacon Period
and CAP slot lengths should be defined according to the worst
transmission environment of any user in the network. Because,
every user which has the lowest supported physical data rate
must be able to use these periods for any operation in the
network. Also in CFP, every user has different transmission
delay. So, coordinator defines different CFP slot length for
every user. Every user can send data package which has
maximum MAC frame length in one superframe as much as
frames in the total assigned CFP slot. Thus, device buffer B,
CFP duration TCF P and the ideal MAC Data Rate Difor the
ith user are respectively given by
B(TB+TCAP +TC F P )Di·(13)
TCF P =
N
X
1
SCF P TSCF Pi·(14)
Di=SCF P Mi
TB+TCAP +TC F P
·(15)
IV. SIMULATION SCENARIOS AND NUMERICAL RES ULTS
In this section, we describe the simulation scenarios and
present the simulation results. The data rate in (15) is the
ideal MAC data rate, which means there are no collisions in
the CAP period for CFP slot allocation. Therefore, it is the
maximum data rate, which will be reached by the user. If
collisions are considered, the data rate decreases and latency
increases.
A. Scenario I
In Scenario I, we select users which have different supported
physical data rates as 9, 12, 18 Mbit/s with OCR20 in Table
1. The ideal data rate and transmission delays according to
Section III are used as simulation inputs. In the simulation,
every user generates data to send in the network as much as
ideal data rate input. But, they can not send all generated data.
This is due to the fact that the ideal data rate is calculated based
on the assumption of no collision. Therefore, collisions cause
extra latency, much package drop and less user data rate.
Fig. 2 demonstrates how the CAP slot lengths and CAP slot
numbers affect the network data rate, latency and collision,
when the maximum 8 number of CFP slot (i.e., the length of
1.8 ms) can be assigned per user, while beacon slot length
is 0.11 ms. Results quantify the performance in data rate and
latency associated with choosing the suitable CAP slot number.
For example, in order to achieve a highest data rate and lowest
latency for this scenario, 4 CAP slot is required and collision
ratio is 0.38 at that point as shown in Fig. 2. While the CAP
slot length is increasing from 0.11 ms to 1.9 ms, the network
gains more stable values in the latency and data rate. With
increasing of CAP slot number, the collision ratio decreases
exponentially and the extremum points becomes flatter in the
data rate and latency.
B. Scenario II
In this scenario, same users in Scenario I are selected. The
maximum CFP slot numbers which can be assigned per user
increases while the total CAP slot number and CAP slot width
are constant.
Fig. 3 shows how the CFP slot numbers affect the data
rate, latency and package drop ratio in the network, when the
maximum CFP slot length is 1.8 ms and beacon slot length
is 0.11 ms. Increasing the number of CFP slots increases the
latency linearly, but also increases the data rate exponentially.
For example, the data rate climbs to 2768.78 kb/s, 3292.63
kb/s, 3406.86 kb/s, 3472.08 kb/s, while latency increases as
6.3 ms, 11.3 ms, 16 ms, 20.3 ms for 2, 8, 14, 20 number of
CFP slots respectively. Thus, the package drop decreases with
increasing CFP slot number. For example, the package drops
are the 0.15, 0.074, 0.054, 0.041 for the 2, 8, 14, 20 number
of CFP slots respectively. Due to the increase in CFP duration
compared to CAP duration, the network begins to saturate in
data rate and packet drop as shown in Fig. 3.
C. Scenario III
In scenario III, the total number of CAP slot increases, CAP
Slot length is constant as 1.1 ms, and users have the same
supported Physical Layer data rate as 24 Mb/s. Longer CAP
slot lengths are selected to investigate the effect of CAP. In
this case, the beacon period is 0.11 ms length and CFP slots
are 0.9 ms length while the maximum number of CFP slot is
8 for per user.
Fig. 4 shows how the CAP slot numbers affect the data rate,
latency and collision ratio with different total active users in
the network. Increasing the number of users also increases the
CFP duration compared to the CAP duration. Therefore, the
network is more stable with a large number of active users.
As the number of active users decreases, the extreme points
in data rate and latency become clear. And, the slope of the
parabolic drop in the collision ratio increases.
V. PROPO SE D ALGORITHM
Slotted Aloha algorithm has no closed loop control mech-
anism to arrange the number of slots. If the network load
demand changes with time, the coordinator needs to estimate
the network demand for network optimization by changing
total slot length and number. To estimate the demand, the
commonly used statistical approach is the Markov Chain.
However, this approach includes many calculation iterations
which might be hard to implement on embedded systems.
To address this, we propose a simple yet computationally
efficient algorithm (see Fig.5) that enables the coordinator to
optimize the network according to load demand with a slotted
Aloha mechanism. In the proposed method, the coordinator
observes the user’s feedback in the CAP. It sets an observation
time and declares the length and beginning of the time in
the Beacon Frame. It collects all feedback of users until
observation time ends. The user requests GTS and sends also
the amount of total request and retry number that occurred in
the observation time. In this way, the coordinator calculates
0 5 10 15 20
Total Number of CAP Slots
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Collision Ratio
0.11 ms CAP Slot
0.35 ms CAP Slot
0.55 ms CAP Slot
1.10 ms CAP Slot
1.90 ms CAP Slot
(a)
0 5 10 15 20
Total Number of CAP Slots
1500
2000
2500
3000
3500
Data Rate (Kb/s)
0.11 ms CAP Slot
0.35 ms CAP Slot
0.55 ms CAP Slot
1.10 ms CAP Slot
1.90 ms CAP Slot
(b)
0 5 10 15 20
Total Number of CAP Slots
0
10
20
30
40
50
60
Latency (ms)
1.90 ms CAP Slot
1.10 ms CAP Slot
0.55 ms CAP Slot
0.35 ms CAP Slot
0.11 ms CAP Slot
(c)
Fig. 2: Simulation Results of Scenario I.
0 5 10 15 20
Total Number Of CFP Slots Per User
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
Package Drop Ratio
0.11 ms x 5 CAP Slot
(a)
0 5 10 15 20
Total Number of CFP Slots Per User
2400
2600
2800
3000
3200
3400
3600
Data Rate (Kb/s)
0.11 x 5 CAP Slots
(b)
0 5 10 15 20
Total Number of CFP Slots Per User
4
6
8
10
12
14
16
18
20
22
Latency (ms)
0.11 ms x 5 CAP Slot
(c)
Fig. 3: Simulation Results of Scenario II.
0 5 10 15 20 25
Total Number of CAP Slots
0
0.2
0.4
0.6
0.8
1
Collision Ratio
15 Users
11 Users
8 Users
5 Users
2 Users
(a)
0 5 10 15 20 25
Total Number of CAP Slots
0
1000
2000
3000
4000
5000
6000
Data Rate (Kb/s)
2 Users
5 Users
8 Users
11 Users
15 Users
(b)
0 5 10 15 20 25
Total Number of CAP Slots
0
100
200
300
400
500
600
Latency (ms)
15 Users
11 Users
8 Users
5 Users
2 Users
(c)
Fig. 4: Simulation Results of Scenario III.
the collision ratio, estimates the load demand according to
collision ratio and arranges a suitable total number of CAP
slot for the network.
To evaluate the performance of proposed algorithm, simu-
lations are carried out following Scenario III. Fig. 6 shows
the maximum throughput according to specific collision prob-
ability range. The collision ratio is controlled by arranging
total CAP slot number according to the user’s collision ratio
feedback in the proposed algorithm. The maximum throughput
is observed at the specific collision ratio. Considering the
simulations, the suitable collision ratio is between 0.4 and 0.6
to obtain maximum throughput. If the collision ratio remains
at the suitable values, the network reaches the maximum
throughput. The performance and time responsiveness of the
algorithm can be changed according to the applied control
approach. The PID Control also can be implemented for more
efficient time responsivity performance.
Fig. 7 shows the total throughput of the network accord-
ing to total CAP slot and total active user number. While
total active user number change with time in the network,
set new
Observation time
(OT) and take ref
max collision
ratio (CR) of the
users
if CR > 0.6
if CR < 0.4
Yes
Yes
No
No
if OT == 0
Yes No
OT = Ob.Time;
CR =
max(CR_Array)
OT = OT - 1;
pushback
(CR_Array)
Decrease the Observation
Time (OT) with each super frame
collect the collision
ratio array (CR_ Array)
on CAP
Wait for Next
Superframe
CAP Slot Number ++
CAP Slot Number --
Increase Cap
Slot
Decrease Cap
Slot
USER
COORDINATOR
BEACON
(Start Observation Time)
CAP GTS Request
(Failed Request / Total Try Number)
Fig. 5: Proposed Algorithm.
0 5 10 15
Total Active User
0.8
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2
Total Network Throughput (kbyte)
10 4
Max Throughput
Proposed Algorithm Throughput
Total Throughput at 0.5 - 1.0 Collision R.
Total Throughput at 0.0 - 0.3 Collision R.
Fig. 6: Total Network Throughput Comparison.
0 5 10 15 20 25
Total Number of CAP Slots
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
Total Network Throughput (kbyte)
10 4
2 Users
5 Users
8 Users
15 Users
Fig. 7: Total Network Throughput.
total throughput and suitable total CAP slot number change
accordingly. When the total network throughput in different
collision probabilities are compared; total network throughput
with probability range between 0.4 and 0.6 is at least 5%
higher than the other probability ranges as shown in Fig. 7.
This indicates at least 5% improvement. This value changes
according to individual conditions. In cases wheres the dif-
ference between the number of active users changes is high,
further improvement can be obtained.
VI. CONCLUSION
In this work, we analyzed the effect of superframe design
on the throughput of IEEE 802.15.13-based VLC networks.
Through simulations, we analyzed the effect of the total
number of CAP slots and active user number. It is shown that
for a specific CAP slot duration and collision ratio, there is
an optimum number of CAP slots either to increase the data
rate or to reduce latency. For short CAP duration or long CFP
duration, the number of CAP slots has a limited effect. If the
CAP duration is the same or larger than CFP duration, the
coordinator should select the number of CAP slots carefully.
Our simulations show that the effect of collision ratio to data
rate and latency is high only if the number of active users is
low. If the number of active users and total number of CAP
Slots is higher, both latency and data rate are stable. We further
investigate the effect of the number of CFP slots. The data rate
increases logarithmically, whereas latency increases linearly
with the increase of the total number of CFP slots.
REFERENCES
[1] https://www.ieee802.org/15/pub/TG13.html, [Accessed: May 7, 2021].
[2] D. Lee, H. Roh, B. Kim, and J. Lee, “Performance analysis of the
IEEE 802.15.4 MAC protocol,” in International Conference on ICT
Convergence (ICTC), 2013, pp. 398–401.
[3] D. A. Dyson and Z. J. Haas, “A dynamic packet reservation multiple
access scheme for wireless ATM,” in 97 MILCOM 97 Proceedings,
vol. 2, 1997, pp. 687–693 vol.2.
[4] V. Nguyen, T. Khoa, T. Oo, N. Tran, C. S. Hong, and E.-n. Huh, “Time
slot utilization for efficient multi-channel mac protocol in VANETs,”
Sensors, vol. 18, p. 3028, 09 2018.
[5] F. Vazquez-Gallego, J. Alonso-Zarate, A. M. Mandalari, O. Briante,
A. Molinaro, and G. Ruggeri, “Performance evaluation of reservation
frame Slotted-ALOHA for data collection M2M networks,” in European
Wireless 2014; 20th European Wireless Conference, 2014, pp. 1–6.
[6] B. Lee, H. Wu, and N. Yu, “A priority based algorithm for adaptive
superframe adjustment and GTS allocation (PASAGA) in IEEE 802.15.4
LR-WAN,” in 2018 IEEE International Conference on Applied System
Invention (ICASI), 2018, pp. 318–320.
[7] R. Zhang, L. Cai, and J. Pan, “Performance analysis of reservation
and contention-based hybrid MAC for wireless networks,” in IEEE
International Conference on Communications, 2010, pp. 1–5.
[8] X. Ling, K.-H. Liu, Y. Cheng, X. S. Shen, and J. W. Mark, “A novel
performance model for distributed prioritized MAC protocols,” in IEEE
GLOBECOM 2007 - IEEE Global Telecommunications Conference,
2007, pp. 4692–4696.
[9] K. Liu, X. Shen, R. Zhang, and L. Cai, “Performance analysis of
distributed reservation protocol for UWB-based WPAN,” IEEE IEEE
Trans. Veh. Technol., vol. 58, no. 2, pp. 902–913, 2009.
[10] R. Zhang and L. Cai, “Joint AMC and packet fragmentation for error
control over fading channels,IEEE Trans. Veh. Technol., vol. 59, no. 6,
pp. 3070–3080, 2010.
[11] Y. Fujiura, C. Sugimoto, and R. Kohno, “BAN MAC protocol for
optimization the ratio between CAP and CFP,” in 6th International
Symposium on Medical Information and Communication Technology
(ISMICT), 2012, pp. 1–4.
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
In vehicular ad hoc networks (VANETs), many schemes for a multi-channel media access control (MAC) protocol have been proposed to adapt to dynamically changing vehicle traffic conditions and deliver both safety and non-safety packets. One such scheme is to employ both time-division multiple access (TDMA) and carrier-sense multiple access (CSMA) schemes (called a hybrid TDMA/CSMA scheme) in the control channel (CCH) interval. The scheme can adjust the length of the TDMA period depending on traffic conditions. In this paper, we propose a modified packet transmitted in the TDMA period to reduce transmission overhead under a hybrid TDMA/CSMA multi-channel MAC protocol. Simulation results show that a MAC protocol with a modified packet supports an efficient packet delivery ratio of control packets in the CCH. In addition, we analyze the hybrid TDMA/CSMA multi-channel MAC protocol with the modified packet under saturated throughput conditions on the service channels (SCHs). The analysis results show that the number of neighbors has little effect on the establishment of the number of time slots in TDMA periods and on SCHs under saturated throughput conditions.
Conference Paper
Full-text available
In this paper, we consider a Machine-to-Machine (M2M) wireless network composed of a group of devices which duty cycle to save energy. These devices operate in low-power sleeping mode for most of the time and, periodically, they wake-up to listen to a poll packet transmitted by a data collector. Upon this broadcast poll, all devices try to get access to the uplink channel to transmit a burst of data packets. Therefore, the idle network is suddenly set into saturation conditions when all devices wake up and attempt to get access to the channel simultaneously. The Medium Access Control (MAC) protocol used to coordinate these transmissions has a strong influence on the energy efficiency of the network, and thus the lifetime of the devices. Frame Slotted ALOHA (FSA) has been identified in the literature as a simple yet efficient MAC protocol for such kind of communications. However, when the devices have to transmit more than one data packet per channel invocation, the Reservation Frame Slotted-ALOHA (RFSA) may be more efficient, since it guarantees the collision-free transmission of data for a device once it succeeds for the first time. Existing analyzes of both FSA and RFSA are valid for steady traffic conditions and not for abrupt idle-tosaturation traffic patterns. Motivated by this fact, in this paper we evaluate the energy efficiency of RFSA through computerbased simulations to show its better performance compared to FSA. Results show that RFSA can attain up to 48% energy gains compared to FSA, thus extending the lifetime of data-collection M2M networks.1
Article
Full-text available
In this paper, we propose an analytical model for the distributed reservation protocol (DRP), which is defined in the WiMedia specification for ultra-wideband (UWB)-based wireless personal area networks (WPANs). We model the tagged user as a discrete-time queue with vacations, which captures the joint behavior of a queue length variation and a time-varying UWB channel due to shadowing under a given reservation pattern. Furthermore, we consider two reservation methods: hard reservation and soft reservation. With the hard reservation, a time slot is exclusively used by the owner, whereas the unused time slots can be accessed by other users using the soft reservation. Closed-form expressions of important performance metrics such as the mean service time, the waiting time, and the throughput are derived. Through numerical results, we validate the accuracy of the proposed analytical model and investigate the interaction between the DRP and various system parameters. This paper should provide insights into the performance of the DRP and useful guidelines to further improve the protocol to support isochronous applications with a tight delay requirement in UWB-based WPANs.
Conference Paper
Full-text available
Distributed prioritized channel access mechanisms have been adopted by the IEEE 802.11e enhanced distributed channel access (EDCA) and the Multiband OFDM Alliance prioritized channel access (PCA) to support service differentiation. In this paper, we propose a novel analytical model for performance study of such mechanisms. The proposed model gives the average frame service time first and then the per station and network normalized throughput, which makes it applicable to both saturated and unsaturated stations. Furthermore, the model is especially helpful in understanding the different effects of the same prioritizing mechanisms in saturated and unsaturated conditions. To the best of our knowledge, there is no similar work reported in the open literature. The accuracy of the analytical model is demonstrated by extensive simulation.
Article
Full-text available
Error control is critical for wireless networks to combat channel fading and ensure efficient resource utilization. Adaptive modulation and coding (AMC) in the physical (PHY) layer and packet fragmentation and automatic repeat request (ARQ) in the link layer are widely used error-control mechanisms. However, how to jointly optimize them in both layers for high-rate wireless networks is still open. In this paper, using the WiMedia ultrawideband (UWB) networks as an example, we first develop a general analytical framework to quantify the link delay and loss performance considering the channel fading, the joint error-control mechanisms, and the arbitrary reservation-based media access control (MAC) protocol. Second, we introduce a cross-layer design to optimize the PHY-layer AMC and the link-layer packet fragmentation and propose a joint-adaptation mechanism that is simple to implement and has near-optimal performance. Numerical results reveal that fragmentation has a greater impact than AMC on the delay and loss performance for marginal links and that the proposed joint-adaptation strategy is efficient for high-rate wireless networks.
Conference Paper
Full-text available
Dynamic packet reservation multiple access (DPRMA) is a medium access control protocol for wireless multimedia applications. It allows the integration of both constant bit rate and variable bit rate traffic through a single access control mechanism that permits users to specify their bandwidth requirements. Users are allowed to repeatedly update this information in order to reflect any changes in their data rates. A base station analyzes the mobiles' requests, determines which can be accommodated, and conveys the resulting bandwidth assignments to the users. The ability of a mobile to initially reserve a portion of the channel capacity and to then dynamically alter this reservation is a primary feature of the system. In DPRMA, an attempt is made to match the capacity assigned to the user with the user generation rate. Furthermore, this capacity can be allocated using fractional or multiple slot assignments. The scheme is shown to provide improved performance over a system with a modified version of the packet reservation multiple access (PRMA) scheme
Conference Paper
In this paper, we analyze the throughput of the IEEE 802.15.4 medium access control (MAC) protocol with considering both contention access period (CAP) and contention free period (CFP). We first analyze the CSMA/CA algorithm in the CAP using a Markov chain considering the guaranteed time slot (GTS) request packets for the CFP. And we provide an analytical model of the GTS allocation algorithm in the CFP using a Markov chain and the results of the CAP. Numerical results show that the ratio of the length of the CAP to that of the CFP affects the throughput of the system.
Article
Recently, a micro or nano size wireless telecommunications device such as a wireless implant capsule endoscope has been researched and developed. These wearable and implanted wireless devices are quite useful to remotely monitor various vital information, for example ECG, EEG, etc. And we use manipulators for medical treatment. In order to connect several medical sensors and actuators in and on human body, wireless body area network (BAN) has been developed. However, wireless radio communications in BAN may occur a large variance of transmission delay due to complicated radio propagation in and on body. In addition, a lot of packets are coming with contention against each other. The large variance of delay may cause a serious problem in remote medical treatment. Therefore medical packets must be guaranteed about the delay and the authenticity. In particular, emergency access packets should be guaranteed a largest delay within a permissible range. On another front, communication efficiency of non-medical packets has to be guaranteed as much as possible. A MAC protocol using hybrid structure of super frame in IEEE802.15.6 has been proposed. However, a ratio between CFP and CAP in a super frame has not been determined in the standard yet. This paper investigates how to determine the ratio according to traffic of medical and non-medical packets. After calculating the best length of CFP by the medical traffic, non-medical information is allocated in CFP as much as possible. In so doing, the total throughput is improved after securing the delay of medical information. Moreover as for traffic monitoring, we proposed the slotted aloha plotocpl for using the different transmition timing as to the priority of packets. The utility of the method was evaluated by computer simulation. On the proposed method, throughput of medical packets is nearly equal to being allocated CFP to upper limit, and non-medical throughput is better than fixed slots method. Therefore, total throughput is improved. And the accuracy of traffic estimation is good performance for using proposed determining the ratio of CAP and CFP slots.
Conference Paper
Hybrid media access control (MAC) protocols use reservation and contention-based approaches simultaneously, so they can provide satisfactory quality-of-service to multimedia applications by resource reservation, and achieve high resource utilization with multiplexing gain during the contention periods. However, reservation can significantly affect the behavior of the contention-based access. How to split channel time between reservation periods and contention periods and how to adjust the contention scheme for hybrid MAC are important, open issues. In this paper, an analytical model for the hybrid MAC with saturated traffic is first proposed and then extended to the unsaturated traffic case. Based on the mean value analysis, the proposed models give the average frame service time and throughput for the contention-based MAC with the presence of reserved channel periods. They are also applicable to online admission control due to their low computational complexity.