2 Improvement of Duffing Oscillator’s Resistance to Narrowband Noise
Tải bản đầy đủ - 0trang
436
W. He et al.
“0x01” means networks management package.
“0x02” means remote command for Simple Management use
“0x03” means remote register read and write for register mapping mode
RES: reserved for future use
Length: length of payload to be transmitted. This length is only included the
payload.
Payload: Data need to be sent to remote site
CRC: Checksum of even BIP8 check.
In addition, we also develop a private Protocol of LLH (Low Level Shake Hands),
which is used to identify the device type include the bit rate and protocol. As
an example, if customer put one site device with OTU2(10.709 Gb/s), while the
peer site with in OTU1e (11.049 Gb/s). In this case, traﬃc will not be OK, and
in-band OSC channel can also not be built up. Low level shake hands (LLH) protocol is designed to solve this kind of mismatch problem. LLH is implemented
by periodically turn on/of long haul laser to generate a low speed optical pules
which carries the shake hands information. Frame of LLH: LLH frame is deﬁned
as Fig. 7. BYTE1 is LLH sync and LLH Command, BYTE2 is the bit negation
of BYTE1 to check if BYTE1 is correct.
SOF0 – SOF2: Start of Frame indicator. “010” means frame start
C/S: Command or Status indicator. “1” C Command; “0” C Status information
D0– D3: Data to transmit.
When D0 = “0”, following 3 bit deﬁnes the Rate mode of LH
“001” LH Rate is OTU2 (10.709 Gb/s)
“010” LH Rate is OTU1e (11.049 Gb/s)
“011” LH Rate is OTU2e (11.1 Gb/s)
“100” LH Rate is OTU2f
When D0 = “1”, following 3 bit deﬁned as other command or information
“0101100” is deﬁned as remote reset command.
Fig. 7. LLH frame
4
Remote Node Device
Remote node device is a small and only 1/4 size of hub device which consists of
Management card, dual-channel 10G card and power module as show in Fig. 5.
Dual-channel 10G card is also show by the Fig. 8, which is constructed by 4
Management of a Hub-Spoken Optical Transmission Network with P2MP
437
SFP+ module, a digital wrapper chip of VSC8492, FPGA, MCU, RAM, FLASH,
etc. Client 10G switch or router sends 10G Signal to SFP+ module. It is then
encapsulated into a 10G OTN frame signal by OTN Framer. 10G OTN framer,
which is designed to provide FEC function and in-band network OAM management capability. Signal is transparently mapped into 10G OTN frame with FEC.
After digital wrapper OTN frame process, signal will be sent to long haul SFP+
module for long distance transmission.
Fig. 8. Management card and dual-channel 10G card for remote device
The software architecture for management card is based on FreeRTOS running on 32 bit-MCU (STM32F103VET6) [5]. The structure of software is shown
as Fig. 9 which includes 4 layers. The lowest layer is Hardware layer. The upper
layer includes OS software which is designed based the on FreeRTOS (Real-time
operating system), software drivers includes Console driver, SFP+ driver, CAN
driver, FPGA driver, etc. The second layer including HWM (hardware monitor),
DB, alarm, etc. The top of the structure is CLI (command line interface) which
receives message including in-band message.
Fig. 9. Software structure based on MCU
After storing information in ping-pong RAM by FPGA, MCU can read the
information form buﬀer through verifying header continually. In order to do
438
W. He et al.
priority process with the receiving information. We set a interrupt generated by
EXTI and send to MCU. And the Fig. 10 is ﬂow chart about Inband information
process.
Fig. 10. Flow chart of Inband information process
Management card is designed based on a MCU of control devices which is
show by Fig. 8, which can control and manage the management card through
CAN bus by UART/USB/LAN interface. It consists of MCU, DDR, FLASH,
RAM, etc. In our design, management card can manage the 10G card through
LAN/Console/USB and provides the Oﬃce alarm functions.
5
Testing Result
P2MP experiment set up is illustrated as Fig. 4. We can use a PC to monitor
and manage multi links status from hub to each remote node. All the PM and
remote node device status can be monitored by the software from both the hub
node location and remote node location. The following Fig. 11 shows the testing
results for comparing.
In this experiment, we use console to monitor the device. The Hub equipment
can get each remote node equipment status information through digital wrapper
in-band management channel. It can be seen that the monitoring status of the
remote equipment by hub node is the same as the result which is got directly
from the remote node, which proves that P2MP method is implemented correctly
and can monitors the multi remote devices eﬃciently.
Management of a Hub-Spoken Optical Transmission Network with P2MP
439
Fig. 11. Local and remote device status
Acknowledgment. We have built an experimental set up for point to multi point
(P2MP) optical network by using a low cost 8x10G device of hub node and a small
remote node device, which are both developed by ourselves. We also define and develop
a software for the control plane and network management system. This P2MP network
management system is based on the fast FPGA processing and the OTN digital wrapper
(ITU-T G.709) to provide in-band communication. Our test results have shown that
we can simultaneously monitor up to 8 remote nodes status from the hub node through
the OTN digital wrapper in-band channel, which has the great advantage comparing
with using multiple OSC of the traditional optical network.
References
1. Liu, Q., Ouyang, C., Shi, C.: A small form factor and low cost 810 Gb/s optical
transmission system based on digital-wrapper technology. In: 2015 Eighth International Symposium on Computational Intelligence and Design, (ISCID), pp. 275–280.
IEEE Computer Society (2015)
2. Sul, D.M., Kim, S.M., Lee, J.H.: LSP merge in point to multipoint in-band OAM.
In: 12th International Conference on Optical Internet (COIN) (2014)
3. Zhen, C., Yingjin, L.: Hub-spoken major appliance logistics net work building and
practice. In: Computational and Information Sciences, (ICCIS), pp. 383–387 (2013)
4. ITU-T Recommendation G.709. Interface for the Optical Transport NetWorks
(2009)
5. Hou, S., Wu, S.: Design and realization of family intelligent interactive terminal
based on STM32. In: 9th International Conference on Fuzzy Systems and Knowledge
Discovery (2012)
Optimal Power Allocations for Full-Duplex
Enhanced Visible Light Communications
Liping Liang, Wenchi Cheng(&), and Hailin Zhang
State Key Laboratory of Integrated Services Networks,
Xidian University, Xi’an, China
lpliang@stu.xidian.edu.cn,
{wccheng,hlzhang}@xidian.edu.cn
Abstract. We consider the two nodes indoor full-duplex transmission over
bidirectional channels with imperfect self-interference cancellation in visible
light communications (VLCs). The light emitting diodes (LEDs) are used for
illumination and data transmission while the photo diodes (PDs) are used for
reception. In this paper, we ﬁrst formulate the sum-capacity maximization
problem for the full-duplex bidirectional VLCs. Then, we develop an optimal
power allocation scheme, which has the closed-form expression, to achieve the
maximum sum-capacity for two nodes indoor full-duplex bidirectional VLCs.
The obtained numerical results verify our developed optimal power allocation
scheme and show that the full-duplex transmission can signiﬁcantly increase the
sum-capacity than the traditional half-duplex transmission.
Keywords: Visible light communications Á Full-duplex Á Bidirectional
transmission Á Power allocations Á Self-interference Á Sum-capacity
1 Introduction
The optical wireless communication, which can efﬁciently overcome the spectrum
deﬁciency problem in radio communications, has attracted a lot of attention in past few
years. It has been shown that the optical wireless communications can be applied in
transdermal communications, transports, mobile medical body area networks, underwater communications, and indoor optical wireless communications [1–4]. Among
various optical wireless communications, the VLC rapidly emerges as an attractive
communication technology [1]. Visible light communication uses visible light spectrum in the range of wavelength 380–780 nm, which is license-free and does not
generate the electromagnetic interference to the existing radio frequency (RF) systems.
In VLC systems, white light emitting diodes (LEDs) are used for illumination as well
as the signal transmitter while photo diodes (PDs) are used as the receiver. The LEDs
This work was supported in part by the National Natural Science Foundation of China
(No. 61401330), the 111 Project of China (B08038), and the Natural Science Foundation of
Shaanxi Province (No. 2016JQ6027).
© ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2018
Q. Chen et al. (Eds.): ChinaCom 2016, Part I, LNICST 209, pp. 440–450, 2018.
DOI: 10.1007/978-3-319-66625-9_43
Optimal Power Allocations for Full-Duplex
441
can support cable free communications as high as several tens Mbps between the LEDs
and the PDs [5].
In half-duplex VLC systems, we need two time-slots to communicate between the
two nodes, resulting in low network throughput. To deal with this issue, we propose to
use full-duplex VLCs in this paper. The full-duplex wireless communication holds the
promise of double spectral efﬁciency, compared with the traditional half-duplex
wireless communication [6, 7], through simultaneous transmission and reception using
the same frequency band at the same time. The wireless full-duplex communication
mode was thought to be unfeasible because of the very large self-interference leaked
from the local transmitter to the local receiver, thus making it difﬁcult to extract the
useful receive signal transmitted from the other node. Therefore, it is critical to be able
to limit the power of the self-interference to implement full-duplex communications in
practice. Recently, a great number of research works have shown the possibility of
using full-duplex transmission in wireless communications using advanced
propagation-domain interference suppression, analog-domain interference cancelation,
and digital-domain interference cancelation [8–12]. Taking the self-interference as a
part of useful information, the authors of [13] applied the full-duplex relay VLC based
system to mitigate the self-interference. Some researchers proposed a self-adaptive
minimum contention window full-duplex MAC protocol to mitigate channel collisions
for VLCs [14]. To employ the full-duplex capacity efﬁciently, the paper [15] proposed
two contention protocols, named UALOHA and FD-CSMA. The authors of [16]
proved that the average electrical SNR per transmit antenna is considerably high in
indoor MIMO optical wireless communications.
In this paper, we consider the two nodes full-duplex bidirectional transmission
under imperfect self-interference cancelation in VLCs. We develop an optimal power
allocation scheme to maximize the sum-capacity of two nodes visible light full-duplex
bidirectional transmission for VLCs. For visible light full-duplex transmission in high
signal-to-interference-noise rate (SINR) region, we derive the closed-form expression
of optimal power allocation scheme.
The rest of this paper is organized as follows. Section 2 describes the system
model. Section 3 formulates sum-capacity maximization problem for the full-duplex
VLCs. Section 4 develops the optimal power allocation scheme in the high SINR
region for visible light full-duplex bidirectional transmission. Section 5 carries out the
numerical results to evaluate our developed optimal power allocation scheme and the
achieved sum-capacity for two nodes visible light full-duplex bidirectional transmission compared with the traditional half-duplex transmission. Section 6 concludes the
paper.
2 The System Model
In this paper, we consider the two nodes wireless full-duplex bidirectional visible light
communications, as illustrated in Fig. 1, where node A and node B transmit their data
to node B and node A, respectively. Each node is equipped with LEDs and PDs. The
LEDs are used for both illumination and transmission while the PDs are used for
reception. The optical modulation and demodulation schemes are set as intensity
442
L. Liang et al.
A
LED
yb
xa
h3
h1
σn2
B
⊕
LED
Transmitter
PD
σn2
⊕
αi
di
h2
h4
xb
ya
PD
βi
LED
PD
Receiver
Fig. 1. The two nodes full-duplex bidirectional
transmission in VLCs.
Fig. 2. The
geometry
transmitter-receiver pair.
of
a
modulation (IM) and direct detection (DD). The channels from node A to node B and
from node B to node A use the same frequency band at the same time.
The shot and thermal noise in the receiver are modeled as additive white Gaussian
noise (AWGN) and are added to the signal in the electrical domain. Thus, we have the
total noise variance as r2n ¼ r2shot þ r2thermal , where r2shot is the shot noise variance and
r2thermal is the thermal noise variance [17]. We assume that all total noise variances are
the same and uniformly denoted by r2n for mathematical simpliﬁcation. We assume that
the optical signal propagates from the transmitter to the receiver directly (Light of
Sight) and the reflections are ignored [16]. We denote by h21 and h22 the channel gains
from node A to node B and from node B to node A, respectively. We denote by h23 and
h24 the self-interference channel gains from local transmitter to local receiver at nodes A
and B, respectively. For one transmitter-receiver pair as shown in Fig. 2, the hi (i = 1,
2, 3, 4) is given by
(
hi ẳ
m ỵ 1ịA
cosm ai Þ cosc ðbi Þ
2pdi2
0
À Á
if ai ; bi 2 0; p2 ;
others,
ð1Þ
where m is the Lambertian emission order and given as follow (We assume that there is
only one Lambertian emitting mode for LED):
mẳ
ln 2
:
lncos w1=2 ị
2ị
The parameter w1=2 is the semi-angle of the LED at half-power. In Eq. (1), A is the
active area of PD and c is the ﬁled-of-view (FoV) coefﬁcient of the PD receiver [18].
For the ith channel (hi ), di , ai and bi (i = 1, 2, 3, 4) are the distance from LED to PD,
the irradiance angle at LED, and the incident angle at PD, respectively. We denote by
Pa and Pb the transmitter power at nodes A and B, respectively.
Optimal Power Allocations for Full-Duplex
443
To evaluate the performance of visible light full-duplex bidirectional transmission,
we employ the concept of capacity region [19]. The capacity region for two nodes
visible light full-duplex transmission is deﬁned as the set of rate pairs (RA , RB ) such that
the transmitters of node A and node B can reliably transmit data to the receivers of
node B and node A simultaneously, where RA and RB denote the channel rates from
node A to node B and from node B to node A, respectively. Because node A and node
B share the same bandwidth, there is a tradeoff between the reliable communication
rates RA and RB Á RA ðRB Þ increases as RB ðRA Þ decreases. The received signal at nodes A
and B, denoted by yb and ya , respectively, can be derived as follows:
&
pﬃﬃﬃﬃﬃ
pﬃﬃﬃﬃﬃ
2
ya ¼ p P
a h1 xa ỵ pP
b h4 xb ỵ rn2 ;
y b ẳ P b h 3 x a ỵ P a h 2 x b ỵ rn ;
3ị
where xa and xb represent the transmitted signal at nodes A and B, respectively, as
illustrated in Fig. 1. Then, we deﬁne the sum-capacity as the scalar performance
measure to evaluate the two nodes visible light full-duplex bidirectional transmission.
The sum-capacity for two nodes visible light full-duplex bidirectional transmission,
denoted by sb , is formulated as follows:
s b ẳ ya ỵ yb :
4ị
3 The Sum-Capacity Maximization Problem Formulation
For the wireless full-duplex bidirectional transmission, there still exists the residue
self-interference even the advanced self-interference mitigation schemes are applied
[8–12]. To characterize self-interference in visible light full-duplex bidirectional
transmission, this section introduces a parameter j (0
j
1), which is deﬁned as
the self-interference mitigation coefﬁcient. Then based on self-interference mitigation
coefﬁcient, we can build up the received residue self-interference model, denoted by
f ðPt ; h; xt Þ as follows:
f ðPt ; h; xt Þ ¼
pﬃﬃﬃﬃﬃﬃﬃ
jPt hxt ;
ð5Þ
where xt , h, and Pt represent the local transmit signal, the channel gain from the local
transmitter to the local receiver, and the transmit power, respectively. Based on Eq. (5),
it is clear that self-interference decreases with self-interference mitigation coefﬁcient j
decreases. When j = 1, self-interference mitigation technique does not work. When
j = 0, it indicates that self-interference between the local transmitter and the local
receiver can be completely canceled. In fact, j = 0 is impossible using the current
techniques.
We assume that all self-interference mitigation coefﬁcients are the same and uniformly denoted by j for mathematical simpliﬁcation. Then, the received signal at nodes
A and B, denoted by yb and ya , respectively, can be formulated as follows:
444
L. Liang et al.
&
pﬃﬃﬃﬃﬃ
pﬃﬃﬃﬃﬃ
pﬃﬃﬃﬃﬃ
pﬃﬃﬃﬃﬃﬃﬃﬃ
2
2
Ya ẳ pP
a h1 xa ỵ f pP
b ; h4 ; xb ị ỵ rn ẳ pP
a h1 xa ỵ pjP
bh4 xb ỵ rn2 ;
2
Yb ẳ Pb h2 xb ỵ f Pa ; h3 ; xa ị ỵ rn ẳ Pb h2 xb ỵ jPa h3 xa ỵ rn :
ð6Þ
We denote by SINRa and SINRb the SINRs of the received signal Ya and Yb . Then,
based on the analyses above, SINRa and SINRb can be expressed as follows:
8
2
< SINRa ẳ 2 Pa h1 2 ;
r ỵ jPb h
n
4
: SINRb ẳ 2 Pb h2 2 :
r ỵ jPa h
2
n
7ị
3
The capacity from node A to node B and from node B to node A, denoted by
Cab ðPa ; Pb Þ and Cba ðPa ; Pb Þ, respectively, can be derived as follows:
&
Cab ẳ log1 ỵ SINRa ị;
Cba ẳ log1 ỵ SINRb Þ:
ð8Þ
Thus, we can derive the achievable sum-capacity of two nodes optical wireless
full-duplex bidirectional transmission, denoted by CB ðPa ; Pb Þ, as follows:
CB ðPa ; Pb Þ ¼ Cab ðPa ; Pb ị ỵ Cba Pa ; Pb ị ẳ log1 ỵ SINRa ị ỵ log1 ỵ SINRb ị:
9ị
Our goal aims at allocating the transmit power to achieve the maximum
sum-capacity for two nodes optical wireless full-duplex bidirectional transmission.
Then, we can formulate the sum-capacity maximization problem, denoted by P1, as
follows:
P1 : arg max fCB ðPa ; Pb Þg
ð10Þ
ð1Þ Pa [ 0; Pb [ 0;
11ị
Pa ;Pb ị
s:t: :
2ị Pa ỵ Pb
P;
12ị
where P represents the average power constraint. Observing CB ðPa ; Pb Þ speciﬁed in
Eq. (9), we know that the capacity CB ðPa ; Pb Þ is a nonconcave function over the space
spanned by ðPa ; Pb Þ. Thus, P1 is not a strictly convex optimization problem. To obtain
the maximum capacity CB ðPa ; Pb Þ, it is desirable that P1 is a strictly convex optimization problem. In the following section, we convert P1 to convex optimization
problem and develop the optimal power allocation scheme to obtain the global optimal
solution for P1.
4 The Optimal Power Allocation Scheme in the High SINR
Region
In the high SINR region, CB ðPa ; Pb Þ can be re-written as follows:
Optimal Power Allocations for Full-Duplex
Pa h21
Pb h22
CB ðPa ; Pb ị ẳ log 2
ỵ log 2
rn ỵ jPb h24
rn þ jPa h23
Pa Pb h21 h22
¼ log 4
:
rn þ r2n jPa h23 ỵ r2n jPb h24 ỵ j2 Pa Pb h23 h24
445
ð13Þ
Because log ðÞ is a monotonically increasing function as the independent variables
increase, we can convert the problem P1 to a new problem P2, which is an equivalent
problem as P1 and can be expressed as follows:
P2 :
max
ðPa ;Pb Þ
Pa Pb h21 h22
r4n ỵ r2n jPa h23 ỵ r2n jPb h24 þ j2 Pa Pb h23 h24
ð14Þ
subject to the constraints given in Eqs. (11) and (12). Then we have the Lemma 1 as
follows:
Lemma 1: The problem P2 is a strictly convex optimization problem.
Proof: We deﬁne the function f ðPa ; Pb ị as follows:
f Pa ; Pb ị ẳ
h21 h22
r4n
Pa Pb
ỵ
r2n jh23
Pb
ỵ
r2n jh24
Pa
ỵ j2 h23 h24
:
15ị
Since (r4n =Pa Pb ỵ r2n jh23 =Pb ỵ r2n jh24 =Pa ỵ j2 h23 h24 ) decreases as Pa and Pb increase,
(r4n =Pa Pb ỵ r2n jh23 =Pb ỵ r2n jh24 =Pa ỵ j2 h23 h24 ) is strictly convex on the space spanned
by ðPa ; Pb Þ when Pa and Pb are subject to the constraints given in Eqs. (11) and (12).
Thus, f ðPa ; Pb Þ is a strictly concave function on the space spanned by (Pa ; Pb ). On the
other hand, it is clear to verify that all inequalities constraints (Eqs. (11) and (12)) are
linear on the space spanned by (Pa ; Pb ). Therefore, the problem P2 is a strictly convex
optimization problem. To derive the optimal solutions of the problem P2, we have the
following Theorem 1.
Theorem 1: The optimal solutions to the problem P2, denoted by PÃa and PÃb , are
determined by
p
8
2
2 h2 ịPr
2
2
2 2
2n ỵ jP
2 h2 ị
>
4
3
4
< Pa ẳ rn ỵ jPh4 rn þ jPh42Þ Àjðh
;
jðh4 Àh23 Þ
pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ
2
2
2
2
2
2
2
2
2
2
þ
Þ Àjðh Àh ÞðPr
n ỵ jP
h ị
>
rn ỵ jPh
3
: P ẳ rn ÀjPh
4
4
3
4
:
b
jðh2 Àh2 Þ
4
ð16Þ
3
Proof: Because problem P2 is a strictly convex optimization problem, it has the unique
optimal solution. It is easy to know that the optimal solutions to problem P2 need to
: Thus, the optimal solution PÃa and PÃb need to satisfy
satisfy Pa ỵ Pb ẳ P
446
L. Liang et al.
8
r4 P2P ịh2 h2 r2 P2 jh21 h22 h23 ỵ r2n PPa ị2 jh21 h22 h24
@f Pa ;Pb ị
>
>
ẳ n r4 þa r12 jP2 h2n þar2 jP
;
2
2 2 2
2
>
@P
a
>
a 3
b h4 þ j Pa Pb h3 h4 Þ
n
n
n
>
>
>
< @f ðPa ;Pb Þ r4n ðPÀ2Pb Þh21 h22 Àr2n P2b jh21 h22 h24 ỵ r2n PPb ị2 jh21 h22 h23
ẳ
;
@Pb
r4n ỵ r2n jPa h23 ỵ r2n jPb h24 ỵ j2 Pa Pb h23 h24 ị2
>
>
>
ẳ 0;
>
Pa ỵ Pb ị P
>
>
>
: Ã
Pj [ 0; j 2 fa; bg:
ð17Þ
Solving Eq. (17), we can derive the optimal solution expressed as shown in Eq. (16).
Thus, Theorem 1 follows.
Theorem 1 gives the expression of the optimal solutions PÃa and PÃb in the high
SINR region. The solutions are very accurate in the high SINR region and have the
simple closed-form expression.
5 Numerical Results
In this section, we conduct numerical results to evaluate the performance of our
developed optimal power allocation scheme for the visible light full-duplex bidirectional transmission. We consider a 4 m Â 4 m Â 3 m space. We assume that the
LEDs are Lambertian sources with the semi-angle at half power w1=2 ¼ 60 (which
determines m = 1). We further assume that the average power constraint P is set as
70 W. We set A = 15 mm2 smaller than 1 cm2 and c = 1.4738 [16].
Figure 3 depicts the developed power allocation scheme for the two nodes visible
light full-duplex transmission, where we set the irradiance angle at LED (ai ) as 15° and
the incident angle at PD (bi ) as 30° (i = 1, 2, 3, 4). As shown in Fig. 3, when the
distance (h3 ) from the local transmitter to the local receiver at node A is very closed to
distance (h4 ) from the local transmitter to the local receiver at node B, the optimal
power PÃa and PÃb are almost half of the average power. The optimal power of node A
increases as h3 decreases and h4 increases. The optimal power of node B increases as h3
Fig. 3. The power allocations versus the distances from local transmitters to local receivers.