1790 IEEE WIRELESS COMMUNICATIONS LETTERS, VOL. 11, NO.

9, SEPTEMBER 2022

Covert Wireless Communication in Multichannel Systems

Bohan Che , Chang Gao , Ruiqian Ma, Xueqiang Zheng, and Weiwei Yang

Abstract—In this letter, we propose a random sub-channel Willie’s optimal detection performance with the knowledge of
selection covert transmission (RSCT) scheme in multi-channel instantaneous channel state information and statistical chan-
systems to enhance the covertness of communication. We derive nel state information, respectively. Besides, the covertness
the expression of minimum detection error probability (DEP) and
the average effective rate, based on which an optimization is con- performance with finite blocklength was investigated under
structed to maximize the average effective covertness rate (ECR). AWGN channels and the effect of blocklength was analyzed
Numerical results indicate that, compared with the single-channel in [6]. On this basis, scholars tried to improve the performance
scheme, the proposed scheme can indeed enhance the covert- of CWC through various techniques. Reference [7] considered
ness of the communication and the performance is improved a full-duplex receiver sending jamming signals to increase the
as the number of sub-channels increases. Besides, it is demon-
strated that there exists an optimal number of sub-channels for DEP at Willie. Reference [8] showed 3D beamforming can
maximizing the average ECR. improve the covertness performance. Reference [9] improved
the covertness performance by sending artificial noise to dete-
Index Terms—Covert wireless communications, multi-channel
system, finite blocklength. riorate Willie’s detection performance in a untrusted relay
network.
We note that the above investigations are based on single-
I. I NTRODUCTION channel systems. However, multi-channel systems are exten-
sively applied in wireless communications due to the advan-
ITH the rapid growth of wireless communications,
W more and more sensitive information such as personal
privacy and financial data is transmitted via wireless chan-
tage of high effective rate. As multi-channel systems can
provide a new dimension of uncertainty to CWC, it gradu-
ally aroused widespread interest in improving the covertness
nels. Naturally, people have put forward higher demands on
performance. The authors of [10] considered a frequency-
the security of wireless communication. In this context, covert
hopping system with probabilistic sub-channel selection, and
wireless communication (CWC) has drawn increasing atten-
the optimal probability vector is obtained for achieving the
tion [1], [2]. The aim of CWC is to make wireless communica-
enhanced CWC.
tion signals or communication behaviors only be detected with
However, all the above studies are based on single-channel
a ultra-low probability. Bash et al. derived the fundamental
systems, and the CWC in the widespread multi-channel
performance bounds of CWC in additive Gaussian white noise
systems such as frequency hopping systems and orthogo-
(AWGN) channels, i.e., square root law (SRL) [3]. SRT shows
nal frequency division multiplexing (OFDM) systems has not
that the transmitter can covertly and reliably transmits zero
been investigated so far. Reference [11] considered the CWC
bits per channel use when the channel uses N is sufficiently
in a multi-channel system with a full-duplex receiver and
large. Fortunately, scholars have found that positive effective
demonstrated that the ECR can be improved under the con-
covertness rate (ECR) can be achieved when Willie has uncer-
sidered scheme, while in some scenarios with strict delay
tainty on the received signals. Scholars first considered the
requirements.
noise uncertainty at Willie and [4] investigated how to jointly
This letter investigates the CWC in multi-channel systems
optimize the UAV’s trajectory and transmit power to maximize
with finite blocklength. During a fixed transmission period,
the covert rate with noise uncertainty. After that, the uncer-
the transmitter can pick one of K sub-channels to deliver pri-
tainty in channel estimation is inevitable and [5] investigated
vate messages, which further disturbs the warden and raises
Manuscript received 23 March 2022; revised 5 May 2022; accepted the covertness of the communication. However, more sub-
3 June 2022. Date of publication 8 June 2022; date of current version channels imply that the system is occupying up more resources
9 September 2022. This work was supported in part by the National Natural and the average ECR is reduced. In this regard, the num-
Science Foundation of China under Grant 62071486; in part by the Key
Research and Development Program of Jiangsu Province Key Project and ber of sub-channels should be designed carefully in order to
Topics under Grant BE2021095; and in part by the Natural Science Foundation balance the covertness and effectiveness of the system. We
on Frontier Leading Technology Basic Research Project of Jiangsu under derive the closed-form expressions of the minimum detection
Grant BK20212001. The associate editor coordinating the review of this
article and approving it for publication was M. Kishk. (Corresponding author: error probability (DEP), the packet error probability and the
Chang Gao.) average effective rate. And the optimization of transmit power
Bohan Che, Ruiqian Ma, Xueqiang Zheng, and Weiwei Yang are and number of sub-channels is tackled to maximize the aver-
with the College of Communications Engineering, Army Engineering
University of PLA, Nanjing 210007, China (e-mail: [email protected]; age ECR. Through our research, we find that the covertness
[email protected]; [email protected]; [email protected]). performance can be enhanced as the number of sub-channels
Chang Gao is with the College of Information and Communication increases and there exists an optimal number of sub-channels
Engineering, Beijing University of Posts and Telecommunications, Beijing
100876, China (e-mail: [email protected]). to maximize the ECR. It is worth noting that there are some
Digital Object Identifier 10.1109/LWC.2022.3180993 differences between the model considered in this letter and the
2162-2345 
c 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: Military Institute of Science and Technology. Downloaded on January 02,2025 at 04:35:41 UTC from IEEE Xplore. Restrictions apply.
 CHE et al.: COVERT WIRELESS COMMUNICATION IN MULTICHANNEL SYSTEMS 1791

where i = 1, 2, . . . , N denotes the index of channel uses,

l ∈ 1, 2, . . . , K denotes the index of sub-channels, nlw [i ] is the
AWGN at Willie with variance σw 2 , x[i ] ∼ CN (0, P ) is
a
the private message signals, 0 < Pa < Pamax and Pamax is
the maximum transmit power.
When Alice is silent, the received signals at Willie can be
written as
l
H0 : y w [i ] = nlw [i ]. (2)
Fig. 1. The channel model of finite blocklength CWC with multi-channels.
The ultimate goal of Willie is to detect Alice’s transmission
behavior by performing some specific decision rules based
frequency-hopping communication (FH) model. FH sacrifices on the observed samples. However, Willie inevitably suffers a
the communication rate as a cost in order to prevent track- DEP which includes the false alarm probability PFA and the
ing and interfering. CWC requires consideration of a trade-off missed detection probability PMD . We consider the assump-
between communication performance and covert performance. tion of equal priori probabilities (i.e., π1 = π0 = 0.5), and the
As a result, FH usually requires many sub-channels, while DEP is expressed as [12]
CWC studied in this letter only needs a small number of
PE = PMD + PFA . (3)
sub-channels to provide uncertainty.
Naturally, Willie desires to have the minimum DEP P∗E which
II. S YSTEM M ODEL is derived at following. Thus, P∗E ≥ 1 −  as the covert-
A. Communication Scenario ness constraint in the optimization problem below, where  is
A typical three-node CWC system consisting of a legitimate an arbitrarily small positive value that denotes the covertness
transmitter-receiver pair (Alice and Bob) and a warden (Willie) tolerance.
is considered. Specifically, Willie monitors the communication
environment to detect whether Alice conveys private mes- III. P ERFORMANCE A NALYSIS
sages, and Alice aims to reliably transmit private messages A. Detection Performance at Willie
to Bob without being detected by Willie. We suppose that sin-
Since Alice’s transmission state is binary (transmitting or
gle antenna is equipped on each node and the channels are
not transmitting private messages), Willie’s best detector is the
subject to AWGN.
likelihood ratio test (LRT) according to the Neyman-Pearson
As illustrated in Fig. 1, the channel model includes K sub-
lemma [13] which can be expressed as
channels with fixed bandwidth and we assume that one time
slot contains N channel uses of which the number is limited Δ f1 (Yw ) D1
by the maximum channel uses Nmax . When Alice decides to Λ(Yw ) = ≷ 1, (4)
f0 (Yw ) D0
transmit private messages in a time slot, she randomly selects
one sub-channel and encodes D nats information into the N where Yw is defined as the total of samples received by Willie
channel uses, and the encoding rate is given by R = D/N in a time slot, D1 and D0 respectively denote Willie’s deci-
nat/channel use. In other words, if there is a private packet to sions in favor of H1 and H0 , and f1 (Yw ) and f0 (Yw ) are the
be emitted, Alice randomly chooses one sub-channel to emit likelihood function of H1 and H0 , respectively.
the packet with N channel uses and remains silent in other According to (2), the likelihood function of the observations
sub-channels. If there is no private packets to transmit, Alice Yw under H0 is given as follows.
will remain silent in all the sub-channels. This is referred to   NK  K

as a random sub-channel selection covert transmission (RSCT) 1 2 1 
f0 (Yw ) = 2
exp − 2 Yl , (5)
scheme. 2πσw 2σw
l=1
N 2
l [i ]| is the power of N channel uses.
B. Hypothesis Testing at Willie where Yl = i |yw
Willie determines whether Alice is transmitting private mes- Since Willie does not know which sub-channel is selected
sages or not by detecting the received signals in each time slot. by Alice, the likelihood function of the observations Yw under
Therefore, Willie faces a binary hypothesis testing problem H1 is given by
based on the observed samples in each time slot, where the K
K Yk l=1,l=k Yl
null hypotheses H0 denotes that Alice is silent and the alterna- k =1 exp(− 2(P +σ 2 ) −
a w 2σw2 )
tive hypothesis H1 indicates that Alice selects a sub-channel to f1 (Yw ) = N (K −1) N
. (6)
2)
K (2πσw 2 2 )) 2
(2π(Pa + σw
send private messages. Therefore, when Alice selects the k-th
sub-channel to transmit private messages, Willie’s received Then, from (4), the LRT statistic is
signals can be given by
 l K
 Pa Y k
yw [i ] = x[i ] + nlw [i ], l = k , Λ(Yw ) = Ψ
2
e 2σw (Pa +σw ) .
2
(7)
H1 : l [i ] = nl [i ], (1)
yw w l = k , k =1

Authorized licensed use limited to: Military Institute of Science and Technology. Downloaded on January 02,2025 at 04:35:41 UTC from IEEE Xplore. Restrictions apply.
 1792 IEEE WIRELESS COMMUNICATIONS LETTERS, VOL. 11, NO. 9, SEPTEMBER 2022

σ2 N
where Ψ = K1 ( P +σ w
2 ) . The logarithm of (7) is the log-
2 Then, we can obtain the minimum DEP P∗E as
a w
likelihood ratio given as follows. ∗ ∗
γ(N , Nστ2 ) γ(N , PN+σ
τ
2 )
K
 P∗E =1−( w K
) (1 − a
∗
w
), (13)
log Λ(Yw ) = log Ψ + log e μYk
,
Γ(N ) γ(N , Nστ2 )
w
k =1 and the corresponding optimal threshold τ ∗ are denoted as
= log Ψ + LogSumExp(μ{Yk }K
k =1 ), 2 2  2 
(Pa + σw )σw P a + σw
(a) τ∗ = ln
≈ log Ψ + max(μ{Yk }K Pa σ2
k =1 ), ⎛ w ⎞
τ∗
= log Ψ + μ max (Yk ), (8) (Pa + σw 2 2
)σw (K − 1)γ(N , PN+σ 2 )
ln⎝K − ⎠. (14)
a
k ∈{1,...,K } + ∗
w
N Pa γ(N , Nστ2 )
 w

where LogSumExp({YK }K k =1 ) = log K k =1 e

YK and
When K = 1, the above theoretical model degenerates into
2 2
μ = Pa /[2σw (Pa + σw )], step (a) is due to that single-channel systems, and the optimal threshold τ ∗ and cor-
LogSumExp({x (i )}) is an analytic approximation of responding minimum DEP degenerate into [15] with fixed
max({x (i )}) [16]. transmit power, which are given by
Further, according to the Neyman-Fisher factorization the-
2 )σ 2  2 
orem [17], the sub-channel with maximum average power ∗ (Pa + σw w Pa + σ w
τK =1 = ln 2
, (15)
Ymax = N1 maxk ∈{1,...,K } (Yk ) is the approximate sufficient Pa σw
∗
N τK N τ∗
statistic of Willie’s LRT, and the approximate sufficient statis- γ(N , =1
) γ(N , P +σ
K =1
2
σw 2 )
tic detection scheme can be found in [18]. Thus, Willie’s P∗E ,K =1 = 1 − + a w
. (16)
binary hypothesis test can be rewritten as Γ(N ) Γ(N )
It means that the RSCT scheme is a general case of [15] with
D1
Ymax ≷ τ. (9) fixed transmit power.
∗
D0 Remark 1: Based on (13), as γ(N , Nστ2 )/Γ(N ) < 1, the
w
where τ is a predetermined threshold at Willie. minimum DEP P∗E monotonically increases with K. It is due
Then, according to (9), the false alarm probability PFA for to that a higher number of sub-channels K leads to a higher
the threshold τ is given by randomness of sub-channel selection. This implies that, com-
paring to single-channel systems, RSCT scheme can improve
PFA = Pr[Ymax > τ |H0 ] the covertness performance and it can be further improved by
increasing K. Besides, we note from (13) that γ(N , PN+στ∗ )
= 1 − Pr[Ymax ≤ τ |H0 ] a
2
w
K decreases with the transmit power Pa , so that the minimum
= 1− Pr(Yl ≤ τ ) DEP P∗E monotonically decreases with Pa . It means the
l=1 covertness performance reduces as increasing Pa .
 
2N τ K
= 1 − Pr Xs ≤ 2 B. Covert Wireless Communication Design
σw
For the RSCT scheme, Alice randomly selects one sub-
(b)γK N , N τ
2
σw channel to send private messages. Therefore, the signals
= 1− , (10)
ΓK (N ) received at Bob can be given by
 
where γ(α, x ) = 0x e −t t α−1 dt, and Γ(N ) = (N − 1)!, and ybk [i ] = Pa xa [i ] + nkb [i ], (17)
step (b) is due to that Xs is a standard chi-squared random where i = 1, 2, . . . , N denotes the index of channel uses and
variable with 2N degrees of freedom. nb [i ] denotes AWGN at Bob with variance σb2 .
Similarly, the miss detection probability PMD is It is worth noting that since Alice sends private messages
PMD = Pr[Ymax < τ |H1 ] with finite blocklength, the packet error probability cannot be
K
ignored. According to [14], we can obtain an approximation
= Pr(Yl |l=k < τ ) Pr(Yl < τ ) of the packet error probability as
√ 
l=1,l=k N (γ + 1) ln(γ + 1) − R
 2   2  K −1 δ≈Q  , (18)
Pa + σ w σw γ(γ + 2)
= Pr Xs < τ Pr Xs < τ
2N 2N
where γ = Pa /σb2 denotes signal-to-noise ratio (SNR) of
γ N, Nτ γ K −1 N, Nτ  t2
=
2
Pa +σw 2
σw
, (11) the sub-channel at Bob, and Q(x ) = x∞ √1 e − 2 dt is the
2π
ΓK (N ) Gaussian Q-function. As mentioned above, the encoding rate
Then, the DEP at Willie can be expressed as can be given by R = D/N nat/channel use.
Based on (18), the average effective rate is given by [15]
γ K (N , Nτ
2 ) γ K −1 (N , Nτ Nτ
2 )γ(N , Pa +σ 2 )
PE = 1 −
σw
+
σw w
. (12)
R(1 − δ)
ΓK (N ) ΓK (N )
η= . (19)
K
Authorized licensed use limited to: Military Institute of Science and Technology. Downloaded on January 02,2025 at 04:35:41 UTC from IEEE Xplore. Restrictions apply.
 CHE et al.: COVERT WIRELESS COMMUNICATION IN MULTICHANNEL SYSTEMS 1793

Remark 2: According to (18) and (19), we can find that

the average effective rate η monotonically increases with Pa .
Similarly, increasing the number of sub-channels K also leads
to the decreasing of the average effective rate. In combina-
tion with Remark 1, there is a trade-off between covertness
and effectiveness affected by the transmission power Pa and
number of channels K.
Next, we maximize the average ECR under the covertness
constraint by optimizing the transmission power Pa and the
number of channels K, and the optimization problem can be
formulated as
(P1): max η (20a)
K ,Pa
s.t. P∗E ≥ 1 − , (20b)
Pamax> Pa > 0, (20c)
K = 1, 2, . . . , Kmax , (20d)
where (20b) ensures that the CWC requirement is satisfied,
(20c) indicates the range of the transmit power Pa , and (20d)
guarantees that the number of sub-channels K is a positive
integer from 1 to Kmax . Fig. 2. The minimum detection error probability P∗
E versus the number of
sub-channels K with SNR = −10 dB.
Proposition 1: The minimum DEP P∗E monotonically
decreases with respect to Pa for a given K.
Proof: From (12), for any τ , the DEP PE decreases as the
transmission power Pa increases. Assuming that the power at
Alice, Pa1 and Pa2 , satisfies Pa1 < Pa2 , we have
PE (τP∗ a1 , Pa1 ) > PE (τP∗ a1 , Pa2 ) > PE (τP∗ a2 , Pa2 ), (21)
where τP∗ a1 and τP∗ a2 are the optimal detection thresholds for
Pa1 and Pa2 , respectively. The first inequity is owning to
the fact that P∗E is a monotone decreasing function of Pa .
The second inequity is due to the fact that, for Pa2 , PE is
minimized when τ takes the optimal value τP∗ a2 .
When the inequality (20b) takes the equal sign, Pa = Paε
can be obtained by solving the following equation.
Nτ ∗
1 γ(N , Pa∗ +σw2 ) − εΓ(N )
ln( − ∗ ) = K. (22)
ε εγ(N , Nστ2 )
w

Base on Proposition 1, when Paε ≤ Pamax , the optimal trans-

mit power Pa∗ = Paε . Otherwise, the optimal transmit power
Pa∗ = Pamax .
Thus, (P1) can be reformulated as
 Fig. 3. The average effective covertness rate η versus the number of
(P2): max η Pa∗ (23a) sub-channels K with D = 12 nats.
K
s.t. K = 1, 2, . . . , Kmax . (23b)
that increasing the number of sub-channels K means more
Finally, the optimization problem (P2) can be solved by a samples similar to the samples of H0 are added to confuse
simple one-dimensional search. Willie. It implies that the power of idle sub-channel is more
likely to exceeds that of the occupied sub-channel. Thus, the
IV. N UMERICAL R ESULTS covertness performance can be enhanced as the number of
In this section, we provide numerical results to illustrate sub-channels increases and the correctness of Remark 1 is ver-
the performance of the considered scheme under a given ified by simulation. We can also observe for a fixed number
covertness constraint. We set the maximum transmit power of sub-channels K, the minimum DEP decreases as block-
Pamax = 200 dBm, the noise variance at Bob and Willie length N increases. This is because a larger blocklength means
σw2 = σ 2 = −114 dBm [21] and the maximum number of Willie can obtain more samples to improve the detection
b
sub-channels Kmax = 10. accuracy.
In Fig. 2, we first observe that Willie’s minimum DEP Fig. 3 depicts the average ECR versus the number of
P∗E monotonically increases with K. This is due to the fact sub-channels K for different blocklength when the exposure
Authorized licensed use limited to: Military Institute of Science and Technology. Downloaded on January 02,2025 at 04:35:41 UTC from IEEE Xplore. Restrictions apply.
 1794 IEEE WIRELESS COMMUNICATIONS LETTERS, VOL. 11, NO. 9, SEPTEMBER 2022

tolerance  equals to 0.003. In the figure, we can first observe [7] M. Zheng, A. Hamilton, and C. Ling, “Covert communications with
that, as the number of sub-channels K increases, the aver- a full-duplex receiver in non-coherent Rayleigh fading,” IEEE Trans.
Commun., vol. 69, no. 3, pp. 1882–1895, Mar. 2021.
age ECR first increases and then decreases, which implies [8] M. Forouzesh, P. Azmi, N. Mokari, and D. Goeckel, “Covert com-
that there exists an optimal number of sub-channels K ∗ for munication using null space and 3D beamforming: Uncertainty of
maximizing the average ECR. This is due to the fact that Willie’s location information,” IEEE Trans. Veh. Technol., vol. 69, no. 8,
pp. 8568–8576, Aug. 2020.
increasing the number of sub-channels K improves the covert- [9] M. Forouzesh, P. Azmi, A. Kuhestani, and P. L. Yeoh, “Covert commu-
ness performance of the communication but deteriorates the nication and secure transmission over untrusted relaying networks in the
communication performance. The optimal number of sub- presence of multiple wardens,” IEEE Trans. Commun., vol. 68, no. 6,
pp. 3737–3749, Jun. 2020.
channels K ∗ reflects the trade-off between covertness and [10] B. Ning, L. Guan, and H. Huang, “A novel frequency-hopping sequence
effectiveness under the RSCT scheme. for covert communication,” IEEE Access, vol. 5, pp. 20157–20163,
2017.
[11] B. Che, W. Yang, and X. Lu, “Covert communication for multi-channel
V. C ONCLUSION transmission with a full-duplex receiver,” in Proc. 13th Int. Conf.
Wireless Commun. Signal Process. (WCSP), Oct. 2021, pp. 1–5.
In this letter, we first investigate CWC in multi-channel [12] R. Ma, X. Yang, G. Pan, X. Guan, Y. Zhang, and W. Yang, “Covert
systems with the RSCT scheme. Analysis and numerical communications with channel inversion power control in the finite
results indicated that, compared with the single-channel trans- blocklength regime,” IEEE Commun. Lett., vol. 10, no. 4, pp. 835–839,
Apr. 2021.
mission scheme, the proposed RSCT scheme in multi-channel [13] E. Lehmann and J. Romano, Testing Statistical Hypotheses, 3rd ed. New
systems can indeed improve the covertness performance. York, NY, USA: Springer, 2005.
Furthermore, there exists an optimal number of sub-channels [14] B. Yu, Y. Cai, and D. Wu, “Joint access control and resource allocation
for short-packet-based mMTC in status update systems,” IEEE J. Sel.
for maximizing the ECR. Areas Commun., vol. 39, no. 3, pp. 851–865, Mar. 2021.
[15] S. Yan, B. He, X. Zhou, Y. Cong, and A. L. Swindlehurst, “Delay-
R EFERENCES intolerant covert communications with either fixed or random transmit
power,” IEEE Trans. Inf. Forensics Security, vol. 14, pp. 129–140, 2019.
[1] W. Yang, X. Lu, S. Yan, F. Shu, and Z. Li, “Age of information for short- [16] S. Boyd and L. Vandenberghe, Convex Optimization. Cambridge, U.K.:
packet covert communication,” IEEE Wireless Commun. Lett., vol. 10, Cambridge Univ. Press, 2004.
no. 9, pp. 1890–1894, Sep. 2021. [17] M. H. DeGroot and M. J. Schervish, Probability and Statistics. London,
[2] X. Jiang et al., “Covert communication in UAV-assisted air-ground U.K.: Pearson Educ., 2012.
networks,” IEEE Wireless Commun., vol. 28, no. 4, pp. 190–197, [18] B. A. Bash, D. Goeckel, and D. Towsley, “Covert communication gains
Aug. 2021. from adversary’s ignorance of transmission time,” IEEE Trans. Wireless
[3] B. A. Bash, D. Goeckel, and D. Towsley, “Limits of reliable communi- Commun., vol. 15, no. 12, pp. 8394–8405, Dec. 2016.
cation with low probability of detection on AWGN channels,” IEEE J. [19] X. Lu, W. Yang, S. Yan, L. Tao, and D. W. K. Ng, “Joint packet
Sel. Areas Commun., vol. 31, no. 9, pp. 1921–1930, Sep. 2013. generation and covert communication in delay-intolerant status update
[4] X. Zhou, S. Yan, J. Hu, J. Sun, J. Li, and F. Shu, “Joint optimization systems,” IEEE Trans. Veh. Technol., vol. 71, no. 2, pp. 2170–2175,
of a UAV’s trajectory and transmit power for covert communica- Feb. 2022.
tions,” IEEE Trans. Signal Process., vol. 67, no. 16, pp. 4276–4290, [20] R. Ma, W. Yang, L. Tao, X. Lu, Z. Xiang, and J. Liu, “Covert
Aug. 2019. communications with randomly distributed wardens in the finite block-
[5] K. Shahzad and X. Zhou, “Covert wireless communications under length regime,” IEEE Trans. Veh. Technol., vol. 71, no. 1, pp. 533–544,
quasi-static fading with channel uncertainty,” IEEE Trans. Inf. Forensics Jan. 2022.
Security, vol. 16, pp. 1104–1116, 2021. [21] X. Lu, W. Yang, S. Yan, Z. Li, and D. W. K. Ng, “Covertness
[6] X. Yu, S. Wei, and Y. Luo, “Finite blocklength analysis of Gaussian and timeliness of data collection in UAV-aided wireless-powered
random coding in AWGN channels under covert constraint,” IEEE Trans. IoT,” IEEE Internet Things J., early access, Dec. 23, 2021,
Inf. Forensics Security, vol. 16, pp. 1261–1274, 2021. doi: 10.1109/JIOT.2021.3137846.

Authorized licensed use limited to: Military Institute of Science and Technology. Downloaded on January 02,2025 at 04:35:41 UTC from IEEE Xplore. Restrictions apply.