Professional Documents
Culture Documents
Point-to-Point Wireless Information and Power Transfer in WBAN
Point-to-Point Wireless Information and Power Transfer in WBAN
fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2017.2695222, IEEE Access
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2017.2695222, IEEE Access
of applications including medical, consumer lifetime and reduce preservation cost in com-
electronics/personal entertainment and other” munications [13]. In [14], the authors investi-
[3] [4]. In [5], The sensor selection problem gate the power management strategies for the
in WBAN is considered, which is dissolved point-to-point communication powered by EH
by sensor array synthesis algorithm via convex sources. Under the assumption that the capacity
optimization. For the single node and cooper- of the battery is infinite or finite, the authors in
ative multiple nodes, spectrum sensing at sub- [15] study the throughput maximization prob-
Nyquist sampling rates to reduce the computa- lem for both two cases with a deterministic
tional complexity is investigated in [6]. Human EH model and a random EH model. In [16], a
body posture recognition based on WBAN is water-filling energy allocation solution where
discussed in [7]. The authors in [8] propose a the so-called water levels follow a staircase
game theoretic approach about the interference is proved to be optimal, which is investigated
caused by multiple WBANs located closely to in energy harvesting with channel fading. The
each other in a small area. In [9] [10], effi- authors in [17] consider the block random EH
cient resource allocation for Device-to-Device model, and derive the optimal power alloca-
is considered. The authors in [11] consider the tion polices of the throughput maximization
problem of energy efficient resource allocation problem via dynamic programming and convex
for the wireless power transfer enabled collab- optimization techniques. In WBAN, a mass of
orative mobile clouds. By assuming that the sensors consume much energy, and the sensors
mobile cloud contains numbers of hybrid ID have limited operation time. When the sensors
and EH user equipment devices, the authors in are exhausted, WBAN will be fail. However,
[12] propose an algorithm that can noticeably for applications in WBAN replacing batteries
obtain energy efficiency performance. is inconvenient or impossible. Thus energy har-
Compared to conventional communication, vesting in WBAN, whereby energy is extracted
wireless communication with energy harvesting from body surrounding environment, becomes
(EH) is flourishing, and the sensor does not an alternative solution. The authors in [18]
have to consider saving energy at work. Thus discuss the propagation channel between two
optimizing power allocation of sensor nodes half-wavelength dipoles which placed near a
with energy harvesting has recently drawn sig- human body and presents an application for
nificant research interests , which can improve cross-layer design in order to optimize the
the efficiency of networks, prolong network energy consumption of different topologies at
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2017.2695222, IEEE Access
2.45 GHz. In [19], the authors make a relatively neous wireless information and power transfer
comprehensive introduction from energy har- (SWIPT) via the approach of dynamic power
vesting principle and characteristics to practical splitting (DPS). The authors in [26] propose a
application, and analyzed the data communi- new channel learning algorithm for the point-
cation. The throughput and delay performance to-point MIMO WET system. By revising that
of the IEEE 802.15.6 standard is presented in the energy receiver (ER) sends back to the en-
[20], where the authors derive some equations ergy transmitter (ET) only one bit per feedback
for the maximum throughput and minimum interval to indicate the increase or decrease of
delay. its harvested energy. Assuming channel reci-
Recently, the research on the radio fre- procity, this paper studies the optimal design
quency energy-based energy harvesting tech- of channel training for MIMO WET systems
nology with wireless communications has at- in Rician fading channels. The forward link
tracted the attention of people. In [21], the channel of WET can be efficiently estimated
power splitting (PS) and time switching (TS) at the ET based on the training signals sent by
protocol for wireless-powered relaying system the ER in the reverse communication link [27].
with simultaneous energy harvesting and infor- In recent years, how to use the harvested
mation transmission are considered. The au- energy and transfer information efficiently
thors in [22] propose a new wireless powered for a point-to-point communication system in
communication (WPC) system with a bidi- WBAN is becoming an urgent and significant
rectional information and energy forwarding research problem. In this paper, we consider
relay. In [23], the authors investigate practi- the joint command transfer, wireless power
cal receiver designs for simultaneous wireless transfer and information transfer based on PS
information and power transfer. Based on dy- and TS protocol, and in this system, sen-
namic power splitting (DPS), they propose two sor transfers command and information with
practical receiver architectures, namely, sepa- the harvested energy. An optimal information
rated and integrated information and energy throughput problem with energy harvesting and
receivers. In [24], optimal time allocation for information transferring is proposed to maxi-
the power receiver to balance wireless power mize the throughput in WBAN.
transfer and information transfer has been stud- In normal circumstance, we consider power-
ied. In [25], under a point-to-point flat-fading splitting architecture at the sensor. In the
SISO channel setup, the authors study simulta- first phase, the AP transmits command and
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2017.2695222, IEEE Access
wireless energy simultaneously, the command- information transfer, in order to maximize the
signal only indicates the sensor, and then the information throughput of wireless information
sensor transfers information in energy which transmission in the uplink.
is harvested by the sensor to the AP. In this
system, the information signal block length is
the same as the command signal, however,
the command signal takes up a small part
of the entire information signal block length,
which can ensure the information signal to
be transmitted successfully. Thus, the time
slot should be divided into energy harvesting
and information portions. It is significant to
determine the optimal command signal time
allocation in this system in WBAN. We attempt Fig. 1. The body channel model
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2017.2695222, IEEE Access
(QHUJ\ ,QIRUPDWLRQDQG(QHUJ\
,QIRUPDWLRQ
As shown in Fig.1, we consider a Wireless ,QIRUPDWLRQ +DUYHVWLQJ
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2017.2695222, IEEE Access
other is used for information processing. The where µ is the coefficient of amplify-and-
′
signal received at sensor during the first phase forward at sensor. ns is the sensor noise mod-
is given by eled as a zero-mean complex Gaussian random
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2017.2695222, IEEE Access
On the other hand, the signal received It is clear that the power-splitting parameters
at the AP from the sensor is y .and the affect the information throughput only through
a
power due to double channel attenuations h approach to infinity, and the optimal solution
and g, power splitting ratio ρ. Thus we have must be within the open interval (0, 1), which
Ps ≪ Pa . As a result, the end-to-end informa- can be obtained by finding stationary points via
tion throughput from the sensor to the AP is solving equation ρ2 + 2W ρ − W = 0. So we
( ) can obtain the optimal power splitting ratio ρ∗
1 (1 − ρ)µ|h|2 |g|2 Pa in close-from as
R (ρ) = log2 1+
2 |g|2 µσn2 + σa2
√
(11) ρ∗ = −W + W2 + W (15)
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2017.2695222, IEEE Access
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2017.2695222, IEEE Access
received signal at the sensor can be given as In the third phase of the duration (1 − τ )T /2,
√ the sensor forwards an amplified signal c ya2 =
ys,ab = Pa hxa + ns (18) √ (√ ′)
g µs Pa hxa + ns +na to the AP. With sim-
According to (18), the total energy harvested ilar analysis as in the Section III, the received
at the sensor during each block is given by SNR at the AP can be expressed as
[ ]
Es,ab = ηE |ys,ab |2 τ T ≈ ηPa |h|2 τ T (19) µs Pa |h|2 |g|2
γab (Ps,ab , τ ) = (24)
|g|2 µs σn2 +σa2
In the third phase of the duration (1 − τ )T /2,
Since γab (Ps,ab , τ ) monotonically decreases
the sensor forwards an amplified signal to the
with Ps,ab for any fixed τ , Ps,ab should
AP, Which can be achieved as
be set to its maximum possible value
√
ya2 = Ps,ab gxsi + na (20) while satisfying the energy consumption
where xsi is the information-signal of the sen- and harvesting balance constraint.
sor, and with (20), the average energy con- i.e., η|g|2 µs E [|ya2 |2 ] ((1 − τ )T /2) =
in the uplink (from the sensor to the AP) can τ η 2 Pa |h|2 |g|2 T 2τ η 2 Pa |h|2 |g|2
Ps,ab = = (25)
be expressed as (1 − τ )T /2 1−τ
Esi = E [|ya2 |2 ] ((1 − τ )T /2) And the end-to-end throughput from the sensor
(21)
= (Ps,ab |g|2 +σn2 ) ((1 − τ )T /2) to the AP is
( )
In this paper we consider the case when all of 1−τ µs Pa |h|2 |g|2
Rab (Ps,ab , τ ) = log2 1+ 2
2 |g| µs σn2 +σa2
the harvested energy is used for information.
(26)
The average energy consumed at the sensor in
Note that in (26), for convenience, we simplify
the uplink (from the sensor to the AP) should
the (22) inequality constraint to an equality
satisfy the energy harvesting constraint at each
constraint. By substituting (23) and (25) into
block in the entire system. We thus have
(26), the end-to-end information throughput as
µs (Esc + Esi ) ≤ Es,ab
(22) a function of τ can be expressed as
( )
where µs denotes the amplification coefficient Rab (τ ) = 1−τ log 1+ 1
2 2
D1 +D2 (1 − τ ) /τ
to satisfy the energy consumption. According
(27)
to (22), we have
∆ ( 2) ∆
2ηPa |h| τ
2 where D1 = (σ 2
a η+σ 2
n ) / P a |h| and D2 =
µs ≤ (23) ( )
(Ps,ab |g|2 + σn2 ) (1 − τ ) σa2 σn2 / 2ηPa2 |h|4 |g|2 .
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2017.2695222, IEEE Access
10
can be obtained. It is what WBAN needed, can be verified that q ∗ =W (c), where
∆
and we can get a higher throughput in this c = exp(−1) (1/d2 −1). In this WBAN
system. Thus, to maximize the throughput of system, it follows that Pa → 0, and thus
three phases should be optimized. i.e., τ T → T , it satisfies that the first phase which
is used to send abnormal command and the
max Rab (τ )
τ (28) second phase used to harvest energy, and
s.t. 0≤τ ≤1 hence most of time should be allocated for
From the objective function (27), the opti- information processing.
mal time ratio for throughput maximization is
then given by τ ∗ = arg max Rab (τ ), which can V. SIMULATION RESULTS
0≤τ ≤1
be obtained numerically via one-dimensional In this section, the simulation results are
searching. According to WBAN power char- presented. Numerical results are provided to
acteristics, to gain some insights, we consider verify the normal circumstance and abnormal
the extreme case with Pa → 0. In this case, circumstance schemes in WBAN. According to
D1 /D2 ∝ Pa , i.e., d1 is dominated by D2 and body surface to body surface channel model
thus D1 can be neglected. The optimization [4], the following path loss model is based on
problem can be solved through the optimization measurements that cover frequencies of 3.1-
algorithm. The first order of (27) necessary 10.6 GHz. Depending on the location of the
condition is equal to zero, the optimal solution AP and the sensor and the parameters of path
of (27) is given as follows model are extracted from the measurements
( ) [4] [28], the path-loss model in front of the
τ∗ 1
ln 1+ = ∗
D2 (1 − τ )
∗ τ +D2 (1 − τ ∗ ) torso is the function of the distance da and
(29) dr . The downlink channel gain distribution
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2017.2695222, IEEE Access
11
Fig. 6. The optimal power-splitting ratio ρ∗ in normal Fig. 7. Information throughput R (τ ) in normal circumstance
circumstance versus d versus power-splitting ratio ρ
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2017.2695222, IEEE Access
12
tween the AP and the sensor is in the interval decreases as the AP moves near to the sensor.
[0.6,1.4], the optimal power-splitting ratio is The optimal time switching ratio τ ∗ is around
around 0.5, whereas its value increases as the 0.057 . The optimal time-switching ratio τ ∗ is
AP moves near to the sensor and decreases as calculated in section IV. The numerical results
the AP moves far to the sensor . This can be agree with our theoretical analysis. The com-
verified to be in accordance with section III. parison of the maximum information through-
Fig.7 demonstrates the relationship between put Rab (τ ) in different time-switching ratio τ ∗
information throughput R (τ ) and power- in abnormal circumstance against d is depicted
splitting ratio ρ from (11) under no constraint. in Fig.10. It is obvious that when the distance
We set da =dr =1m. In this figure, we can see is far, the information throughput difference is
that, in the interval [0, 0.5], the throughput not notable, and when the distance is close,
increases with ρ , and in the interval [0.5, 1], the values are different from each other. As
the throughput decreases with the increase in ρ. d increases, the power gain decreases. This
This implies that there exists a unique optimal means that the channel condition is going to
power-splitting ratio. become bad, which leads to the decrease in the
In Fig.8, we compare the information throughput. When the time-switching ratio τ
throughput performance between the optimal is optimal, the throughput is higher than that
and the fixed power-splitting policies. Gener- under the fixed time allocotion.
ally, the throughput performance is notable,
particularly when the power-splitting ratio is
optimal. This confirms that it is important to
differentiate the optimal power-splitting ratio
and fixed ratio when designing the power-
splitting policy for good throughput perfor-
mance.
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2017.2695222, IEEE Access
13
Fig. 10. Information throughput Rab (τ ) under different time- Fig. 12. Information throughput Rab (τ ) in abnormal circum-
switching ratios in abnormal circumstance versus d stance versus η under different time-switching ratio τ
Fig.11 reports the comparison results of the In this paper, an information throughput
throughput under the optimal time-switching maximization of point-to-point link in WBAN
ratio τ ∗ , time-switching ratio τ =0.2 and time- over path loss channel model with energy har-
switching ratio τ =0.5 versus different values of vesting is considered. In this case, we focus
transmission power Pa in the interval [0,1mw]. on the design of a power-splitting protocol in
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2017.2695222, IEEE Access
14
The optimal power splitting policies and the RATIO OF EQUATION (15)
optimal time switching policies in point-to- By taking equation (10) into equation (8),
point transmission are derived. Then, an opti- we can get
mization problem in this system is proposed to ηρPa |h|2
µ= (31)
maximize the information throughput with en- (1 − ρ) η 2 ρPa |h|2 |g|4 + σn2
ergy constraint. Additionally, numerical results By taking equation (31) into equation (9), we
in different system parameters are provided to can get
verify that our proposed point-to-point designs (1 − ρ)ρ
γ (ρ) = U (32)
in WBAN are reasonable. ρ + V ρ (1 − ρ) + W
∆ |h|2 Pa ∆ σa2 |g|2 ∆
This work only considered a power split- Where U = σn2 , V = ησn 2 and W =
σa2
ting scheme in the normal scenario. As for |h| |g|2 ηPa
2 . To get the optimal power splitting
future work, the time switching scheme will ∗
ratio ρ , the function of ρ can be given as
be considered in the normal scenario, and the (1 − ρ)ρ
f (ρ) = (33)
performance comparison will be provided. An- ρ + V ρ (1 − ρ) + W
other interesting extension is to further study It also can be given by f (ρ) =
1
a multiuser system with wireless Information 1/(1−ρ)+V +W /(ρ(1−ρ))
Now by calculating
and power transfer. the first derivative of the g (ρ) =
1/(1−ρ) + V + W /(ρ (1−ρ)), the g(ρ)′
is achieved as
ACKNOWLEDGMENT 1 W (2ρ − 1)
g(ρ)′ = 2 + 2 (34)
(1 − ρ) ρ (1 − ρ)2
This work was supported by the Na- From (34), the optimal power splitting ratio ρ∗
tional Natural Science Foundation of China can be given by equation (15)
(No.61671219 and 61273064), Jilin Provin-
cial Science and Technology Department R EFERENCES
Key Scientific and Technological Project [1] S. Ullah, H. Higgins, B. Braem, B. Latre, C. Blondia,
I. Moerman, S. Saleem, Z. Rahman, and K. S. Kwak, “A
(No.20140204034GX) and Jilin Province De-
comprehensive survey of wireless body area networks,”
velopment and Reform Commission Project Journal of medical systems, vol. 36, no. 3, pp. 1065–
(No.2015Y043). 1094, 2012.
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2017.2695222, IEEE Access
15
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2017.2695222, IEEE Access
16
2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.