Content uploaded by Saud Althunibat
Author content
All content in this area was uploaded by Saud Althunibat on Feb 04, 2022
Content may be subject to copyright.
Secure Key Distribution for IoT Networks Based on
Physical Layer Security
Tasneem Alshamaseen∗, Saud Althunibat†, and Marwa Qaraqe‡
∗Department of Computer Engineering, Faculty of Engineering, Al-Hussein Bin Talal University
Ma’an, Jordan, e-mail: tasneem 989@yahoo.com
†Department of Communications Engineering, Faculty of Engineering, Al-Hussein Bin Talal University
Ma’an, Jordan, e-mail: saud.althunibat@ahu.edu.jo
‡Division of Information Computing Technology, College of Science and Engineering, Hamad Bin Khalifa University
Doha, Qatar, e-mail: mqaraqe@hbku.edu.qa
Abstract—Internet of Things (IoT) technology has spread across
many fields. IoT networks contain a large number of devices
that work together and exchange a large amount of information,
which increases the risk of attacks. Data is usually encrypted
using a secret key shared among the communicating nodes to
avoid being revealed to an attacker. Thus, key distribution should
be securely accomplished. However, conventional key distribution
schemes may not be sufficient in the context of IoT networks as
they consume both time and energy from a resource constrained
network. Therefore, there arises a need to distribute different
keys in a more efficient and secure manner. In this paper, a new
scheme is proposed to improving IoT networks security by using
a physical-layer key distribution mechanism to distribute keys
among all nodes within the network. In the proposed scheme,
all the keys are sent simultaneously and only the intended
node can successfully obtain its own key through its estimated
channel phase. Analytical and simulation results prove the high
performance of the proposed scheme and its immunity against
eavesdroppers.
Keywords: Physical layer security · Key distribution · Inter-
net of things
I. INTRODUCTION
The Internet of Things (IoT) technology broadens the tradi-
tional concept of the Internet to become a ubiquitous network
where various objects or devices in the physical world can
interact with each other over the Internet [1]. Such devices are
equipped with sensors, actuators, processors, and transceivers
and communicate with each other (i.e., communication net-
work) to do a wide variety of tasks related to different ap-
plications. The applications of IoT networks cover many fields
including the industrial sector, public safety, smart cities, smart
home applications, traffic management, smart agriculture, for-
est management, tourism and military and security fields [2]. It
has been widely reported that using IoT networks in all of these
applications will reduce human efforts, streamline processes,
and improve the quality of life. However, one of the main
challenges in IoT networks is the attainable security level [3].
As wireless IoT networks contain a huge number of dis-
tributed and connected devices, securing such a network
presents a serious challenge. Among the different security
aspects, focus is usually paid on data confidentiality and
authentication. Data confidentiality is defined as the protection
level given for the exchanged data against potential eavesdrop-
pers located in the surrounding area, while authentication is
defined as the verification of the identity of the message sender.
Both aspects are very important to maintain an acceptable level
of communication security.
The main element of achieving data confidentiality or au-
thentication is to share a secret key among the communicating
ends. This secret key is used for data encryption and decryption
or in verifying the identity of the transmitter for authentication
purposes. Regardless of the purpose of using the key, protect-
ing the key is essential in maintaining network security and
must be distributed to nodes over a secure channel to avoid
eavesdropper interception [4].
Conventional secret key distribution mechanisms might be
not suitable in IoT networks due to the large size of the
IoT networks, which makes key distribution a costly process
[5]. Specifically, because IoT networks are associated with
a large number of connected devices, transmitting the key
to each node individually is time consuming and unrealistic.
Additionally, securing the key during transmission via complex
encryption methods is not feasible due to the limited processing
ability of IoT devices. Moreover, as IoT devices have limited
resources, including energy, increasing the transmit power to
ensure correct reception of the key at the IoT node is energy
inefficient and will impact the battery lifetime of the IoT
device.
Physical layer security (PLS) presents both efficient and
lightweight secure solutions for the different aspects of security
[6], [7]. In PLS, parameters of the physical layer, such as
channel characteristics, modulation, channel coding, bit-to-
symbol mapping, power control and others, are exploited to
attain a secure link between communicators [4]. In this work,
PLS is used to propose an efficient key distribution mechanism
that is suitable for IoT networks. The proposed scheme conveys
all keys to all nodes (or to a subset of the nodes) at the same
time by using the concept of non-orthogonal multiple access
(NOMA) [8], [9]. NOMA allows for concurrent transmission
to multiple nodes, which aids in reducing the time required to
accomplish the process in very efficient time resources. Also,
NOMA shares the transmit power among the intended nodes
[10], [11], which prolongs the network lifetime by improving
energy efficiency. However, conventional NOMA transmission
does not attain the required security, and thus, a key of a node
might be accessible by another node. Therefore, the proposed
key distribution scheme is built on utilizing the physical-layer
parameters to secure the key transmissions without inducing
additional resource consumption or increasing the computa-
tional complexity. Specifically, the proposed scheme relies on
the channel phase between each node an the Central Entity
(CE), which is extracted via a channel estimation process, to
induce a secret phase shift on each transmitted key symbol
such that only the intended node can decode its own key.
The proposed key distribution scheme achieves a high level
of security, minimizes the required overhead, minimizes the
time required to accomplish distribution, minimizes the power
consumption, and minimizes the computational complexity
required either at the CE or the nodes.
A. Related Works
Among the literature works on PLS-based key distribution
mechanisms, the authors in [12] proposed generating a secret
key scheme for each pairwise communication link. Both pairs
estimate their common channel to get the channel coefficients
then generate their identical keys by directly quantizing the
estimate channel phases. In [13], the authors proposed a secret
key agreement method which depends on identical channel
characteristics at the two ends of a channel, such as channel
impulse response or received signal strength (RSS) values and
spatial decorrelation. They use the difference between two RSS
values to extract the value of a secret bit.
In [14], author presented a secret key extraction from the
RSS variations in wireless channels at an adaptive environ-
ment that uses an adaptive lossy quantizer in conjunction
with Cascade-based information reconciliation and privacy
amplification. In [15], authors proposed an efficient physical
layer key generation scheme between transceivers based on
the RSS of signals by applying an adaptive quantization
algorithm for quantifying the measurements. Subsequently,
they implement a randomness extractor to further increase key
generation rate and ensure the randomness of the generated
keys. In another work, the authors of [16] exploited the Radio
Channel Variability (RCV) properties in practical environments
to extract key bits by quantizing the complex channel radio
coefficients and their randomness, then look at the impact of
the carrier frequency, the channel variability in the space, time,
and frequency degrees of freedom used to construct a long
secret key. In [17], the authors developed a key generation
scheme by combining the unique carrier frequency offset
(CFO) with the channel estimates of a pair of nodes. The
resulting key provides a high level of security between the
communicating nodes. Authors in [18] proposed a novel secret
key generation scheme for environments with long coherence
time by controlling fluctuations of the virtual channel. Then
they exploit frequency diversity to guarantee key secrecy. In
[4], authors propose a novel physical-layer key distribution
mechanism for IoT networks. The CE broadcasts several ran-
dom signals as different encryption keys to all nodes within
the network at the same time by exploits the channel diversity
and uniqueness between any two communicators to deliver
the key at each node based on a different modulation order.
This modulation order is chosen at each node based on the
magnitude of channel between the node and the CE, and each
node will have a different key with a different length depending
on the selected modulation. Authors in [19] proposed a key
generation scheme where the key agreement process is divided
into three phases. First, they used the difference between each
legitimate relay node random channels and the eavesdropping
channels to generate a secret key. Secondly, they integrated
the remaining common randomness associated with each relay
pair to generate another key. In the last phase, they join the
generated keys from the previous two phases to get the final
key. For further reading, readers may refer to [20]–[25]
II. SY ST EM MO DE L
The adopted system includes an IoT network of NTnodes
that are grouped into smaller clusters each of Nnodes that are
managed by a CE, as shown in Figure 1. The cluster nodes are
selected such that sufficient spacing between them is achieved,
and all cluster nodes are within the communication range of
the CE. The channel impact between node u(1≤u≤N)
and the CE is represented by the path loss and channel fading,
denoted by guand hu, respectively. huis modeled as a complex
Gaussian random variable with zero mean and unity variance
(Rayleigh fading model), and the path loss guis expressed as
gu=du
d0−η
,(1)
where duis the distance between node uand the CE, ηis the
path loss exponent and d0is a reference distance at which the
path loss is absent.
The exchange of information between the CE and the nodes
must be secured against eavesdroppers. To this end, a secret
key should be used, which is usually sent from the CE towards
nodes. Each node will have its own key that includes Lbits.
The key length is usually a mean feature that prevents them
to be computed, hence, keys are usually long. Therefore, a
key is usually segmented into Nssymbols to be transmitted
consecutively towards the intended node. The symbol length
follows the modulation order, denoted by M, where the symbol
length is equal to log2M.
A. NOMA Overview
As mentioned earlier, NOMA is an efficient multiple access
that allows for serving multiple users on the same frequency
and time resources. Specifically, NOMA implies superimpos-
ing the different modulated symbols into one symbol after
being scaled by different power levels. The resultant symbol is
emitted to towards all nodes, where each node can extract its
Fig. 1: The adopted system model
own symbol by performing Successive Interference Cancella-
tion (SIC) [8], [9].
SIC implies that each node should start decoding the sym-
bols of all nodes that have power levels higher than its level
successively till reaching its own symbol, while the symbols
of other nodes that have power levels lower than its level will
be considered as interference.
In the context of key distribution, key symbols will super-
imposed into one symbol that is transmitted towards intended
nodes, and each node successfully extract its own key by
performing SIC described above. Therefore, employing NOMA
will improve resources efficiency in terms of time and power
resources, where the CE can deliver keys for multiple nodes on
the same time and with the same transmit power. However, the
challenge here is that a node can easily reveal the other nodes’
keys given that keys are un-encrypted. To this end, a secure
key distribution mechanism based on the PLS is proposed in
the next section.
III. PROP OS ED S CH EM E
The proposed scheme relies on a secure channel estimation
process between the CE and each node described as follows.
First, the CE emits a set of channel estimation signals to the
node which uses them to extract the channel gain. The node
then transmits the channel estimation signals back to the CE,
which also extracts the channel gain with the corresponding
node. Based on the assumption that the channel gain will
remain constant, the two estimated values of the channel gains
can be considered identical. Notice that none of the estimated
values will be exchanged over a feedback channel, which
avoids being revealed by an eavesdropper.
Upon estimating the channel phases of all nodes, a phase
shift, to be later used to secure the transmitted key symbol of
each node, is computed for each node (θu) as follows
θu=φu
2π
M∗2π
M,(2)
where φuis the channel phase of the node u. This phase
shift, i.e. θu, is induced to the modulated key symbol of the
corresponding node.
Prior to composing the transmit signal based on NOMA
[9], [10], [26], The CE should compute the power level of
each node as NOMA implied. Among the different methods
of power allocation, we chose to rely on the node distance as
a stable parameter. Therefore, the power level of the node u,
denoted by pu, is given as follows [26]
pu=d2
u
PN
u=1 d2
u
.(3)
Now, the transmitted signal at the tth (1≤t≤Ns)
transmission time, denoted by xt, can be composed as follows
xt=
N
X
u=1 √pus(u)
teθu,(4)
where s(u)
tis the modulated key symbol of the node uat the
tth transmission time.
At the nodes’ sides, each node will receive the signal from
the CE and extracts its own key symbol by using SIC, as
described earlier. Mathematically, the received signal at the
node ucan be expressed as follows
y(u)
t=hupP guxt+wu,(5)
where Pis the total transmit power, and wuis the additive
white complex Gaussian noise at the node uwith zero mean
and Novariance. The detected key symbol at node ucan be
obtained as follows
˜s(u)
t= arg min
s∈S
˜y(u)
t−hupP gupus
2
,(6)
where Sis the set that includes all the possible options
for the transmitted symbol that is drawn from the M-PSK
constellation, and ˜
y(u)
tis given as follows:
˜y(u)
t=y(u)
t−hupP gu
N
X
k=u+1
√pk˜s(k)
t.(7)
Notice that each node after detecting its own key symbol
will multiply it by e−jθuin order to cancel the added phase
shift at the CE. Finally, after detecting all key symbols, each
node will combine the retrieved bit blocks from each detected
key symbol to form the detected key.
It is worth highlighting here that the secret phase shift has
been able to secure the transmitted key symbols such that only
the intended node can get its own key, while other nodes will
not be able to correctly decode the key symbol due to the secret
phase shift.
IV. PERFORMANCE ANALYSIS
In this section, the performance of the proposed scheme is
evaluated mathematically by deriving the Key Error Probability
(KEP), which can be defined for the node uas follows
KEP(u)= 1 −1−SEP(u)Ns
,(8)
where Nsis the number of key symbols and SEP(u)is the
symbol error probability of the node uwhich can be expressed
as
SEP(u)=
MN
X
i=1
Pr.(xi)
MN
X
j=1
Pr.(xi→xj)α(u)
ij (9)
where xiand xj, represent the transmitted signals from the CE
if the decimal value of the whole symbols is equal to iand
j, respectively, Pr.(xi)is the probability that xiis transmitted
and it is considered to identical for all signals, i.e., Pr.(xi) =
1
MN. the probability Pr.(xi→xj)represents the pairwise error
probability that is defined as the probability that xjis detected
given that xihas been transmitted. Finally, α(u)
ij represents a
binary value (either 0or 1) that indicates if the symbols of the
node uin xiand xjare identical (α(u)
ij = 0) or not (α(u)
ij = 1).
Based on (6), Pr.(xi→xj)can be formulated as
Pr.(xi→xj) =Pr.
˜y(u)−hupP gupus(u)
i
2
>
˜y(u)−hupP gupus(u)
j
2,
(10)
where it can be rewritten using (5) and (7) and expanding the
norms to yield
Pr.(xi→xj) = Pr. 2<nhupP gupu∆(u)
ij wuo
>
hupP gupu∆(u)
ij
2
+2<n|hu|2P gu√pu∆(u)
ij (I(u)
i−E(u)
ij )o.
(11)
where
∆(u)
ij =s(u)
i−s(u)
j,
I(u)
i=
u−1
X
k=1
√pks(k)
i,
E(u)
ij =
N
X
k=u+1
√pk(s(k)
i−s(k)
j).
Notice that I(i)
urepresents the interference from the near
nodes, while E(ij)
urepresents the possible error which might
be happened at the far nodes detection process. Also, notice
that 2<nhu√P gupu∆(u)
ij wuois a normal random variable
with zero mean and a variance of 2
hu√P gupu∆(u)
ij
2
No.
Therefore, Q-function can be used to represent the probability
in (10) for a given hu, i.e., Pr.(xi→xj|hu), as follows
Pr. (xi→xj|hu) = Qq|hu|2γ,(12)
where
γ=
P gu√pu
∆(u)
ij
2+ 2Re n∆(u)
ij I(u)
i−E(u)
ij o2
2
∆(u)
ij
2
No
Now, to find the value for Pairwise Error Probability of
any value of hu, we the integral of (12) over the PDF of the
Rayleigh fading channel |hu|2, this can be expressed as
Pr.(xi→xj) = Z∞
0
Qpζγ f|hu|2(ζ).dζ, (13)
where f|hu|2for Rayleigh fading channels is given as
f|hu|2(ζ) = exp(−ζ)for ζ≥0.(14)
Accordingly, to compute Pr.(xi→xj), we substitute (14) in
(13) to yield
Pr.(xi→xj) = Z∞
0
Qpζγ exp(−ζ).dζ, (15)
which can be rewritten using the alternative Craig formula of
the Q-function [27] as follows
Pr.(xi→xj) = 1
πZπ
2
0Z∞
0
exp −ζγ
2 sin2β−ζdζ dβ, (16)
where the inner integral can be solved to yield
Pr.(xi→xj) = 1
πZπ
2
0
2 sin2β
γ+ 2 sin2βdβ, (17)
which can be solved using [28, Eq.65] to be as follows
Pr.(xi→xj) = 1
21−rγ
γ+ 2.(18)
Now, by substituting (18) into (9), SE P (u)can be expressed
in a closed form as follows
SEP(u)=1
MN
MN
X
i=1
MN
X
j=1 1
21−rγ
γ+ 2α(u)
ij ,(19)
Therefore, KEP(u)can be expressed as follows
KEP(u)= 1−
1−
1
2MN
MN
X
i=1
MN
X
j=1 1−rγ
γ+ 2 α(u)
ij
Ns
.
(20)
V. SIMULATION RESU LTS
In this section, results obtained using Monte Carlo Simulation
regarding the performance of the proposed secure key distribution
scheme are explored. Moreover, the results obtained using the derived
are also validated and compared to the simulation results. For the
results included in this section, the key length is set to L= 8 bits, the
modulation used is BPSK, the noise power is set to No=−20dBW,
the path loss exponent is set to η= 3, and the reference distance is
set to do= 100m.
In the first set of results, i.e., Fig. 2, all nodes are assumed honest
nodes where each node aims to detect its own key only. In Fig. 2-(a)
and (b), the KEP at each node is plotted versus the transmit power P
in dBW for the case of N= 2 nodes and N= 3 nodes, respectively.
The first two users in both figures are located at d= 100m, and
d= 500m, while the third user in Fig. 2-(b) is located at d= 1000m.
For sake of comparison and validation, the analytical results obtained
by (20) have been included in both figures. The initial observation
can be made is that the KEP is improved for all users as the transmit
power increases, which is due to improving the received power at each
node. Another observation is that the KEP differs at each node, where
as the node becomes closer to the CE, its KEP performance enhances
which due to the decrease in path loss impact. The last observation
on Fig. 2 is that the analytical results provide good matching with
the simulation curves. The gap at low values of Pis expected due to
the fact the obtained mathematical formula is derived based on the
union bounding technique which acts as an upper bound. However,
as can be seen in all curves, the gap diminishes as the transmit power
increases to reach the exact matching with the simulation curves.
The second set of results are depicted in Fig. 3. In these results,
we consider that one of the IoT nodes, specifically node 2, is an
eavesdropper that attempts to decode the other nodes’ keys. As such,
the KEP performance of Node 2when decoding all keys is shown in
Fig. 3-(a) and (b) for the case of N= 2 and N= 3, respectively.
It can be clearly seen that proposed scheme can prevent node 2from
decoding the other nodes keys where the error performance at node 2
when trying to decode the keys of node 1and node 3is very bad.
On the other hand, the proposed scheme does not impact of the
performance of node 2when decoding its key.
VI. CONCLUSIONS
In this paper, a new scheme has been proposed to distribute the
key in IoT networks based on the physical-layer key. The proposed
scheme has leveraged on the well-known NOMA systems to attain
fast and efficient delivery of the keys. The proposed scheme added a
security level on the top of NOMA transmission to ensure that a key
is delivered to the intended node only. Specifically, the instantaneous
channel phase between a node and the central entity has been used
to induce a secret shift on the transmitted key symbol that cannot
be recovered by any other nodes. The proposed scheme achieves a
high level of security, minimizes the required overhead, minimizes
the time required to accomplish distribution, minimizes the power
consumption, and minimizes the computational complexity required
either at the central entity or the nodes. Simulation and analytical
results prove the high performance of the proposed scheme and its
immunity against eavesdroppers.
ACKNOWLEDGMENT
This research was sponsored in part by the NATO Science for Peace
and Security Programme under grant SPS G5797-PHYSEC.
REFERENCES
[1] K. Zhang et al., “Sybil attacks and their defenses in the internet of
things,” IEEE Internet of Things Journal, vol. 1, no. 5, pp. 372–383,
2014.
[2] P. Sethi et al., “Review Article: Internet of things: architectures, proto-
cols, and applications,” Journal of Electrical and Computer Engineering,
vol. 2017, p. 25, 2017.
[3] A. Tandon et al., “Sentinel based malicious relay detection in wireless
IoT networks,” Journal of Communications and Networks, vol. 21, no. 5,
pp. 458–468, 2019.
[4] M. Alhasanat et al., “A physical-layer key distribution mechanism for
IoT networks,” Mobile Networks and Applications, vol. 25, no. 1, pp.
173–178, 2020.
[5] A. Mukherjee, “Physical-layer security in the Internet of Things: Sensing
and communication confidentiality under resource constraints,” Proceed-
ings of the IEEE, vol. 103, no. 10, pp. 1747–1761, 2015.
[6] S. Althunibat et al., “A physical-layer security scheme by phase-
based adaptive modulation,” IEEE Transactions on Vehicular Technology,
vol. 66, no. 11, pp. 9931–9942, 2017.
[7] S. Althunibat, “Physical-layer entity authentication scheme for mobile
mimo systems,” IET Communications, vol. 12, pp. 712–718(6),
April 2018. [Online]. Available: https://digital-library.theiet.org/content/
journals/10.1049/iet-com.2017.0518
[8] B. M. ElHalawany et al., “Physical-layer security of NOMA systems un-
der untrusted users,” in 2018 IEEE Global Communications Conference
(GLOBECOM). IEEE, 2018, pp. 1–6.
[9] L. Dai et al., “A survey of non-orthogonal multiple access for 5G,” IEEE
communications surveys & tutorials, vol. 20, no. 3, pp. 2294–2323, 2018.
[10] Z. Zhang et al.,Non-orthogonal Multiple Access. Cham: Springer
International Publishing, 2018, pp. 1–4. [Online]. Available: https:
//doi.org/10.1007/978-3-319-32903-1 57-1
[11] Z. Xiang et al., “NOMA-assisted secure short-packet communications in
IoT,” IEEE Wireless Communications, vol. 27, no. 4, pp. 8–15, 2020.
[12] A. Sayeed et al., “Secure wireless communications: Secret keys through
multipath,” in 2008 IEEE International Conference on Acoustics, Speech
and Signal Processing. IEEE, 2008, pp. 3013–3016.
[13] B. Zan et al., “Improving robustness of key extraction from wireless
channels with differential techniques,” in 2012 International Conference
on Computing, Networking and Communications (ICNC), 2012, pp. 980–
984.
[14] S. N. Premnath et al., “Secret Key Extraction from Wireless Signal
Strength in Real Environments,” IEEE Transactions on Mobile Com-
puting, vol. 12, no. 5, pp. 917–930, 2013.
[15] R. Lin et al., “Efficient physical layer key generation technique in wire-
less communications,” EURASIP Journal on Wireless Communications
and Networking, vol. 2020, no. 1, pp. 1–15, 2020.
[16] T. Mazloum et al., “Analysis of Secret Key Randomness Exploiting
the Radio Channel Variability,” International Journal of Antennas and
Propagation, vol. 2015, p. 13, 2015.
[17] M. Jacovic et al., “Evaluation of physical layer secret key generation
for iot devices,” in 2019 IEEE 20th Wireless and Microwave Technology
Conference (WAMICON). IEEE, 2019, pp. 1–6.
[18] P. Huang et al., “Fast secret key generation in static wireless networks: A
virtual channel approach,” in 2013 Proceedings IEEE INFOCOM. IEEE,
2013, pp. 2292–2300.
0 10 20 30 40 50 60
10−4
10−3
10−2
10−1
100
Transmit Power P[dBW]
Key Error Probability (KEP)
Node 1 - Anal ytical
Node 2 - Anal ytical
Node 1 - Simulations
Node 2 - Simulations
0 10 20 30 40 50 60 70
10−4
10−3
10−2
10−1
100
Transmit Power P[dBW]
Key Error Probability (KEP)
Node 1 - Anal ytical
Node 2 - Anal ytical
Node 3 - Anal ytical
Node 1 - Simulations
Node 2 - Simulations
Node 3 - Simulations
(a) (b)
Fig. 2: The average KEP at all nodes versus the transmit power Pin the proposed scheme considering (a) N= 2 and (b) N= 3.
0 10 20 30 40 50 60
10−4
10−3
10−2
10−1
100
Transmit Power P[dBW]
KEP at Node 2
KEP of Node 1’s key at Node 2
KEP of Node 2’s key at Node 2
0 10 20 30 40 50 60
10−4
10−3
10−2
10−1
100
Transmit Power P[dBW]
KEP at Node 2
KEP of Node 1’s key at Node 2
KEP of Node 2’s key at Node 2
KEP of Node 3’s key at Node 2
(a) (b)
Fig. 3: The average KEP of all nodes at an eavesdropper (node 2) versus the transmit power Pin the proposed scheme considering (a)
N= 2 and (b) N= 3.
[19] P. Xu et al., “Physical-Layer Cooperative Key Generation with Correlated
Eavesdropping Channels in IoT,” in 2020 International Conferences on
Internet of Things (iThings) and IEEE Green Computing and Commu-
nications (GreenCom) and IEEE Cyber, Physical and Social Computing
(CPSCom) and IEEE Smart Data (SmartData) and IEEE Congress on
Cybermatics (Cybermatics). IEEE, 2020, pp. 29–36.
[20] S. Jana et al., “On the effectiveness of secret key extraction from wireless
signal strength in real environments,” in Proceedings of the 15th annual
international conference on Mobile computing and networking, 2009, pp.
321–332.
[21] H. Liu et al., “Fast and practical secret key extraction by exploiting
channel response,” in 2013 Proceedings IEEE INFOCOM. IEEE, 2013,
pp. 3048–3056.
[22] B. Azimi-Sadjadi et al., “Robust key generation from signal envelopes
in wireless networks,” in Proceedings of the 14th ACM conference on
Computer and communications security, 2007, pp. 401–410.
[23] J. Zhang et al., “Securing wireless communications of the internet of
things from the physical layer, an overview,” Entropy, vol. 19, no. 8, p.
420, 2017.
[24] O. Abualghanam et al., “A survey of key distribution in the context
of internet of things,” Journal of Theoretical and Applied Information
Technology, vol. 97, no. 22, pp. 3217–3241, 2019.
[25] H. Jin et al., “Physical-layer secret key generation with correlated
eavesdropping channel,” in 2018 IEEE 4th International Conference on
Computer and Communications (ICCC). IEEE, 2018, pp. 226–231.
[26] A. Almohamad et al., “On the Error Performance of Non-orthogonal
Multiple Access Systems,” in 2020 International Conference on Infor-
mation and Communication Technology Convergence (ICTC), 2020, pp.
116–121.
[27] N. Beaulieu, “Generalization of Craig’s Second Formula,” IEEE Com-
munications Letters, vol. 17, no. 3, pp. 433–434, 2013.
[28] M.-S. Alouini et al., “A unified approach for calculating error rates
of linearly modulated signals over generalized fading channels,” IEEE
Transactions on Communications, vol. 47, no. 9, pp. 1324–1334, 1999.
