Cyber Security and Applications 1 (2023) 100003

Contents lists available at ScienceDirect

Cyber Security and Applications

An efficient IoT group association and data sharing mechanism in edge

a r t i c l e i n f o a b s t r a c t

Keywords: Despite its benefits and promising future, security and privacy challenges for the IoT wireless communication of
Edge Computing edge computing environment remain unaddressed. As a result, proper authentication mechanisms are critical, es-
Wireless Communication pecially in the extreme scenario where some edge facilities are not functional. For the above consideration, in this
IoT
paper we develop an efficient IoT group association and updating mechanism in edge computing paradigm. The
Security
proposed scheme can provide data transmission and communication guarantees for special practical scenarios.
Privacy-Preserving
The group key updating process in our architecture only necessitates minor changes on the EI side, whereas the
decryption information of some IoT devices remains constant if the devices have not been revoked. The proposed
strategy can accomplish the desired security features, according to the security analysis.

1. Introduction telligent analytical processing services closer to the data source. For ex-
ample, in 2014, more than 30 percent of smart meters from Spain’s three
Because of the inherent open wireless communication features [1– largest electricity service providers were found to have severe security
3], IoT data sharing is vulnerable to a variety of security and privacy vulnerabilities that could be exploited by attackers. Therefore, the at-
problems, particularly in IoT group communication environments with tackers are able to commit electricity fraud or even manage the entire
a large number of participating devices. Advanced security measures circuit systems. Security attacks have been even more damaging in the
and privacy preservation mechanisms are critical for edge infrastruc- industrial IoT space, with the 2018 TSMC manufacturing site attack, the
ture in this circumstances [4,5]. Effective and efficient authentication 2017 ransomware incident, and the 2015 Ukraine massive power out-
mechanism between IoT terminal users (TUs) and the regarding edge age all causing significant damage to targeted industrial IoT systems and
infrastructure (EI) could provide preliminary protection for IoT data ex- individual devices.
change. As a result, various known and unknown secure threats such As for the various types of applications and implementations in the
as eavesdropping, impersonation, and the malicious replaying opera- scope of Internet of Things (IoT), the participating IoT devices will
tions conducted by adversaries can be prevented, which is important for record/generate/transmit a large amount of data involving user privacy.
the current booming 5G industry and the future of 6G communications. Therefore, data security risks are becoming more serious. More than a
Nowadays, because of the proliferation of edge computing techniques, million families and children’s data, conversation recording data, and
the large storing and instant data processing requirements of IoT net- action track data have been leaked since 2015. Hundreds of thousands
work can be satisfied with the cloud and edge infrastructure. of users’ credit card accounts, biometric data, and other personal infor-
Meanwhile, the edge computing infrastructure is recognized as a de- mation were stolen when a vending machine company in the United
centralized computing architecture that shifts applications, correspond- States was hacked in July 2017. Additionaly, a Chinese security firm
ing data, and service processing from the network’s central nodes to the that makes IoT cameras warns that hackers can use the commonly used
network’s logical terminal devices. Edge computing is capable of divid- default credentials to directly access the user devices and even acquire
ing large services that would otherwise be processed and managed to- the sensitive cameras’ live data, which leads to significant security dan-
tally by the central server or node into much smaller, more manageable ger to every customer.
chunks. In the next, it will distribute them to the edge devices for pro- Wireless IoT communication, as an important data transmission
cessing. As a matter of fact, the edge nodes are the local servers that are channel of IoT, has very limited energy, processing capacity, storage
closer to the user-end devices, which can reduce latency and speed up capacity and communication capacity. The overall security reliability
data processing and delivery. Edge computing reduces latency, increases of IoT is harmed by the fact that a large number of traditional devices
efficiency, and improves security and privacy protection by bringing in- lack synchronous configuration protection. Meanwhile, the convergence

∗
Corresponding author.

https://doi.org/10.1016/j.csa.2022.100003
Received 9 February 2022; Received in revised form 16 April 2022; Accepted 16 June 2022
Available online 5 July 2022
2772-9184/© 2022 The Authors. Published by Elsevier B.V. on behalf of KeAi Communications Co., Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/)
H. Tan Cyber Security and Applications 1 (2023) 100003

and diversification of IoT terminals and applications have increased the terward, the pseudonymous authentication-based conditional privacy
security risk associated with the IoT industry. Attackers have a large and protocol PACP [19] is presented by Huang et al.. The improvement in
extensive entry point for network attacks thanks to the growing variety terms of computation and storage cost during the message validation
of IoT interconnected devices, which has resulted in a slew of problems process is achieved. Thereafter, two privacy-preserving authentication
and challenges for IoT. In this case, edge computing technologies can mechanisms [20,21] for secure vehicular communication are respec-
be used to significantly improve the security level of IoT systems based tively proposed in 2014. Subsequently, He et al. developed an identity-
on this, allowing users’ private data to be protected during both the based VANET authentication method without pairing [22]. Accordingly,
interaction and storage processes. the computational complexity of the verification session can be signif-
Many research accomplishments have been accomplished recently, icantly reduced. In 2020, Feng et al. applied the blockchain-assisted
focusing on the IoT safe authentication issue using edge comput- authentication framework in [23] for privacy preservation. Dynamic
ing, which employs a variety of cryptographic design and verification revocation and conditional tracking towards the misbehaving devices
methodologies. It’s worth noting that in some cases, the key generation are enabled. Another attribute-based encryption (ABE) model [24] is
center (KGC) arranges all of the keying information for specific IoT de- developed in order to meet the responding time requirement of edge
vices, potentially causing the key escrow problem. Therefore, it is crit- intelligence-empowered IoV. The proposed ABEM-POD adopts the par-
ical for the IoT device to construct its own partial secret pair and then allel outsourced decryption process, which is of specific usage for the
retain the key information hidden from all other entities, even the key tree access structure. Another attribute-based verification scheme for
generation center. By adopting the unique partial confidential secrets secure data sharing is proposed in [25].
from both the KGC and the device itself, certificateless encryption out-
performs other approaches in this regard. It’s worth noting that neither 3. Preliminaries
the KGC nor the IoT device has access to the other party’s partial secret.
We assume the particular IoT data sharing scenario intended for ex- In this part, the fundamental principles and preliminary knowledge
treme environments, where the edge facilities may be disabled or com- are given so as to facilitate the reader’s understandings.
promised [6–8]. In fact, in most isolated natural landscapes, such as
mountains, desert areas or tropical rain forests, especially in sparsely 3.1. Lagrange polynomial interpolation
populated areas, it’s possible that the edge infrastructures are not al-
ways available. That is, most of the IoT devices of this region that origi-
nally rely on the wireless communication for data sharing may be out of Definition 1 (Degree of Polynomial over. 𝔽𝑝 ) Let 𝔽𝑝 be a finite field,
𝑡
∑
touch. In this case, we utilizes the nearby other IoT devices that are still 𝑃 (𝑥) = 𝜖𝑖 𝑥𝑖 be a non-zero polynomial, where 𝜖𝑡 ≠ 0, the arbitrary pos-
in contact with the edge infrastructure to conduct the message forward- 𝑖=0
ing service. In this case, even in out-of-service region, the IoT devices itive integer 𝑡 is defined as the degree of 𝑃 (𝑥) such that deg 𝑃 (𝑥) = 𝑡.
can still maintain instant data sharing with the remote cloud server. Accordingly, define {(𝑥0 , 𝑦0 ), … , (𝑥𝑗 , 𝑦𝑗 ), … , (𝑥𝑘 , 𝑦𝑘 )} as a set of 𝑘 + 1
Meanwhile, all the nearby IoT devices are capable of constructing the distinctive data points such that ∀𝑚 ≠ 𝑗, 𝑥𝑚 ≠ 𝑥𝑗 . The polynomial 𝑄𝑘 (𝑥)
randomized data forwarding and delivering networks if necessary. of the degree 𝑘 over the finite field 𝔽𝑝 is built according to
𝑘
∑
2. Related work
𝑄𝑘 (𝑥) = 𝑎𝑖 𝑥𝑖 ,
𝑖=0
Nowadays, the topics of data security and user privacy protection to-
wards IoT environments are extensively investigated. Numerous studies where 𝑄𝑘 (𝑥𝑖 ) = 𝑦𝑖 for all 𝑖 = 0, … , 𝑘. The unique Lagrange basis polyno-
regarding authenticated key management and reliable vehicular data mials 𝓁𝑗 (𝑥) (0 ≤ 𝑗 ≤ 𝑘) of degree at most 𝑘 are computed as
exchange of different IoT sceanrios have been conducted so far. As for (𝑥−𝑥0 ) (𝑥−𝑥𝑗−1 ) (𝑥−𝑥𝑗+1 ) (𝑥−𝑥𝑘 )
𝓁𝑗 (𝑥) = (𝑥 −𝑥0 )
…
(𝑥𝑗 −𝑥𝑗−1 ) (𝑥𝑗 −𝑥𝑗+1 )
… (𝑥𝑗 −𝑥𝑘 )
the vehicular communication, in 2012, Lu et al. [9] developed a dy- ∏𝑗𝑘 .
𝑥−𝑥𝑚
namic key updating protocol DIKE to satisfy the privacy-preserving and = 𝑚=0,𝑚≠𝑗 𝑥𝑗 −𝑥𝑚
reliability requirements of location-based VANET services (LBS). The
The corresponding interpolation polynomial 𝐿𝑘 (𝑥) in the Lagrange form
distributed session keys are cooperatively updated by the involved de- ∑
can be defined as 𝐿𝑘 (𝑥) = 𝑘𝑗=0 𝑦𝑗 𝓁𝑗 (𝑥). That is,
vices whenever the revocation process initializes. In [10], the validating
process towards certificate revocation lists (CRLs) in terms of vehicu- 𝑘
( 𝑘 )
∑ ∏ 𝑥 − 𝑥𝑚
lar message authentication is improved with the adopted hash chains. 𝐿𝑘 (𝑥) = 𝑦𝑗 .
𝑥 − 𝑥𝑚
𝑗=0 𝑚=0,𝑚≠𝑗 𝑗
Subsequently, a scalable group key management scheme with message
encryption is proposed by Aliev et al. [11]. Notably, the matrix-based Accordingly, for ∀𝑖 ≠ 𝑗,
encryption algorithm is utilized in the distributed architecture so that
𝑘
∏
enhanced security characteristics and efficiency can be guaranteed. Sim- 𝑥𝑖 − 𝑥𝑚
𝓁𝑗 (𝑥𝑖 ) = = 0,
ilarly, Aman et al. [12] developed a robust IoV authentication scheme 𝑚=0,𝑚≠𝑗
𝑥𝑗 − 𝑥𝑚
with unclonable functions. The approaching devices are verified by the
gateway instead of each RSU. In 2021, Cai et al. [13] proposed a con- and
𝑘
∏
ditional privacy protection mechanism adopting ring signcryption and 𝑥𝑗 − 𝑥𝑚
identity-based cryptosystem. Identities of the misbehaving nodes can be 𝓁𝑗 (𝑥𝑗 ) = =1
𝑚=0,𝑚≠𝑗
𝑥𝑗 − 𝑥𝑚
revealed with the assigned tracking marks. Recently, several authenti-
cated key management (AKM) schemes are developed [14–16]. hold. Hence, the reconstruction of the polynomial 𝑄𝑘 (𝑥) can be per-
Specifically, identity-based and attributed-based cryptographic tech- formed with 𝑘 + 1 distinctive data points on the graph of polynomial
niques have been widely adopted in the authenticated key management 𝑄𝑘 (𝑥) and 𝐿𝑘 (𝑥).
process. A cooperative message authentication and key management
framework is developed in 2011 [17], where decentralized message 3.2. Bilinear pairing
verification tasks are allocated to each legitimate device. Meanwhile,
with the aim to enhance the communication efficiency of emergency Let 𝔾1 and 𝔾2 be the cyclic additive group and multiplicative group
services, Yeh et al. [18] proposed an attributed-based access control generated with prime order 𝑞. The mapping 𝑒̂ ∶ 𝔾1 × 𝔾1 → 𝔾2 is defined
scheme ABACS so that data confidentiality property is provided. Af- as a bilinear pairing with the following characteristics:

2
H. Tan Cyber Security and Applications 1 (2023) 100003

Fig. 1. System Model.

1. Bilinearity: ∀𝑃 , 𝑄, 𝑅 ∈ 𝔾1 and ∀𝑎, 𝑏 ∈ ℤ∗𝑞 , there is among the participating two or more parties, so that the further data
{ exchange can be preserved with the shared keys.
𝑒̂(𝑎𝑃 , 𝑏𝑄) = 𝑒̂(𝑃 , 𝑏𝑄)𝑎 = 𝑒̂(𝑎𝑃 , 𝑄)𝑏 = 𝑒̂(𝑃 , 𝑄)𝑎𝑏 • Conditional Privacy-Preserving: Conditional privacy-preserving is
.
𝑒̂(𝑃 , 𝑄 + 𝑅) = 𝑒̂(𝑄 + 𝑅, 𝑃 ) = 𝑒̂(𝑃 , 𝑄)𝑒̂(𝑃 , 𝑅) made up of two important privacy-related criteria: user privacy pro-
tection and TU identity retrieval. On the one hand, personal data
2. Non-degeneracy: ∃𝑃 , 𝑄 ∈ 𝔾1 such that 𝑒̂(𝑃 , 𝑄) ≠ 1𝔾2 , where 1𝔾2 is the about a user’s genuine identity should be kept safe at all times in
identity element of 𝔾2 . order to prevent malevolent activities like illegal tracing and eaves-
3. Computability: ∀𝑃 , 𝑄 ∈ 𝔾1 , there is an efficient algorithm to calculate dropping on individual TUs. In emergency situations, however, the
𝑒̂(𝑃 , 𝑄). law enforcement agency (LEA) should be able to expose the true TU’s
identity. In other words, user privacy is presumed to be conditional
on the detection and disclosure of compromised or disabled TUs.
3.3. Security objectives
• Mutual Authentication: Mutual authentication is the most basic but
most important security feature of the edge environment, ensuring
The objectives of our design are to enhance the vital security proper-
that all edge entities check each other before the data delivery ses-
ties for edge computing scenarios. The following security requirements
sion begins. The absence of mutual authentication before the com-
should be satisfied.
munication process could result in serious security flaws.
• User Anonymity: Normally, messages originated from the same TU
carry identical or regular patterns, which is likely to expose the link- 3.4. System model
ability. Hence, by analyzing large quantities of eavesdropped mes-
sages, the non-negligible details can be deduced, which seriously In our design, the entire edge infrastructure consists of the cloud
compromise the user privacy. Therefore, anonymous message deliv- server, the edge infrastructure (EI) and multiple terminal users (TUs).
ery is utterly necessary for all the participating TUs. The utilized architecture of the proposed design is shown in Figure 1,
• Session Key Establishment: The shared session keys between edge de- which is considered as a specific edge communication scenario devoted
vices and the system should be established after mutual validation to emergency situations [26].

3
H. Tan Cyber Security and Applications 1 (2023) 100003

In our design, the edge infrastructure (EI) is the essential part of the Thereafter, the data ⟨ð𝑖 , 𝙰𝑖 , 𝙴𝑖 ⟩ is distributed to individual TU 𝐼𝐷𝑖 .
Internet of Things (IoT) system. EI is in charge of significant processes According to our design, it is worth noting that the acquired secret in-
such as system setup, user registration, key management, and verifica- formation ð𝑖 for 𝐼𝐷𝑖 are considered as the unique partial key that is
tion, among others. EI is assumed to be robust to all types of attacks and independently issued by EI, while the remaining partial key value 𝜘𝑖 is
to remain authentic at all times in our design [27]. Because the role of preserved during the entire authentication process. Following this way,
EI is undertaken by commercial groups in the proposed plan, it cannot for 𝐼𝐷𝑖 ∈ 𝑆, the key set
be entirely trusted. We regard the EI to be a trustworthy but suspicious
{(ð1 , 𝜘1 ), (ð2 , 𝜘2 ), … , (ð𝑡 , 𝜘𝑡 )} (3)
authority, where all major generating and identification processes are
carried out properly. Note that the EI only generates a portion of the pri- with 𝑡 participating TUs will be one-to-one related to 𝐼𝐷𝑖 . That is, each
vate key for the registered TUs, while each TU is designed to produce legitimate IoT device maintains the partial key pair for further manage-
the remaining portion of the private key on its own. In this method, the ment. At this point, the offline registration is completed. All the legiti-
key escrow issue is avoided. EI, in particular, provides direct wireless mate TUs collect the distinctive partial key ð𝑖 , along with the interme-
connection for certified devices within its effective range, while devices diate value ⟨𝙰𝑖 , 𝙴𝑖 ⟩. In this phase, the essential communication rounds
outside of its coverage can obtain cellular connectivity indirectly [28– between EI and TUs are performed in order to offer mutual verification.
30]. In our design, the proposed group authentication process is assumed
The terminal users (TUs) are envisioned to be the IoT communica- to be initialized with one broadcast operation conducted by EI. That is,
tion’s terminal users. TUs are involved in data transmission that is routed EI computes Λ = 𝑔 ℑ and broadcast ⟨𝑅𝑒𝑞𝑢𝑒𝑠𝑡, Λ⟩. In the next, each TU
over the EI framework in hostile environments. That is, the participating adopts the stored ⟨ð𝑖 , 𝙰𝑖 , 𝙴𝑖 ⟩, along with the derived Λ to verify whether
TUs not only deliver the messages they generate, but they also convey the following equation holds:
the routed data from other TUs. Even though some devices are outside of
?
the EI coverage, a particular TU’s interaction with surrounding devices 𝑒̂(𝙰𝑖 , Λ)𝙴𝑖 = 𝑒̂(ð𝑖 , Λ). (4)
ensures good connectivity to all remaining TUs. It’s worth emphasizing
If the equation fails, TU stops the procedure and discards the data it
that, for security reasons, the TUs should be verified before accessing
has received. Otherwise, TU produces its own partial secret key 𝜗𝑖 ∈ ℤ𝙿
the edge network [31–33].
at random and computes 𝚃𝑖 = Λ𝜗𝑖 for the next verification session. In
4. Proposed scheme this case, the complete secret key set of TU is presented as ⟨ð𝑖 , 𝜗𝑖 ⟩. The
previous authentication result is stored as
The certificateless group authentication technique is provided in this 𝜂𝑖 = 𝑒̂(𝙰𝑖 , Λ)𝙴𝑖 . (5)
paper with the goal of offering an enhanced authentication scheme for
𝜘
edge communication in an IoT paradigm. The proposed scheme is di- Therefore, the identity 𝑇 𝑖𝑑 𝑖 is derived as 𝑇 𝑖𝑑 𝑖 = 𝙰𝑖 ð𝑖 = −1
𝜁𝑖 𝑖 .
Mean-
vided into two parts: certificateless authentication and group key distri- while, the certificate 𝐴𝑢𝑡ℎ𝑖 can be calculated as 𝐴𝑢𝑡ℎ𝑖 = 𝐻2 (𝑇 𝑖𝑑 𝑖 , 𝚃𝑖 , 𝜂𝑖 ).
bution. In this case, the packet ⟨𝑇 𝑖𝑑 𝑖 , 𝚃𝑖 , 𝐴𝑢𝑡ℎ𝑖 ⟩ is delivered to EI. At this point,
The approach of certificateless authentication and group key man- EI compares the value 𝑇 𝑖𝑑 𝑖 that was delivered before with its database
agement is discussed, with a focus on verification for participating TUs. in order to search for the target TU that matches. It is worth noting that
𝜘 𝜘
The registration, verification, and group key distribution are the three the set {ℵ1 1 , … , ℵ𝑖 𝑖 } are previously calculated such that the repetitive
steps of our authentication design. As a result, the TU registration, as operations are prevented. Afterwards, EI is responsible for validating the
well as certain non-trivial key initialization preparation, takes place dur- correctness of the received 𝐴𝑢𝑡ℎ𝑖 , where the value 𝜂𝑖 can be calculated
ing the offline registration. It’s worth noting that all TUs must register according to 𝜂𝑖 = 𝑒̂(ð𝑖 , Λ). If matches, EI computes
with EI before accessing the edge networks.
𝑔 𝜗𝑖 = 𝚃𝑖 𝑔 −ℑ (6)
In the authentication step, major certificateless authentication solu-
tions are presented. The group key is then generated and transmitted and uses it in the next group key distribution. The design assigns a
in a timely and trustworthy manner. Furthermore, the technique for widely shared secret key to create a universal group communication
group key update is introduced, allowing TUs to manage their member- channel between EI and all legitimate TUs. Message broadcasting be-
ship quickly and efficiently. It’s worth mentioning that our solution uses comes available in this way for practical purposes such as emergency
a certificateless encryption strategy for mutual verification between EI rescue and medical service. Instead of sending the keying message to
and TUs, which eliminates the need for key escrow. The goal of bilin- each device individually, EI sends out a single broadcast to all devices,
ear pairing is to provide increased security features. For practical cir- making key distribution more efficient.
cumstances in a complicated environment, the proposed certificateless We’ll assume that 𝑡 TUs (𝐼𝐷𝑖 ∈ 𝑆) passed the prior EI verification. As
authentication for edge communication is suitable. a result, the group key should be successfully delivered to all TUs, and
The offline registration phase is intended for the initialization includ- outsiders should be unable to deduce the group key via eavesdropping.
ing the essential key information management, and TU registration. Ini- Accordingly, for 𝑖 ∈ [1, 𝑡], EI computes
tially, the bilinear group (𝙿, G, G𝚂 , 𝑒̂) is defined, where 𝙿 is a 𝜆-bit prime,
G and G𝕊 denote two multiplicative cyclic groups with the prime order †𝑖 = 𝐻3 (𝑒̂(𝑔 𝜗𝑖 , ð𝑖 ), 𝜂𝑖 ). (7)
𝙿. Hence the bilinear map 𝑒̂ is constructed as 𝑒̂ ∶ G × G → G𝚂 . 𝑔, 𝑤 ∈ G Hence †𝑖 is related to certain TU so that EI randomly generates the group
are the generators and ℑ ∈ ℤ𝙿 is the randomly generated master key. At key 𝜁 ∈ ℤ𝙿 and constructs
first, the unique license 𝐼𝐷𝑖 is assigned to each TU so that 𝐼𝐷𝑖 ∈ {0, 1}∗
𝑡−1
∑
and 𝑖 ∈ [1, 𝑡], where 𝑡 denotes the total number of the registered TUs.
𝑓 ( 𝑥 ) = ( 𝑥 − †1 ) … ( 𝑥 − †𝑡 ) + 𝜁 = 𝑥 𝑡 + 𝑑𝑖 𝑥𝑖 + 𝑑0 . (8)
Hence the edge device set is defined as 𝑆 = {𝐼𝐷1 , … , 𝐼𝐷𝑡 }. Moreover,
𝑖=1
the cryptographic hash functions 𝐻1 ∶ {0, 1}∗ → G, 𝐻2 ∶ G × G × G𝚂 →
ℤ𝙿 , 𝐻3 ∶ G𝚂 × G𝚂 → G𝙿 and 𝐻4 ∶ G × {0, 1}∗ → ℤ𝙿 are defined. EI com- Notably, for ∀𝑖 ∈ [1, 𝑡], 𝑓 (𝑥) = 𝜁 holds. Subsequently, EI calculates𝛿 =
putes ℵ𝑖 = 𝐻1 (𝐼𝐷𝑖 ) for 𝐼𝐷𝑖 ∈ 𝑆, and then selects ð𝑖 ∈ G and 𝜘𝑖 ∈ ℤ𝙿 , 𝑤ℑ and
with 𝑖 ∈ [1, 𝑡]. EI computes:
℘ = 𝐻4 (ℵ𝑖 , 𝑑0 , … , 𝑑𝑡−1 ) (9)
𝜘
𝙰𝑖 = ð𝑖 ℵ𝑖 𝑖 (1)
and delivers ⟨𝛿, ℘, 𝑑0 , … , 𝑑𝑡−1 ⟩. The validation of the following equation:
and
?
𝙴𝑖 = 𝑒̂(ℵ𝑖 , 𝑔 −𝜘𝑖 )ℑ (2) 𝑒̂(𝛿, 𝑔 ) = 𝑒̂(𝑤, Λ) (10)

4
H. Tan Cyber Security and Applications 1 (2023) 100003

𝜘
is conducted in the next. At this point, TU computes is derived as 𝑇 𝑖𝑑 𝑖 = ℵ𝑖 𝑖 , which contains the randomly generated param-
eter 𝜘𝑖 . In the EI side, 𝜘𝑖 was previously picked. The sending message
†𝑖 = 𝐻3 (𝑒̂(𝑔 𝜗𝑖 , ð𝑖 ), 𝜂𝑖 ) (11)
⟨𝑇 𝑖𝑑 𝑖 , 𝚃𝑖 , 𝐴𝑢𝑡ℎ𝑖 ⟩ bears no resemblance to the data exchange that follows.
and adopts †𝑖 into the calculation as In this method, tracing to a specific device is avoided.
Furthermore, user unlinkability for all participating TUs should be
𝑓 (†𝑖 ) = 𝜁 , (12)
assured, so that the adversary cannot link messages sent by the same TU.
where the distributed group key 𝜁 is derived in TU side. It should be The suggested authentication system ensures that TUs are unlinkable
noted that only validated TUs can obtain the right group key 𝜁 using and resistant to illicit tracing. That is, the attacker will be unable to track
the self-computed †𝑖 . The group key is preserved in this way. down specific TUs by examining the featured identity in transmitted
messages.
5. Security analysis
5.5. Conditional privacy-preservation
The proposed authentication design for edge computing environ-
ment is evaluated in terms of vital security properties as follows. The conditional privacy preservation property is supplied for all le-
gitimate TUs during the whole authenticated key management and data
5.1. Certificateless authentication transfer process. If necessary, the central system can reveal the true iden-
tify of hostile or compromised entities while maintaining the anonymous
Our technique includes the certificateless authentication feature, identity. In the suggested technique, the anonymous session identities
which eliminates the requirement for key escrow. For IoT devices, the are used in each session, but the confidential original TU identities are
proposed protocol can allow certificateless authentication. Malicious in- kept secret at all times.
dividuals are unable to expose the TU’s private key message. Further- Furthermore, as a partial key, the TU chooses a random integer that
more, using the collected knowledge, EI is unable to imitate authentic is only valid and effective for the duration of the current session. On
automobiles. According to the previously introduced authentication pro- a certain TU, impersonation and forgery will fail the final validation.
cess, the self-generated random partial key 𝜗𝑖 ∈ ℤ𝙿 from TU side is kept As a result, the privacy of users is safeguarded. Meanwhile, message
secret to EI all the moment. Moreover, according to DBDH, EI cannot retrieval and identity tracking for the selected suspicious TUs can be ac-
decrypt the correct value of 𝜗𝑖 from the received 𝚃𝑖 = Λ𝜗𝑖 as well. In this complished, revealing the malicious user behaviors and patterns. Over-
scenario, impersonation on a certain TU is not achievable. all, a quite effective authentication and key management system is used
to maximize the privacy-preserving property.
5.2. Session key construction As a result, conditional privacy is preserved. To enhance the privacy-
preserving property, an effective authentication and key management
To provide data confidentiality and transmission security in an IoT design, as well as a unique edge communication approach, are imple-
environment, a shared session key must be produced between the EI mented. When necessary, the central cloud server may also swiftly di-
and all TUs. Our authentication scheme provides the shared session key vulge each device’s true identity. As a result, conditional privacy pro-
𝜁 ∈ ℤ𝙿 between EI and all the validated TUs. In our design, the final tection is achieved.
universal key 𝜁 for group communication is managed by EI and then
delivered to each valid TU in a secure channel. In this case, the 𝜁 is 6. Conclusion
adopted as the group key between EI and all the legitimate devices that
have been verified via mutual authentication. If the existing legitimate In this research, we describe an edge computing-based IoT group
IoT evice has not been revoked by EI, the newly distributed session key association and updating design. The out-of-range IoT devices can be
𝜁 𝑛𝑒𝑤 can easily be derived by the existing devices with the usage of connected to the edge network thanks to the assistance of nearby other
formula 𝑓 (𝑥), where devices. The EI side of our architecture’s group key updating procedure
simply requires modest adjustments, whereas the decryption informa-
𝑓 (†𝑖 ) = 𝜁 𝑛𝑒𝑤 (13)
tion of some IoT devices remains constant if the devices have not been
holds for all the validated TUs. revoked. According to the security assessment, the recommended tech-
nique can achieve the desired security features.
5.3. Replay attack resilience
Acknowledgment
In terms of authentication, the certificate authentication technique
is used, with two partial secret keys created by EI and TU respec- The authors would like to thank...
tively: ⟨ð𝑖 , 𝜗𝑖 ⟩. As a result, after the successful validation of 𝐴𝑢𝑡ℎ𝑖 =
References
𝐻2 (𝑇 𝑖𝑑𝑖 , 𝚃𝑖 , 𝜂𝑖 ), the used partial secret key 𝜗𝑖 ∈ ℤ𝙿 is chosen at random.
In each authenticating session, the partial secret key 𝜗𝑖 , as well as the [1] S. Naveen, M.R. Kounte, Key technologies and challenges in iot edge computing, in:
group key 𝜁 ∈ ℤ𝙿 , is considered the randomly generated value. As a re- 2019 Third International conference on I-SMAC (IoT in Social, Mobile, Analytics and
sult, the previous session’s provided ⟨𝛿, ℘, 𝑑0 , … , 𝑑𝑡−1 ⟩ cannot pass the Cloud) (I-SMAC), 2019, pp. 61–65, doi:10.1109/I-SMAC47947.2019.9032541.
[2] Z. Zhang, X. Guo, Y. Lin, Trust management method of d2d communication based
current validation process. In this approach, the replay attack by the on RF fingerprint identification, IEEE Access 6 (2018) 66082–66087.
malicious entities could be prevented. [3] M. Wang, Z. Yan, Privacy-preserving authentication and key agreement protocols
for d2d group communications, IEEE Trans. Ind. Inf. 14 (8) (2018) 3637–3647.
[4] P. Mendki, Docker container based analytics at iot edge video analytics usecase,
5.4. Identity privacy preservation
in: 2018 3rd International Conference On Internet of Things: Smart Innovation and
Usages (IoT-SIU), 2018, pp. 1–4, doi:10.1109/IoT-SIU.2018.8519852.
An adversary (insider/outsider attacker) could conduct illicit track- [5] M. Cao, L. Wang, H. Xu, D. Chen, C. Lou, N. Zhang, Y. Zhu, Z. Qin, Sec-d2d: A secure
and lightweight d2d communication system with multiple sensors, IEEE Access 7
ing towards a specific device in a real-world communication setting,
(2019) 33759–33770.
risking the user’s privacy. As a result, during the communication pro- [6] M.A. L’opez Peña, I. Muñoz Fernández, Sat-iot: An architectural model for a high-
cess, the true identity of the TU cannot be revealed. performance fog/edge/cloud iot platform, in: 2019 IEEE 5th World Forum on Inter-
The true identity 𝐼𝐷𝑖 of a particular TU 𝑖 is always hidden in our net of Things (WF-IoT), 2019, pp. 633–638, doi:10.1109/WF-IoT.2019.8767282.
[7] D. Kelly, K. Curran, B. Caulfield, Automatic prediction of health status using
design. The one-way hashing function 𝐻1 is used in the authentication smartphone-derived behavior profiles, IEEE J. Biomed. Health Inf. 21 (6) (2017)
phase, as demonstrated. The newly constructed temporary identity 𝑇 𝑖𝑑𝑖 1750–1760.

5
H. Tan Cyber Security and Applications 1 (2023) 100003

[8] M. Alrowaily, Z. Lu, Secure edge computing in iot systems: Review and case stud- [23] Q. Feng, D. He, S. Zeadally, K. Liang, BPAS: blockchain-assisted privacy-preserving
ies, in: 2018 IEEE/ACM Symposium on Edge Computing (SEC), 2018, pp. 440–444, authentication system for vehicular ad hoc networks, IEEE Trans. Ind. Inf. 16 (6)
doi:10.1109/SEC.2018.00060. (2020) 4146–4155.
[9] R. Lu, X. Lin, X. Liang, X. Shen, A dynamic privacy-preserving key management [24] C. Feng, K. Yu, M. Aloqaily, M. Alazab, Z. Lv, S. Mumtaz, Attribute-based encryp-
scheme for location-based services in VANETs, IEEE Trans. Intell. Transp. Syst. 13 tion with parallel outsourced decryption for edge intelligent iov, IEEE Trans. Veh.
(1) (2012) 127–139. Technol. 69 (11) (2020) 13784–13795.
[10] A. Wasef, X. Shen, EMAP: expedite message authentication protocol for vehicular ad [25] K. Fan, Q. Pan, K. Zhang, Y. Bai, S. Sun, H. Li, Y. Yang, A secure and verifiable data
hoc networks, IEEE Trans. Mobile Comput. 12 (1) (2013) 78–89. sharing scheme based on blockchain in vehicular social networks, IEEE Trans. Veh.
[11] H. Aliev, H. Kim, S. Choi, A scalable and secure group key management method for Technol. 69 (6) (2020) 5826–5835.
secure v2v communication, Sensors 20 (21) (2020) 6137. [26] M.O. Ozcan, F. Odaci, I. Ari, Remote debugging for containerized applications in
[12] M.N. Aman, U. Javaid, B. Sikdar, A privacy-preserving and scalable authentication edge computing environments, in: 2019 IEEE International Conference on Edge
protocol for the internet of vehicles, IEEE Internet Things J. 8 (2) (2021) 1123–1139. Computing (EDGE), 2019, pp. 30–32, doi:10.1109/EDGE.2019.00021.
[13] Y. Cai, H. Zhang, Y. Fang, A conditional privacy protection scheme based on ring [27] S.K. Datta, C. Bonnet, An edge computing architecture integrating virtual iot devices,
signcryption for vehicular ad hoc networks, IEEE Internet Things J. 8 (1) (2021) in: 2017 IEEE 6th Global Conference on Consumer Electronics (GCCE), 2017, pp. 1–
647–656. 3, doi:10.1109/GCCE.2017.8229253.
[14] T. Miao, J. Shen, C.-F. Lai, S. Ji, H. Wang, Fuzzy-based trustworthiness evaluation [28] J. Xu, S. Wang, A. Zhou, F. Yang, Edgence: A blockchain-enabled edge-computing
scheme for privilege management in vehicular ad hoc networks, IEEE Trans. Fuzzy platform for intelligent iot-based dapps, China Commun. 17 (4) (2020) 78–87,
Syst. 29 (1) (2021) 137–147. doi:10.23919/JCC.2020.04.008.
[15] P. Wang, Y. Liu, SEMA: secure and efficient message authentication protocol for [29] D. Loghin, L. Ramapantulu, Y.M. Teo, On understanding time, energy and cost
vanets, IEEE Syst. J. 15 (1) (2021) 846–855. performance of wimpy heterogeneous systems for edge computing, in: 2017
[16] L. Wei, J. Cui, Y. Xu, J. Cheng, H. Zhong, Secure and lightweight conditional pri- IEEE International Conference on Edge Computing (EDGE), 2017, pp. 1–8,
vacy-preserving authentication for securing traffic emergency messages in VANETs, doi:10.1109/IEEE.EDGE.2017.10.
IEEE Trans. Inf. Forensics Secur. 16 (1) (2021) 1681–1695. [30] J. Xu, B. Palanisamy, H. Ludwig, Q. Wang, Zenith: Utility-aware resource allocation
[17] Y. Hao, Y. Cheng, C. Zhou, W. Song, A distributed key management framework with for edge computing, in: 2017 IEEE International Conference on Edge Computing
cooperative message authentication in VANETs, IEEE J. Sel. Area. Commun. 29 (3) (EDGE), 2017, pp. 47–54, doi:10.1109/IEEE.EDGE.2017.15.
(2011) 616–629. [31] M. Goudarzi, H. Wu, M. Palaniswami, R. Buyya, An application placement technique
[18] L.-Y. Yeh, Y.-C. Chen, J.-L. Huang, ABACS: an attribute-based access control system for concurrent iot applications in edge and fog computing environments, IEEE Trans.
for emergency services over vehicular ad hoc networks, IEEE J. Sel. Area. Commun. Mobile Comput. 20 (4) (2021) 1298–1311, doi:10.1109/TMC.2020.2967041.
29 (3) (2011) 630–643. [32] Q. Liu, L. Cheng, T. Ozcelebi, J. Murphy, J. Lukkien, Deep reinforcement learning
[19] D. Huang, S. Misra, M. Verma, G. Xue, PACP: an efficient pseudonymous authentica- for iot network dynamic clustering in edge computing, in: 2019 19th IEEE/ACM
tion-based conditional privacy protocol for vanets, IEEE Trans. Intell. Transp. Syst. International Symposium on Cluster, Cloud and Grid Computing (CCGRID), 2019,
12 (3) (2011) 736–746. pp. 600–603, doi:10.1109/CCGRID.2019.00077.
[20] X. Zhu, S. Jiang, L. Wang, H. Li, Efficient privacy-preserving authentication for ve- [33] X. Liu, J. Yu, Z. Feng, Y. Gao, Multi-agent reinforcement learning for resource alloca-
hicular ad hoc networks, IEEE Trans. Veh. Technol. 63 (2) (2014) 907–919. tion in iot networks with edge computing, China Commun. 17 (9) (2020) 220–236,
[21] M. Chuang, J. Lee, TEAM: trust-extended authentication mechanism for vehicular doi:10.23919/JCC.2020.09.017.
ad hoc networks, IEEE Syst. J. 8 (3) (2014) 749–758.
[22] D. He, S. Zeadally, B. Xu, X. Huang, An efficient identity-based conditional priva-
cy-preserving authentication scheme for vehicular ad hoc networks, IEEE Trans. Inf.
Forensics Secur. 10 (12) (2015) 2681–2691.