IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 72, NO.

1, JANUARY 2023 1383

Intelligent Reflecting Surface-Aided Wireless Powered I. INTRODUCTION

Hybrid Backscatter-Active Communication Networks In the future Internet of things (IoT), massive power-constrained
sensor devices (SDs) will be deployed to collect data for supporting
Sun Mao , Kun Yang , Senior Member, IEEE,
various intelligent applications, such as forest fire monitoring, intelli-
Jie Hu , Senior Member, IEEE, Dong Li ,
gent automatic manufacturing, smart city, and so forth [1]. Due to the
Dapeng Lan , Graduate Student Member, IEEE,
size restrictions and the cost constraints, the battery capacity of SDs
Ming Xiao , Senior Member, IEEE, and Youzhi Xiong
is generally small and the battery replacement will be inconvenient
and high-cost, especially for the application that equips SDs in the
Abstract—Wireless powered hybrid backscatter-active communications remote or dangerous areas. Therefore, it is a vital research topic
are recognized as the promising technique to provide the sustainable to prolong the SDs’ operation time in a low-cost manner. Wireless
communication service for wireless devices. In this paper, we consider an powered communication network (WPCN) has been envisioned as
intelligent reflecting surface-aided wireless powered hybrid backscatter-
active communication network, where multiple sensors devices powered
a promising solution, due to its ability to power the SDs by ex-
by a power beacon, transmit their information to an access point through ploiting radio-frequency (RF)-based wireless energy transfer (WET)
backscatter and active communications. In particular, an intelligent re- technique [2].
flecting surface is properly deployed to improve the communication quality Wireless powered active communications (WPAC) and backscatter
between sensor devices and access point by utilizing the composite channel communications (BackCom) are two transmission paradigms for
gain and phase beamforming gain. Furthermore, we formulate a system
energy efficiency maximization problem under the minimum throughput WPCNs [3]. In WPAC, the SDs first harvest energy from the RF energy
requirements of sensor devices, to jointly optimize the phase shifts of in- signal transmitted by the power beacon, and then utilize the harvested
telligent reflecting surface, time allocation for backscatter communications energy to transmit their information to the access point (AP). Due to
and active transmissions, and transmit power of sensor devices and power the extra time required for WET, the time length for information trans-
beacon. To solve the formulated non-convex problem, we first unveil the
optimal phase shifts of intelligent reflecting surface, and then utilize the
missions will be inevitably reduced, which takes a negative effect on
variable substitutions and Dinkelbach-based method to develop an iterative communication performance. Different from WPAC, BackCom enables
method to solve the energy efficiency maximization problem. Simulation the SD to split the received signal into two parts, one is modulated and
results reveal that the proposed method outperforms the existing several backscattered for carrying its own information to the intended receiver,
benchmark methods in terms of system energy efficiency. and the rest is used to collect energy for maintaining its circuit power
Index Terms—Backscatter communications, intelligent reflecting consumption. Besides, the SDs do not require any energy-consuming
surface, wireless energy transfer. active RF circuits in BackCom, so it can greatly extend the operation
time of SDs. Despite the benefits brought by the BackCom technique, it
also reduces the achievable throughput of SDs compared to WPAC [4].
Therefore, the BackCom technique can be integrated with WPAC for
balancing the energy efficiency and throughput performance of current
Manuscript received 20 April 2022; revised 9 July 2022 and 6 September WPCNs, and it will introduce a new design paradigm, namely wireless
2022; accepted 6 September 2022. Date of publication 12 September 2022; date powered hybrid backscatter-active communication networks. In [4]
of current version 16 January 2023. This work was supported in part by the
Natural Science Foundation of China under Grants 62132004 and 62101370, and [5], the authors investigated the throughput maximization-based
in part by the Natural Science Foundation of Sichuan Province under Grants and energy efficiency optimization-based resource allocation methods
2022NSFSC0479 and 2022NSFSC0480, in part by Sichuan Science and Tech- for wireless powered hybrid backscatter-active communication
nology Program under Grant 2022YFH0022, in part by Sichuan Major R&D networks.
Project under Grant 22QYCX0168, and in part by The Science and Technology
Development Fund, Macau SAR, under Grants 0018/2019/AMJ, 0110/2020/A3,
Nevertheless, wireless powered hybrid backscatter-active commu-
and 0029/2021/AGJ. The review of this article was coordinated by Prof. Vittorio nication is fundamentally restricted by the low efficiency of WET and
Degli-Esposti. (Corresponding authors: Kun Yang; Jie Hu.) double pathloss of BackCom. Recently, intelligent reflecting surface
Sun Mao is with the College of Computer Science, Sichuan Normal Univer- (IRS) has been proposed as a promising technique to overcome this
sity, Chengdu 610101, China (e-mail: [email protected]). bottleneck. In general, IRS is a smart thin panel integrated with massive
Kun Yang is with the School of Computer Technology and Engineering,
Changchun Institute of Technology, Changchun 130012, China, and also with active/passive reflection units, and it can achieve fine-grained reflect
the School of Information and Communication Engineering, University of beamforming to improve the efficiency of WET and BackCom [6],
Electronic Science and Technology of China, Chengdu 611731, China (e-mail: [7]. The recent works in [8], [9], [10], [11], [12] integrated the
[email protected]). IRS technique with WET or BackCom, and they demonstrated that
Jie Hu is with the School of Information and Communication Engineering,
University of Electronic Science and Technology of China, Chengdu 611731,
the IRS has great potential to improve the efficiency of WET and
China (e-mail: [email protected]). BackCom.
Dong Li is with the School of Computer Science and Engineering, Although the resource scheduling problems for IRS-assisted
Macau University of Science and Technology, Macau 999078, China (e-mail: BackCom or IRS-aided WPAC have been investigated separately in
[email protected]). existing works. However, few researchers considered the integration
Dapeng Lan is with the Department of Informatics, University of Oslo, 0316
Oslo, Norway (e-mail: [email protected]). of IRS technique with wireless powered hybrid backscatter-active
Ming Xiao is with the Department of Information Science and Engineer- communications, which will make the resource allocation problem
ing, KTH Royal Institute of Technology, 10044 Stockholm, Sweden (e-mail: more difficult, due to the highly coupled among the phase shift matrix
[email protected]). of IRS, time-slot allocation for backscatter and active communications,
Youzhi Xiong is with the College of Physics and Electronic Engineer-
ing, Sichuan Normal University, Chengdu 610101, China (e-mail: yzx-
and transmit power control. Meanwhile, energy efficiency defined as
[email protected]). bits-per-joule is an important metric to measure the performance of
Digital Object Identifier 10.1109/TVT.2022.3205950 WET-based networks.

0018-9545 © 2022 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: Indraprastha Institute of Information Technology. Downloaded on June 12,2023 at 09:43:28 UTC from IEEE Xplore. Restrictions apply.
 1384 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 72, NO. 1, JANUARY 2023

Fig. 1. IRS-assisted wireless powered hybrid backscatter-active communication networks.

Therefore, this paper studies the energy-efficient resource allocation power beacon, an IRS with M reflection elements, and K energy-
and phase beamforming optimization for an IRS-aided wireless pow- constrained SDs. In addition to the IRS, all the other wireless terminals
ered hybrid backscatter-active communication network including an are equipped with a single antenna. The sets of reflection elements
AP, a power beacon, an IRS, and multiple SDs. The main contributions and SDs are denoted by M = {1, 2, . . . , M } and K = {1, 2, . . . , K},
of this paper are outlined as follows. respectively. It is assumed that each SD is equipped with the radio-
r We formulate a system energy efficiency maximization problem frequency energy harvesting circuit, the backscatter circuit and wire-
under the minimum transmission bits requirements of sensors less transceiver circuit, in order to achieve hybrid backscatter-active
devices, and study the joint scheduling of the time-slot allocation communication mode.
for backscatter communications and active transmissions, the Furthermore, all the channels follow the time block-based flat fading
transmission power of sensor devices and power beacon, and the model, namely the channel keeps constant during the considered time
phase shifts of the IRS. block but changes over different blocks. The channels from the power
r To solve the formulated non-convex fractional optimization prob- beacon to the k-th SD, from the k-th SD to the IRS, from the k-th SD
lem, we adopt the variable substitution technique to tackle the to the AP, and from the IRS to the AP are denoted by gD,k ∈ C 1×1 ,
coupled optimization variables, and utilize a Dinkelbach-based hHT,k ∈ C
1×M
, hD,k ∈ C 1×1 , hR ∈ C M ×1 , respectively.
iterative algorithm to solve the energy-efficient resource manage- The whole time block will be divided into two phases, namely
ment problem. In addition, we derive the optimal transmit power backscatter phase and active transmission phase. In the backscatter
of SDs in closed-form expressions by exploiting the Karush- phase, the SDs can not only harvest energy from the RF signal sent
Kuhn-Tucker (KKT) conditions, and find that each SD decides by the power beacon, but also modulate and reflect the received signal
to transmit its information to the AP via active communications for carrying their own information to the AP. In the active transmission
only when the composite channel gain from the SD to the AP is phase, the SDs will exploit the harvested energy to conduct active
larger than a threshold. information transmissions to the AP. In particular, the IRS is deployed
r Numerical results verify that the proposed hybrid backscatter- to assist both the backscatter and active communications between the
active communication networks exhibit a higher energy efficiency AP and SDs. The detailed transmission procedure will be illustrated in
performance, as compared to the pure backscatter/active commu- the following subsections.
nication networks. Moreover, we also find that the IRS has great
potential to improve the system energy efficiency via optimizing
its phase shifts.
A. Backscatter Phase
As shown in Fig. 1, the backscatter phase is further partitioned into
K sub-slots. During the k-th sub-slot τk , the k-th SD will modulate and
II. SYSTEM MODEL
backscatter its received RF signal sent by the power beacon, in order to
As illustrated in Fig. 1, we consider an IRS-aided wireless powered carry its information the AP. In the other sub-slots, the k-th SD is able
hybrid backscatter-active communication network including an AP, a to harvest energy from downlink RF signals. Hence, the total harvested

Authorized licensed use limited to: Indraprastha Institute of Information Technology. Downloaded on June 12,2023 at 09:43:28 UTC from IEEE Xplore. Restrictions apply.
 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 72, NO. 1, JANUARY 2023 1385

energy by the k-th SD will be consumption of power beacon. Therefore, the system energy efficiency
 can be expressed as
Ek = ηPa |gD,k |2 τi , (1)
K
i∈K,i=k Rtot k=1 (RB,k + RA,k )
ηEE = =  . (6)
where Pa denotes the transmission power of the power beacon, and
Etot Pa K k=1 τk
η ∈ (0, 1] stands for the energy conversion efficiency. For notation
convenience, we assume that all energy harvesting devices have the III. PROBLEM FORMULATION AND SOLUTION
same energy conversion efficiency. In this paper, we aim at maximizing the system energy efficiency
√ In the time Hslot τk , the signal received at the AP is given by via jointly optimizing the time-slot allocation {τk , tk , ∀k ∈ K}, the
ζνPa gD,k (hT,k ΓB,k hR + hD,k )se sbk + n, where se represents the phase beamforming {ΓA,k , ΓB,k , ∀k ∈ K} of IRS, the transmission
normalized energy signal transmitted by the power beacon, sbk is the power {Pk , ∀k ∈ K} of SDs, and the transmission power Pa of power
normalized information signal backscattered by the k-th sensor device, beacon. This problem can be formulated as
n denotes the Gaussian noise generated at AP, ν ∈ (0, 1] represents the
scattering efficiency at the sensor devices, ζ denotes the throughput maximize ηEE
gap between the backscatter communications and active transmis- {τk ,tk ,Pk ,Pa ,ΓA,k ,ΓB,k }

sions, and ΓB,k = diag{ejφB,k,1 , . . . , ejφB,k,M }, where {φB,k,m ∈

s.t. C1: Rtot,k ≥ Rk,min , ∀k,
[0, 2π], ∀m ∈ M} represents the phase shift of the IRS during τk . As
we know, the Shannon capacity is derived based on the Gaussian signals. C2: Ekcons ≤ Ek , ∀k,
However, the multiplicative signal se sbk does not follow the Gaussian 
distribution. Therefore, it will result in a performance gap between the C3: (τk + tk ) ≤ T,
channel capacity of backscatter communications and Shannon capacity. k∈K

According to [4], [13], the amount of bits transmitted by the k-th SD C4: 0 ≤ φB,k,m ≤ 2π, ∀k, m,
in the backscatter phase can be expressed as
  C5: 0 ≤ φA,k,m ≤ 2π, ∀k, m,
ζνPa |gD,k |2 |hHT,k ΓB,k hR +hD,k |
2
RB,k = Bτk log2 1+ , C6: 0 ≤ Pk ≤ Pk,max , ∀k,
δ2
(2) C7: 0 ≤ Pa ≤ Pa,max ,
where B is the system total bandwidth, and δ 2 denotes the Gaussian C8: τk , tk ≥ 0, ∀k, (7)
noise power.
where C1 ensures that the total throughput achieved by SDs should
B. Active Transmission Phase be higher than their minimum throughput requirements, C2 represents
In the active transmission phase, the SDs use the harvested RF energy that the total energy consumption of SDs cannot exceed their harvested
to transmit information to the AP in a time-division manner. During the RF energy, C3 denotes the maximum time length constraints, C4-C5
sub-slot tk for active communications, the amount of bits transmitted are the phase shift constraints at IRS, C6 and C7 restrict the maximum
by the k-th SD is given by transmission power at SDs and the power beacon, respectively.
As observed, the formulated problem (7) is a non-convex fractional
 
Pk |hH optimization problem, which is hard to obtain the globally optimal
T,k ΓA,k hR +hD,k |
2
RA,k = Btk log2 1 + , (3) solution in polynomial time. To solve it, we first derive the optimal
δ2
phase shifts of IRS in the following Lemma 1.
Lemma 1: The system energy efficiency is maximized when the
where Pk denotes the k-th SD’s transmission power, and ΓA,k =
phase beamforming of IRS satisfies the following equations
diag{ejφA,k,1 , . . . , ejφA,k,M }, where {φA,k,m ∈ [0, 2π], ∀m ∈ M}
stands for the phase shift of the IRS during tk . Therefore, the total
φ∗B,k,m = φ∗A,k,m = arg(hD,k ) − arg(hk [m]), (8)
throughput achieved by the k-th SD via backscatter communications
and active transmissions will be
where hk = diag(hH T,k )hR .
Rtot,k = RB,k + RA,k . (4) Proof: According to the result presented in [6], the phase shift of IRS
should be customized such that the cascaded link via IRS aligns with
Furthermore, the total energy consumption of the k-th SD includes its direct link in scenarios without multiuser interference. Therefore,
two parts, i.e., the energy consumption for backscatter and active the optimal phase shift of IRS can be derived as given in (8). 
communications, which can be expressed as Substituting the optimal phase beamforming {Γ∗A,k , Γ∗B,k } into (7),
the energy efficiency optimization problem will be simplified as
Ekcons = Pcbac τk + tk (Pk + Pcac ), (5)
K
k=1 Rtot,k (τk , tk , Pk , Pa )
where Pcbac
and Pcac
indicate the circuit power consumption during the maximize  (9a)
{τk ,tk ,Pk ,Pa } Pa K k=1 τk
backscatter phase and the active transmission phase, respectively.
s.t. Rtot,k (τk , tk , Pk , Pa ) ≥ Rk,min , ∀k, (9b)
C. System Energy Efficiency 
Ekcons ≤ ηPa |gD,k |2 τi , ∀k (9c)
Since SDs need to harvest energy from the power beacon to conduct i∈K,i=k
the information transmission, so we define the system energy efficiency
as the ratio of the total amount of bits transmitted by SDs and the energy C3, C6-C8. (9d)

Authorized licensed use limited to: Indraprastha Institute of Information Technology. Downloaded on June 12,2023 at 09:43:28 UTC from IEEE Xplore. Restrictions apply.
 1386 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 72, NO. 1, JANUARY 2023

In (9), we have
 
Pk |ĥk |2
Rtot,k (τk , tk , Pk , Pa ) = Btk log2 1+
δ2
 
ζμPa |gD,k |2 |ĥk |2
+ Bτk log2 1+ , (10)
δ2

where ĥk = hH ∗ H ∗
T,k ΓB,k hR + hD,k = hT,k ΓA,k hR + hD,k . Based on
the Dinkelbach-type method, the optimal solution of (9) can be obtained
by iteratively solving the following problem:


K 
K
maximize Rtot,k (τk , tk , Pk , Pa ) − uPa τk (11a)
{τk ,tk ,Pk ,Pa }
k=1 k=1

s.t. (9b)–(9d), (11b)

where u is an iteration parameter. Due to the coupled optimization

variables, e.g., Pk and tk , the problem (11) is still non-convex. It
motivate us to introduce the following auxiliary variables, i.e., bk =
Pk tk and ck = Pa τk , and (11) will be converted to the following
problem:
K 
 
 ζμck |gD,k |2 |ĥk |2
Bτk log2 1+ complexity
√ of Algorithm 1 can be calculated as O(16Nc (12K +
k=1
τk δ 2 1)K 2 8K + 1 log( 1 )), where denotes the tolerance factor. It should
maximize    (12a) be noted that the optimal phase beamforming is derived as closed-form
{τk ,tk ,bk ,ck }
bk |ĥk |2 K
+ Btk log2 1 + −u ck expression, the corresponding complexity is much smaller than that of
tk δ 2 k=1 the iteration procedure, and it can be ignored.
 
ζμck |gD,k |2 |ĥk |2
Bτk log2 1+ B. Derivation of Optimal Solution
τk δ 2
s.t.   (12b) For the purpose of gaining more insights, we adopt the KKT con-
bk |ĥk |2 dition to derive the structure of optimal solutions. The Lagrangian
+ Btk log2 1 + ≥ Rk,min , ∀k function of (12) can be expressed as
tk δ 2
   
Pcbac τk +bk +tk Pcac ≤ η|gD,k |2 ci ,∀k, (12c) 
K
ζμck |gD,k |2 |ĥk |2
i=k
L= − Bτk log2 1+
k=1
τk δ 2
0 ≤ ck ≤ τk Pa,max , ∀k, (12d)  
bk |ĥk |2 
K

0 ≤ bk ≤ tk Pk,max , ∀k, (12e) +Btk log2 1+

tk δ 2
+u ck
k=1
C3, C8. (12f)   

K
ζμck |gD,k |2 |ĥk |2
According to the property of perspective functions, we can prove that + λ1,k Rk,min − Bτk log2 1+
k=1
τk δ 2
(12) is a convex problem, which can be solved by a variety of toolbox,
  
such as CVX [14]. bk |ĥk |2 
K

The detailed process for solving (7) is outlined in Algorithm 1. In −Btk log2 1+ + λ2,k Pcbac τk +bk +tk Pcac
tk δ 2 k=1
summary, the optimal phase shift matrix of IRS is obtained according
to Lemma 1 in step 2, and then we can utilize the variable substitution   
 
K
technique and Dinkelbach’s method to design an iterative method to − η|gD,k | 2
ci + λ3 (τk + tk ) − T
obtain the optimal transmit power and time-slot allocation strategy in i∈K,i=k k=1
step 3–11.

K 
K
+ λ4,k (ck − τk Pa,max ) + λ5,k (bk − tk Pk,max ), (13)
A. Convergence and Complexity Analysis k=1 k=1

According to [15], the proposed Dinkelbach’s method will converge

where λ1,k , λ2,k , λ3,k , λ4 , and λ5,k are the non-negative Lagrangian
to the globally optimal system energy efficiency with the super-linear
multipliers.
rate. Next, we will analyze the computational complexity of Algo-
Theorem 1: The optimal transmission power of k-th SD during the
rithm 1, which mainly includes the iteration number and per-iteration
active transmission phase will be
complexity. We assume that Algorithm 1 is able to converge to the
optimal solution after Nc iterations. Moreover, in each iteration, the  +
interior-point method-based toolbox is adopted to solve (12) with b∗ (1 + λ∗1,k )B δ2
Pk∗ = ∗k = ∗ ∗
− (14)
4 K variables and 8K + 1 constraints. Hence, the total computational tk (λ2,k + λ5,k ) ln 2 |ĥk |2

Authorized licensed use limited to: Indraprastha Institute of Information Technology. Downloaded on June 12,2023 at 09:43:28 UTC from IEEE Xplore. Restrictions apply.
 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 72, NO. 1, JANUARY 2023 1387

Proof: By taking the partial derivative of L to bk , we have

|ĥk |2
∂L (1 + λ1,k )B δ2
=− + λ2,k + λ5,k (15)
∂bk ln 2 bk |ĥk |2
1+ δ 2 tk

∂L
By letting ∂b = 0, we can derive the optimal transmission power Pk∗
k
as given in (14). 
Remark 1: As observed from Theorem 1, we find that the k-th
SD decides to transmit its information to the AP during the active
transmission phase only when the composite channel gain |ĥk |2 is larger
δ 2 (λ∗2,k +λ∗5,k ) ln 2
than (1+λ∗ )B
.
1,k
Theorem 2: The optimal transmission power of power beacon for
maximizing the total energy efficiency is given by
c∗k
Pa∗ =
τk∗
⎡ ⎤+
⎢ (1+λ∗1,k )B δ2 ⎥
=⎣ ∗  − ⎦
(λ4,k +u∗ + λ∗2,i η|gD,i |2 ) ln 2 ζμ|gD,k |2 |ĥk |2
i∈K,i=k Fig. 2. System energy efficiency ηEE versus performance gap ζ.
(16)

Proof: By taking the partial derivative of L to ck , we have

ζμ|gD,k |2 |ĥk |2
∂L (1 + λ1,k )B δ2
= −
∂ck ln 2 ζμck |gD,k |2 |ĥk |2
1+ δ 2 τk

+ λ2,i η|gD,i |2 + λ4,k + u (17)
i∈K,i=k

∂L
By letting ∂c = 0, we can derive the optimal transmission power Pa
k
of the power beacon as illustrated in (16). 
Theorem 3: For the total energy efficiency maximization problem
(7), the optimal solution is obtained by using up all the available
transmission time T .
Proof: We assume that (τk∗ , t∗k , Pk∗ , Pa∗ ) achieves
K the maximum en-
∗ ∗ ∗
ergy efficiency ηEE of (7) and satisfies k=1 (τk + tk ) < T . We
∗ ∗ ∗ ∗
can construct another feasiblesolution (xτk , xtk , Pk , Pa ) with x =
K T ∗ ∗ > 1 such that K k=1 x(τ k + t k ) = T . In addition, sub-
(τ +t )
k=1 k k
stituting this solution into the objective function of (7), we can derive
that ηEE (xτk∗ , xt∗k , Pk∗ , Pa∗ ) = ηEE
∗
. Therefore, the maximum energy
Fig. 3. System energy efficiency ηEE versus maximum time block length T .
efficiency can always be obtained by using up all the available trans-
mission time T . 
the receive node, α = 2.5 stands for the path-loss factor, and h follows
IV. NUMERICAL RESULTS Rayleigh fading.
In this section, we demonstrated the superiority of the proposed In Fig. 2, we show the system energy efficiency against the per-
energy-efficient resource scheduling strategy for IRS-aided wireless formance gap ζ between the backscatter communications and active
powered hybrid passive-active communication networks, as compared transmissions. It is observed from this figure that except for the pure
with considered benchmark methods. In the simulations, the system active transmission mode, all the other schemes increase with the
bandwidth is B = 1 MHz, the noise power is δ 2 = 10−9 W, the energy performance gap ζ. Because that the energy efficiency of backscatter
conversion efficiency is η = 0.8, the scattering efficiency is ν = 0.8, communications will increase with ζ, while the active transmissions
and the circuit power consumption of SDs for backscatter communi- are not influenced by ζ. By comparisons, we find that the proposed
cations and active transmissions are set as Pkbac = 10μ W and Pkac = 1 method can achieve the highest system energy efficiency as compared
mW, respectively, the time block length is set as T = 1 Seconds, with the other benchmark methods, because the IRS will improve
the number of reflection elements is M = 30, and the minimum re- the communication efficiency between SDs and AP, and the pure
quired throughput of SDs is Rk,min = 100 Kbits, and the performance backscatter mode and pure active transmission mode are the special
gap between backscatter communications and active transmissions is cases of our proposed hybrid backscatter-active communication mode,
ζ = −20 dB. Besides, we consider a typical scenario with two sensor e.g., the proposed method will be simplified to the pure backscatter
nodes. The AP, power beacon, IRS and SDs are located at [0, 0], [5, 0], mode when we set {tk = 0, ∀k ∈ K}.
[2, 1] and ([3, −1],[4,
√ 0]),respectively. In addition, all the channels can Fig. 3 reveals the impact of the time block length T on the system
be modeled as χ = C0 d−α h, where C0 = −30dB is the path loss at energy efficiency. As expected, we find that the system energy efficiency
an unit distance, d denotes the distance between the transmit node and of all schemes increases with the time block length T . This is because
Authorized licensed use limited to: Indraprastha Institute of Information Technology. Downloaded on June 12,2023 at 09:43:28 UTC from IEEE Xplore. Restrictions apply.
 1388 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 72, NO. 1, JANUARY 2023

requirements of SDs will lead to an unreasonable energy utilization.

Therefore, as shown in Fig. 5, the energy efficiency of all schemes
decreases with the increase of Rk,min .

V. CONCLUSION
This paper investigated the optimal resource scheduling method for
IRS-aided wireless powered hybrid backscatter-active communication
networks. We formulated an energy efficiency maximization problem
with the joint optimization of phase shifts of IRS, time-slot assign-
ment for backscatter communications and active transmissions, and
power control of SDs and power beacon. Furthermore, we proposed
a Dinkelbach-type algorithm to obtain the globally optimal solution
of the formulated energy efficiency optimization problem, and derived
the closed-form expressions for the optimal phase shifts of IRS, and
transmission power of SDs and power beacon. Finally, numerical results
demonstrated that the proposed method can achieve a notably higher
energy efficiency than the baseline methods.

REFERENCES
Fig. 4. System energy efficiency ηEE versus number of reflection
elements M . [1] Z. Qin, F. Y. Li, G. Y. Li, J. A. McCann, and Q. Ni, “Low-power wide-area
networks for sustainable IoT,” IEEE Wireless Commun., vol. 26, no. 3,
pp. 140–145, Jun. 2019.
[2] X. Lu, P. Wang, D. Niyato, D. I. Kim, and Z. Han, “Wireless networks with
RF energy harvesting: A contemporary survey,” IEEE Commun. Surv. Tut.,
vol. 17, no. 2, pp. 757–789, Secondquarter 2015.
[3] D. Li, “Two birds with one stone: Exploiting decode-and-forward relaying
for opportunistic ambient backscattering,” IEEE Trans. Commun., vol. 68,
no. 3, pp. 1405–1416, Mar. 2020.
[4] Y. Ye, L. Shi, X. Chu, and G. Lu, “Throughput fairness guarantee in
wireless powered backscatter communications with HTT,” IEEE Wireless
Commun. Lett., vol. 10, no. 3, pp. 449–453, Mar. 2021.
[5] Y. Xu, B. Gu, and D. Li, “Robust energy-efficient optimization for
secure wireless-powered backscatter communications with a non-linear
EH model,” IEEE Commun. Lett., vol. 25, no. 10, pp. 3209–3213,
Oct. 2021.
[6] Q. Wu and R. Zhang, “Intelligent reflecting surface enhanced wireless
network via joint active and passive beamforming,” IEEE Trans. Wireless
Commun., vol. 18, no. 11, pp. 5394–5409, Nov. 2019.
[7] C. Huang, A. Zappone, G. C. Alexandropoulos, M. Debbah, and
C. Yuen, “Reconfigurable intelligent surfaces for energy efficiency in
wireless communication,” IEEE Trans. Wireless Commun., vol. 18, no. 8,
pp. 4157–4170, Aug. 2019.
[8] B. Lyu, P. Ramezani, D. T. Hoang, S. Gong, Z. Yang, and A. Jamalipour,
“Optimized energy and information relaying in self-sustainable IRS-
empowered WPCN,” IEEE Trans. Commun., vol. 69, no. 1, pp. 619–633,
Fig. 5. System energy efficiency ηEE versus minimum required throughput Jan. 2021.
Rk,min . [9] C. Pan et al., “Intelligent reflecting surface aided MIMO broadcasting for
simultaneous wireless information and power transfer,” IEEE J. Sel. Areas
that a larger T will result in a more reasonable energy utilization. Commun., vol. 38, no. 8, pp. 1719–1734, Aug. 2020.
Moreover, our proposed scheme will achieve higher energy efficiency [10] J. Zuo, Y. Liu, L. Yang, L. Song, and Y.-C. Liang, “Reconfigurable
intelligent surface enhanced NOMA assisted backscatter communica-
gain than the Without IRS scheme, as the time block length T increases. tion system,” IEEE Trans. Veh. Technol., vol. 70, no. 7, pp. 7261–7266,
This observation indicates that the IRS has great potential to improve Jul. 2021.
the energy efficiency of considered systems, especially when the time [11] X. Jia, X. Zhou, D. Niyato, and J. Zhao, “Intelligent reflecting surface-
block length is large. assisted bistatic backscatter networks: Joint beamforming and reflection
design,” IEEE Trans. Green Commun. Netw., vol. 6, no. 2, pp. 799–814,
In Fig. 4, we show the relationship between the system energy
Jun. 2022.
efficiency and the number of reflection elements. As seen from this [12] P. Ramezani and A. Jamalipour, “Backscatter-assisted wireless pow-
figure, we find that except for without IRS scheme, the energy efficiency ered communication networks empowered by intelligent reflecting sur-
achieved by the other IRS-assisted schemes increases with the number face,” IEEE Trans. Veh. Technol., vol. 70, no. 11, pp. 11908–11922,
of reflection elements. Because the additional reflection elements will Nov. 2021.
[13] S. H. Kim and D. I. Kim, “Hybrid backscatter communication for
provide extra phase beamforming gain to improve the energy efficiency wireless-powered heterogeneous networks,” IEEE Trans. Wireless Com-
of backscatter communications and active transmissions. In addition, it mun., vol. 16, no. 10, pp. 6557–6570, Oct. 2017.
is observed that the proposed strategy is able to achieve highest system [14] M. Grant and S. Boyd, “CVX: Matlab software for disciplined convex
energy efficiency than the other benchmark methods, especially when programming, version 2.1,” 2014.
[15] A. Zappone and E. Jorswieck, “Energy efficiency in wireless networks
the IRS equips with a larger number of reflection units.
via fractional programming theory,” Found. Trends Commun. Inf. Theory,
In Fig. 5, we plot the system energy efficiency versus the mini- vol. 11, no. 3/4, pp. 185–396, 2015.
mum required throughput of SDs. In general, the heavy throughput

Authorized licensed use limited to: Indraprastha Institute of Information Technology. Downloaded on June 12,2023 at 09:43:28 UTC from IEEE Xplore. Restrictions apply.