Secured fog-based vehicular crowd-sensing protocol by Modified Attribute based Encryption Model

Abstract

The contemporary fog computing paradigm evolved from traditional cloud computing to give storage and computation resources at the network’s edge. Fog enabled vehicular processing is anticipated to be a fundamental component that may expedite a variety of applications, notably crowd-sensing, when applied to vehicular networks. As a result, the confidentiality and safety of vehicles participating in the crowd-sensing platform are now recognized as critical challenges for smart police and cyber defence. Furthermore, sophisticated access control is essential to meet the requirements of crowd-sensing users of information. This work presents a novel secured fog-based protocol for vehicular crowd sensing utilizing an attribute-based encryption model. The proposed framework incorporates a two-layered fog architecture termed fog layer A and fog layer B. Fog layer A includes data owners (vehicles) and fog nodes, while fog layer B consists of fog nodes integrated with Road Side Units (RSUs). The Transport Triggered Architecture (TTA) governs the retrieval of regular or specialized data as per the request of data users. The data collection procedure varies based on the type of data being processed. Regular data is gathered from data owners, aggregated using the Multiple Kernel Induced in Kernel Least Mean Square (MKI-KLMS) technique, which employs kernel least mean square and hierarchical fractional bidirectional least mean square methods to handle redundancy in data aggregation. Subsequently, the aggregated data is encrypted using the Proposed Fusion Key for Encryption (PFKE) technique, which employs a fusion key for message encryption. Additionally, an enhanced Blowfish algorithm is applied for an extra layer of encryption on the already encrypted data. The encrypted data is then transmitted to the TTA. The decryption process utilizes the same key to retrieve the original message. The Modified Attribute based Encryption Model (MAEM) scheme achieves the highest efficiency of 90.9476.

Keywords

Data encryption decryption process efficiency protocol redundancy and Blowfish algorithm

Nomenclature

Abbreviation

Description

AES

Advanced Encryption Standard

BPPSVC

Blockchain-based Privacy-Preserving SVM Classification

CP-ABE-UP

Ciphertext-Policy Attribute-Based Encryption with access Update Policy

CCN

Content Centric-Networking

CCA

Chosen Ciphertext Attack

CPA

Chosen Plaintext Attack

DPDR

Differentially Private Double Auction With Reliability

DRMCS

Data Relay Mule–based Collection Scheme

DONA

Data-Oriented Network Architecture

DES

Data Encryption Standard

FEVC

Fog-Enabled Vehicle Crowd-Sensing

ICN

Information Centric Networks

KCA

Key Correlated Attack

KPA

Known Plaintext Attack

LA-SPR

Location Authentication based Secure Participant Recruitment

MCS

Mobile Crowd-Sensing

MMDC

Micro Mobile Data Center

MAEM

Modified Attribute based Encryption Model

MKI-KLMS

Multiple Kernel Induced in Kernel Least Mean Square

PFKE

Proposed Fusion Key for Encryption

RSU

Road Side Unit

TTA

Transport Triggered Architecture

1. Introduction

Cisco announced the concept of fog computing for the first time in 2012 as an infrastructure, which accepts heavy data acquired by final equipment, especially vehicle sensors, and transmits it to local fog nodes situated next to these devices rather than transferring it to the central cloud [1]. Following then, fog computing has been used in a variety of disciplines and studied by researchers in academia as well as industry. Moreover, the fog computing architecture is expected to tackle the problem of expanding resources for computing through conducting statistical analysis at network edges rather than centralized cloud servers [2,3]. Also, Fog-Enabled Vehicle Crowd-Sensing (FEVC) represents a novel idea that enhances Mobile Crowd-Sensing (MCS) by permitting vehicles to gather information collected through embedded sensors and distribute it for ITS-based or additional purposes. Fog-Enabled Vehicle Crowd-Sensing (FEVC) permits fog devices to gather and evaluate data quickly [4,5]. Further, the vehicles gather information and then transmit it to fog devices in FEVC, where it is processed according to the demands of data users consisting of police agencies, traffic authorities, and rescue organizations [6]. As a result, depending on the underlying circumstance, data can be gathered and evaluated at the fog level or core cloud layer [7].

Despite the fact that fog computing infrastructure provides significant capacity for handling shared data, the extensive number of vehicles in the network as well as the shifting wireless connectivity pose a danger to the successful delivery of these data [8,9]. Various academic and business studies have highlighted the constraints of global Internet Protocol(IP) address assignments for vehicle networks. As a result, the vehicular network will probably implement Information Centric Network (ICN) with the goal to increase data transmission. Numerous underpinning structures, which incorporate Content-Centric Networking (CCN) and Data Oriented Network Architecture (DONA), have also been suggested [10,11]. Several research studies accepted and endorsed CCN as an effective method of facilitating data distribution and querying for volatile networks that have substantial mobility, including vehicles networks [12].

A spatial crowd-sensing architecture inspired by fog that strives to accomplish exact assignment of tasks as well as safe crowd-sensing data de-duplication. Despite addressing typical security issues, appropriate authentication and authorization was not addressed, which is critical for the crowdsensing technology considering its function in reducing the communications strain [13,14]. Such fog nodes, tiny clouds, are anticipated to provide reliable information outsourced activities despite compromising the data owner’s confidentiality or privacy. A further major investigation was given to secure crowd-sensing protocol for the internet of vehicles, which addressed security and privacy concerns for crowd-sensing devices for fog-based vehicular networks [15,16]. Nevertheless, the protocol they utilized was built on substantial collective signatures, which compromises the system’s performance [17]. The system they created also took into account traditional control over access, which does not depend on encryption using attributes and hence cannot provide precise authorization for many data users. This makes figuring out the key that was used to encrypt a message the attacker’s primary goal [18]. The aforementioned protocol also collapses to solve the problem of data outsourcing that constitutes a critical idea for crowd-sourced solutions [19,20]. For the greatest degree of the information we have, no complete method for vehicle-based crowd-sensing has been suggested that solves safety, confidentiality, extremely fine control over access, as well as contracting concerns [21]. As a result of the vehicles not interacting with the upper fog nodes and the effective alleviation of the bandwidth deficiency issue when compared to the conventional IP network, the lower level fog nodes can generate a content naming policy that will be utilized by the DU to pull the data from the upper fog layer in a content-centric manner, while also performing ciphertext update functions for data outsourcing. Hence, this paper proposes secured fog based vehicular crowd sensing protocol through attribute based encryption model.

The main contribution of this work is as follows:

–
Proposes PFKE technique for encrypting the message, in which fusion key is induced for encrypting the message and performs decryption with the same key.
–
Developsmodified regular data collection, in which the parameters of encryption are improved with the combination of chaotic map including modified logistic map and tent map.
–
Deploys MKI-KLMS technique, in which it employs kernel least mean square and hierarchical fractional bidirectional least mean square for aggregating redundancy data.

The main organization of this work is as follows:

The rest of these work studied on fog based vehicular crowd sensing protocol, which are implemented with various conventional strategies are tabularized in Section 2. The Section 3 elaborates the proposed model of secured fog based vehicular crowd sensing protocol through attribute based encryption model. The assessment of numerous statistical measures analyzed by comparing various strategies is exhibited in Section 4. Eventually, the conclusion of proposed work is concluded in Section 5.
2. Literature review

In 2021, Foschini, L., et al. [4] has proposed Participant platform, which designs edge nodes to evaluate appropriate complicated crowd capabilities and evaluated the network of federated blockchain to save the state of prices. Edge nodes had been informed of any hazardous situations within their coverage area and may alert smartphone users through a smart push announcement system that prevented users from sending excessive messages through adjusting the cautionary frequency based on the mode of travel and the precise subarea that involved customers were situated.

In 2021, Nkenyereye, L., et al. [7] has developed privacy protection and secure crowd sensing strategy based on fog activated vehicular evaluation. This work comprises of two layered fog nodes, which had the capability to produce crowd sensing work for vehicles that gathers, assessment and aggregates the data in accordance with user specification. Moreover, Ciphertext-Policy Attribute-Based Encryption with access Update Policy (CP-ABE-UP) approach was proposed to encrypt the data. This proposed work ensured both privacy protection and authentication with the utilization of ID based signature approach together with pseudonymous mechanism.

In 2020, Ren, Y., et al. [6] has suggested Micro Mobile Data Center (MMDC) technique for minimizing the completion as well as delay rate of sensing tasks. MMDC minimizes data loss as opposed to typical data gathering methods. Further, MMDC operates as a mobile data centre with a set schedule and itinerary. Data Relay Mule–based Collection Scheme (DRMCS)’s MMDC selection technique depends on a computerized annealing process that takes into account both redundancy and gathering rate.

In 2021, Ni, T., et al. [2] has introduced Differentially Private Double Auction With Reliability (DPDR) approach, in which the incentive technique was deployed with exponential technique for choosing clearing worth tuple. For all clearing price tuples, numerous consistent workers were considered as candidates to gather accurate sensory data. Also, social welfare function was modified with the implementation of utility function together with deploying various practical pricing techniques as well as minimal sensitivity.

In 2020, Smahi, A., et al. [1] has implemented Block chain-based Privacy-Preserving SVM Classification (BPPSVC) approach among data owners of mutually distrusted. Moreover, this proposed work framed the design at first, then adversary system and implement the goals of proposed BPPSVC. Further, the privacy assessment represented that the suggested work was secure and it ensured privacy against Denial of Service (DoS) attacks.

In 2018, Nsikak P. Owoh, et al. [8] has framed security assessment of mobile crowd sensing platform. This paper illustrated the unauthenticated access to position data of user at the time of sensing because of defective security techniques in major sensing strategy. Also, this work suggested advanced encryption 256-Galois counter mode (AES 256-GCM) to guaranteed authentication as well as encryption of sensed information. It also guaranteed peer-to-peer security position as well as moving information from smart phone sensors.

In 2020, Wang, D., et al. [3] has introduced Location Authentication based Secure Participant Recruitment (LA-SPR) strategy to activate all vehicle to execute participant recruitment procedure without showing the exact position as well as sensitive data to requester and revealing key transmission and position authenticity lacking of prior distributed secret. Moreover, this work employed TA to gather location tag of anchor as well as vehicle candidates. Thus, the proposed work attained security, feasibility as well as efficiency with assessment of performance and theoretical validation.

In 2019, A. P., Amritha and J., Sandeep [9] has framed centralized travel scheduling model that links the vehicles to centralized sink to obtain best solution of routes. A prediction model was used to schedule routes. The environment was sampled for several aspects such as congestion, crowd, and the arrival and departure times of other vehicles. For control and management, the system included modules such as mobile nodes, static nodes, sink nodes, and host nodes. A major technique for designing any Vehicular Ad-hoc Network (VANET) system was the choice of a route and protocol.

In 2024, Peng et al. [22] has presented a unique technique for privacy-preserving truth discovery based on multi-party computation that is secure. We use the Secret Sharing method to safely breakdown data and build a Secure Multi-party Computation protocol to compute the ground truth, with the goal of achieving high efficiency and robust privacy protection. Furthermore, a data quality-driven incentive mechanism is constructed by using the weight value produced by truth discovery as a quantitative data quality indicator that dynamically modifies the user’s rewards.

In 2023, Yin et al. [23] has developed, a data collection method (VDCM) for vehicle ad hoc networks (VANETs) crowd sensing is proposed. The mechanism chooses suitable techniques to cut consumption and incorporates two mechanism assumptions. When there are relatively few vehicles or the coalition cannot be formed, it chooses sub mechanism 1; otherwise, it chooses sub mechanism 2. In sub mechanism 1, data is collected using a single aggregate. Submechanism2 employs multi aggregation for data collection, coalition formation method, and auction cooperation agreement for the selection of cooperative vehicles.

In 2022, Omar A et al. [24] has developed a machine learning-enabled intrusion detection system (ESOML-IDS) model and an effective seeker optimization approach is presented for the FC and EC settings. The main purpose of the ESOML-IDS model is to provide a novel ESO-based feature selection (FS) method that selects the best subset of features for detecting intrusions in the FC and EC environments. For the purpose of detecting intrusions, we also used a complete learning particle swarm optimization (CLPSO) with a denoising autoencoder (DAE). Improved detection efficacy and efficiency are the outcome of the creation of the DAE algorithm for parameter optimization and the ESO algorithm for feature subset selection.

Table 1
Features and challenges on traditional techniques correspond to fog based vehicular crowd sensing protocol.

Author [Citation]	Methodology	Features	Challenges
Foschini, L., et al. [4]	– ParticipAct	– It improved the distributed ledger platforms’ robustness.	– It doesn’t incorporate the ability of edge with this crowd-sensing platform system.
Nkenyereye, L., et al. [7]	– CP-ABE-UP	– It increases security and network resources with rejecting unauthorized data owners.	– Need to add participation dependent on incentive to ensure less cryptographic fixed cost.
Ren, Y., et al. [6]	– MMDC	– It achieved better sensing task finishing value.	– Need to leverage the redundancy value and data collection value.
Ni, T., et al. [2]	– DPDR	– It generates greater social welfare rather than other techniques.	– It is necessary to expand the work with concerning other kinds of task.
Smahi, A., et al. [1]	– BPPSVC	– It offers security against DoS attacks.	– Need to analyze blockchain technology with other edge computing approaches.
Nsikak P. Owoh, et al. [8]	– Advanced Encryption Standard 256-Galois Counter Mode	– It guarantees confidentiality as well as authenticity of sensor data.	– Requires high memory utilization to reduce the execution time.
Wang, D., et al. [3]	– LA-SPR	– It accomplished time efficacy and ensures privacy.	– It doesn’t concern privacy preservation of data gathering as well as transmission.
Amritha and J., Sandeep [9]	– Centralized travel scheduling model	– It attains greater packet delivery ratio.	– Need to concern data limitation.
Peng et al. [22]	– Truth discovery scheme	– high efficiency and strong privacy protection.	– Larger datasets are used.
Yin et al. [23]	– data collection mechanism (VDCM)	– Reduce time consumption and get a higher amount of available data.	– More complexity.
Omar A et al. [24]	– Effective Seeker Optimization model	– improved detection efficiency and effectiveness.	– More complexity.

2.1. Problem statement

Table 1 reveals that the features and challenges on traditional techniques correspond to fog based vehicular crowd sensing protocol. The paper [4] improved the distributed ledger platforms’ robustness by deploying ParticipAct approach. Nevertheless, incorporating the ability of edge with this crowd-sensing platform system is complicated issue. Using paper [7], the author achieved increased security and network resources with rejecting unauthorized data owners with CP-ABE-UP technique. Yet, including participation dependent on incentive to ensure less cryptographic fixed cost is difficult. The author employs MMDC mechanism [6] and achieved better sensing task finishing value but still leveraging the redundancy value and data collection value is complicated. Though the author generates greater social welfare rather than other techniques by employing DPDR [2] still it is difficult to expand the work with concerning other kinds of task. Using paper [1], the BPPSVC technique offers security against DoS attacks but still it is necessary to analyze blockchain technology with other edge computing approaches. The paper [8] employs advanced encryption standard 256-Galois counter mode and guaranteed confidentiality as well as authenticity of sensor data. However, it requires high memory utilization to reduce the execution time is difficult. Moreover, the paper deploys LA-SPRtechnique [3] and accomplished time efficacy and ensures privacy. Nonetheless, concerning privacy preservation of data gathering as well as transmission is complicated issue. Consequently, the author utilizes centralized travel scheduling model [9] and attained greater packet delivery ratio but still it is vague in concerning data limitation. As a result, the author uses a centralized journey scheduling model and achieves a higher packet delivery ratio, however the data limitation is still unclear. In order to overcome these limitations, a plan for vehicle-based crowd-sensing devices needs to be investigated. It is necessary to investigate fog computing architectural responses to security, privacy, fine-grained access control, and outsourcing constraints. This work aims to present a modified attribute-based encryption model for a protected fog-based vehicular crowd-sensing system.

3. Proposed model of secured fog based vehicular crowd sensing protocol through PFKE technique

3.1. System model

In fog based vehicular cloud system, the central cloud storages are retained with TTA that stores data and data user that are considered as governmental or non-governmental companies. After this layer, there are two fog layers namely fog layer B and fog layer A, which are implemented sequentially. Both fog layers A and B comprises of fog nodes that are assumed as ${f o g}_{i}^{A}$ and ${f o g}_{i}^{B}$ , respectively. In the fog layer B, there is several RSU represented as fog nodes that are placed in the roadside cloud; whereas, Consider several vehicles can be represented as data owner $D_{i}^{o w n}$ in that $D_{i}^{o w n} : i \to {1, 2, \dots N^{A}}$ where, $N^{A}$ is the number of data owners that gives information via sensors, which are installed in their vehicles in fog layer A and numerous fog nodes can be represented as ${f o g}_{i}^{A}$ such that ${f o g}_{i}^{A} : i \to {1, 2, \dots N^{A}}$ are placed in the fog layer A. The ${f o g}_{i}^{B}$ nodes directed towards data users butare more closeness to vehicles. That is, the upper fog nodes ${f o g}_{i}^{B}$ are adopted to offer proximity to data users; whereas, the lower fog nodes ${f o g}_{i}^{A}$ are adopted to offer proximity to data owners that are vehicles. Initially, TTA registers and stores all information of entities regarding to $D_{o w n}$ and ${f o g}_{i}^{A}$ . The registered vehicles only take part in crowd-sensing and those are permitted to send data. The data user in the central cloud may require crowd-sensing information from the vehicles including national weather agency or authority of transportation. Moreover, the data user may demands two type of data such as regular data or special data. The regular data constitutes a survey on driving behaviour in a particular region, parking, traffic and accident management, etc. On the other hand, the special data constitutes information on instant scenario like natural disasters. If the data user demands regular data, it sends request to TTA and to RSU in fog layer B and then to $D_{i}^{o w n}$ . The data owner provides the corresponding information to ${f o g}_{i}^{A}$ that aggregates the data using MKI-KLMS technique and sends to RSU and then to TTA and then back to data user. Conversely, if the data user demands any special data, it sends request to TTA and sends to RSU and then directly to ${f o g}_{i}^{A}$ . The fog nodes ${f o g}_{i}^{A}$ provide relevant information with aggregated data to RSU and send to TTA and then return to data user. The foremost role of these devices is to offer authentication of data owner and also, ${f o g}_{i}^{A}$ gives the certification of data. Thereby, the data is encrypted via modified cipher policy attribute based encryption approach while transmitting the data. The architecture of proposed secured fog based vehicular crowd sensing protocol is illustrated in Fig. 1. The proposed architecture along with the entities’ core function is elaborated as follows:

–
Data owner: The data owner $D_{i}^{o w n}$ is considered as vehicles (public or private vehicle) including ambulances, public bus or any private vehicles. Each vehicle is registered in TTA under a sign of participation agreement and those vehicles have the capability to communicate with each other due to the sensors installed on the vehicles. While communication, the data (message) are send to ${f o g}_{i}^{A}$ .
–
TTA: The entire system is initialized by the trusted authority. In the system, each entity has particular keys and security parameters that are offered by TTA. Let us consider that TTA is entirely trusted and has the ability of better computation of the system. Also, TTA displays the original entities’ identity that those are participated in the argument.
–
Fog layer A: Along the road side, certain fixed server is positioned in this layer and also it including fog nodes together with sufficient ability of wireless communication and other estimating abilities. According to the demand of data user, these fog nodes gather raw information like regular or special data from the sensors of vehicles. These fog nodes ${f o g}_{i}^{A}$ offer proximity to $D_{i}^{o w n}$ . The major role of ${f o g}_{i}^{A}$ is offering techniques, which guarantees confidentiality and security of data owner and authenticating malicious $D_{i}^{o w n}$ .
–
Fog layer B: This layer comprises of fog nodes ${f o g}_{i}^{B}$ that are also referred as road side cloud nodes and this layer is placed following by TTA. The nodes ${f o g}_{i}^{B}$ are related to an electrical power supply along with adequate processing and storing abilities. This is vital layer due to supplying data to data user via TTA. Moreover, there is another way of providing data to data user. That is, consider that the data user demands data from TTA, the ${f o g}_{i}^{B}$ nodes gather data from ${f o g}_{i}^{A}$ nodes and practice the data to their level of process and then send to TTA and back to data user.
–
Data user: Let assume that the data user be governmental or non-governmental organization including regional disaster management, accident prevention or management department, or electrical power department. According to their needs, the diverse companies collect relevant data that is crowd sensing data, which are provided by data owners. In case of crowd sensing action, the data user determines that the offered information should be pulled in the form of ICN format. That is, the ${f o g}_{i}^{A}$ nodes create a content naming policy, which aids data users to pull information against ${f o g}_{i}^{B}$ nodes.

Figure 1.
Architecture of proposed secured fog based vehicular crowd sensing protocol.
3.2. Proposed PFKE approach for data encryption

PFKE strategy can be composed with the following five sub-algorithms including setup, update, encryption, key generation as well as decryption. The data encryptionprocedures employed previously is not encrypted the message securely. Hence, in this paper, a novel PFKE approach is employed for securely encrypting the message that is elaborated as follows:

–
Setup- $A \lg (1^{ϑ}, P^{U})$ : This setup process is performed with TTA and the findings reveal the parameters of public key andwith deploying an implicit confidentiality parameter and universe attribute $P^{U} .$
–
Key generation- $A \lg (B, M s k)$ : This type of strategy is processed with the authentication, which intakes access structure asalong with mater key Msk. The resultant of this strategy is decryption key pk, which is incorporated with the access structure $B$ .
–
Encryption: The encryption strategy takes input as message Msg and the resultant is $C^{T e x t}$ and it includes $B$ with number of attributes $A_{S} \in P^{U}$ . Also, it needs to assure the parameter of public key.
–
Update- $A \lg (p a r a m s, C^{T e x t}, O_{i n d}, P^{U})$ : This strategy intakes parameter, operation indicator $O_{i n d}$ , ciphertext and universal attributes as input and the resultant can be modified fresh $C^{T e x t^{'}}$ .
–
Decryption: This strategy is processed by the destination side, which takes encrypted ciphertext as input and used decryption key considered as fusion key that decrypts the ciphertext and provides the resultant message as Msg.

3.3. Proposed PFKE scheme implementation

The implementation of PFKE strategy can be described with the following five sub-algorithms including setup, update, encryption, key generation as well as decryption that are elaborated as follows:

3.3.1. Pfke.Setup()

The Pfke.Setup() utilize certain parameter as input like universe attribute as

p^{U} = {1, 2, \dots N}

with the parameter implicit security and provides the resultant as master secret key and Pfke.params. Consider that two multiplicative classes as

G_{1}

and

G_{2}

, which describes bilinear pairing as

\hat{e} : G_{1} \times G_{2} \to G_{T}

such that the prime order

q, g \in G_{1}

and

h \in G_{2}

. Thus, this Pfke.Setup()function creates the parameters with the following strategies.

Step 1: Choose the parameter that encodes and transmits the attribute in $U$ to the components $τ (a t) = δ \in Z_{q}^{*}$

Step 2: Then random values are selected like $α, γ \in Z_{q}^{*}$ and assign $u = g^{α γ}$ as well as $v = e^{(h, g^{α})}$ .

Step 3: Allocate the master secret key as $M s k = (α, γ, g)$ ,

Step 4: Create the parameter as $P f k e . p a r a m s = {u, v, h, G_{T}, G_{1}, G_{2}, τ, U, (h^{α γ^{i}}) i = 0, 1, \dots, n}$ and,

Step 5: Return $⟨ M s k, P f k e . p a r a m s ⟩$ .

3.3.2. Pfke.Keygen()

This strategy takes input as Msk together with

B

can be defined as

P f k e . K e y g e n (P f k e . p a r a m s, M s k, B)

. Moreover, it returns pk, which is incorporated with access structure and that can be defined as follows:

Step 1: hoose $m \in Z_{q}^{*}$ and evaluate $S_{B}^{K} = [{g^{\frac{m}{γ + τ ({a t}_{i})}}} {a t}_{i} \in B, h^{\frac{m - 1}{γ}}]$

Step 2: Return $S_{B}^{K}$ .

3.3.3. Proposed Pfke.Encrypt() steps

This strategy is implemented for encrypting the message Msg, in which it includes

B

with number of attributes

A_{S} \in P^{U}

can be defined as

P f k e . E n c r y p t (W, A_{S}, P f k e . p a r a m s)

. Then the resultant is returned as ciphertext message

C^{M s g}

and this traditional encryption model is described in the following steps.

Step 1: select the random variable $k \in Z_{q}^{*}$ and evaluate the key ${E n c}_{0} = h^{(k - α)} . \prod_{a t \in S} (γ + τ (a t))$ and ${E n c}_{1} = {E n c}_{0}^{γ}, \dots {E n c}_{n - S} = {E n c}_{n - S - 1}^{γ} .$

Step 2: Create $C_{1} = u^{- k}$ .

Step 3: Evaluate $C^{M s g} = v^{k} . M s g$ .

These steps are improved due to lack of confidentiality in data encryption by including fusion key in the ciphertext message to encrypt the message more securely with the following points:

–
The random value is selected from the positive integer set $Z_{q}^{*}$ and evaluates the key ${E n c}_{0}$ with the following formulation as per Eqs (1) and (2).
${E n c}_{0} = h^{(k - α)} . \prod_{a t \in S} (γ + τ (a t))$
(1)

${E n c}_{1} = {E n c}_{0}^{γ}, \dots {E n c}_{n - S} = {E n c}_{n - S - 1}^{γ}$
(2)
–
Create $C_{1} = u^{- k}$ and generate fusion key as $z_{k} = ({E n c}_{0} + C_{1})$ .
–
Evaluate $C^{M s g} = v^{k} . M s g . z_{k}$ .
–
Then return $C^{T e x t} = ({E n c}_{0}, C_{1}, C^{M s g})$ , where $C^{T e x t}$ refers to ciphertext and $C^{M s g}$ refers to cipher message and this $C^{T e x t}$ can be transmit directly for decryption.
–
The encryption extracts the rate $α$ through public parameter.

3.3.4. Pfke.update()

The implementation of update procedure takes ciphertext

C^{T e x t}

and attributes set

p^{U} = {{a t}_{1}^{^{'}}, \dots, {a t}_{n}^{^{'}}}

. Then the resultant is novel ciphertext

C^{T e x t^{'}}

with new access key

S^{'}

are as follows:

Step 1: Evaluate ${E n c}_{0}^{^{'}} = {E n c}_{0}^{F (γ)} = \prod_{i = 0}^{n} {E n c}_{i}^{f i}$ for the functional polynomial can be represented as $J (X) = \prod_{a_{i} \in P^{U}} (X + τ (a t^{'})) = f_{n} X^{n} + f_{n - 1} X^{n - 1} + \dots + f_{0},$

Step 2: Return $C^{T e x t^{'}} = ({E n c}_{0}^{^{'}}, C_{1}, C^{M s g})$ .

3.3.5. Proposed Pfke.dec()

This strategy decrypts

C^{T e x t}

with the parameter

P f k e . d e c (C^{T e x t}, S_{B}^{K}, f k e . p a r a m s)

and this is applicable only if

A_{S}

assures access treeand the traditional decryption model is described as in the following steps.

Step 1: Evaluate aggregate $(g^{\frac{m}{\prod_{a t \in S} (γ + τ ({a t}_{i}))}})$ .

Step 2: For decryption, aggregate strategy is employed and the decryption key relevant to LSSS protocol is resolved as secret key $S_{B}^{K}$ .

Step 3: Evaluate $e^{(h, g)^{k . α . γ}}$ .

Step 4: The decryption evaluates $L_{i} = e^{(C_{1}, h^{\frac{m - 1}{γ}})} . e^{(h, g)^{^{k . α . γ}}}$ and the resultant is $L_{i} = e^{(h, g)^{^{k . α}}}$ .

This step 4 is improved by evaluating $L_{i} = e^{(C_{1}, h^{\frac{m - 1}{γ}})} . e^{(h, g)^{^{k . α . γ}}} . z_{k}$ (i.e. including fusion key to decrypt the ciphertext message) and the resultant is $L_{i} = e^{(h, g)^{^{k . α}}} . z_{k}$

Step 5: The return the message Msg is extracted as $M s g = \frac{C^{M s g}}{e^{(h, g)^{^{k . α}}} . z_{k}}$ .

3.4. Format of ID-based signature

The implementation of an ID-based signature strategy can be described with the following four sub-algorithms including setup, key, sign, and verification which are elaborated as follows:

3.4.1. IdSig.Setup()

With the parameterk, the two multiplicative classes as $G_{1}$ and $G_{2}$ , which describes bilinear pairing as $\hat{e} : G_{1} \times G_{2} \to G_{T}$ such that the prime order $q, g \in G_{1}$ and $h \in G_{2}$ . Then the strategy of IdSig.Setup() provides resultant as Msk and the public attributes are as follows:

Step 1: Select random variable $w \in Z_{q}^{*}$ and evaluate master public key ${P u}^{K} = w . P .$

Step 2: Choose double cryptographic hash function as $V_{1} : [0, 1]^{*} \to G_{1}$ and $V_{2} : [0, 1]^{*} \to Z_{q}^{*}$ .

Step 3: Assign the parameters as $I d S i g . p a r a m s = {G_{1}, G_{2}, q, \hat{2}, V_{1}, V_{2}, P, {P u}^{K}}$ .

Step 4: Return $⟨ w, I d S i g . p a r a m s ⟩$ .

3.4.2. IdSig.key()

The secret value is returned for the provided ID as follows:

Step 1: Create $R_{I D} = V_{1} (I D)$

Step 2: Evaluate $S_{I D}^{K} = w . R_{I D}$

Step 3: Return $⟨ I D, S_{I D}^{K} ⟩$

3.4.3. IdSig.Sign()

This strategy returns message signature ${M s g}^{S}$ with $I d S i g . S i g n ({M s g}^{S}, S_{I D}^{K})$ are as follows:

Step 1: Select random variable $r \in Z_{q}^{*}$ and evaluate $P^{U} = r . P .$

Step 2: Evaluate $c = V_{2} (I D, {M s g}^{S}, P^{U})$ as well as $c_{1} = S_{I D}^{K} + c . r . {P u}^{K} .$

Step 3: Return the message as $χ = (P^{U}, c_{1}) .$

3.4.4. IdSig.verf()

The signature is validated with the function IdSig.verf() such that $η = (P^{U}, c_{1})$ for the message Msg as follows:

Step 1: Evaluate $R_{I D} = V_{1} (I D)$ and $c = V_{2} (I D, {M s g}^{S}, P^{U}) .$

Step 2: verify whether $\hat{e} (c_{1}, P) = \hat{e} (R_{I D} + c . P^{U}, {P u}^{K})$ hold or not.

Step 3: $\hat{e} (c_{1}, P) = \hat{e} (S_{I D}^{K} + c . r, {P u}^{K}, P) = \hat{e} (R_{I D} + c . P^{U}, {P u}^{K})$

3.5. Description of proposed protocol

3.5.1. System initialization

The TTA [7] offers the attributes for each entity in this initialization stage within the system are determined in the following step:

TTA produces IdSig.Setup() that stores the parameters as $⟨ {p k}_{T T A}, I d s i g . p a r a m s ⟩ \leftarrow I d S i g . S e t u p ()$ that retains ${p k}_{T T A}$ , the master private key. Then the parameters IdSig.params are broadcasted openly to everyone.

3.5.1.1. Node registration

The TTA registers all entities within the system and executes the following the steps for data owners $D_{i}^{o w n}$ .

Step 1: TTA assigns a pseudonym psudo $D_{i}^{o w n}$ to all vehicles and for each $D_{i}^{o w n}$ .

Step 2: The TTA process ${p k}_{D_{i}^{o w n}} \leftarrow I d S i g . k e y (D_{i}^{o w n})$ to publish signing key pk that offers pk to $D_{i}^{o w n}$ confidentially.

Step 3: TTA assigns a pseudonym ${p s u d o}_{{f o g}_{i}^{A}}$ for all ${f o g}_{i}^{A}$ .

Step 4: TTA publishestocan be represented as and send the key for eachsecurely.

Step 5: TTA assigns a pseudonym to all vehicles and for each.

Step 6: TTA publishes sk to ${f o g}_{i}^{A}$ can be represented as ${p k}_{{f o g}_{i}^{A}} \leftarrow I d S i g . k e y ({f o g}_{i}^{A})$ and send the key ${p k}_{{f o g}_{i}^{A}}$ for each ${f o g}_{i}^{A}$ securely.

3.5.1.2. Crowd sensing request

In this crowd sensing request, two types of data gathering is performed like regular data or special data. Initially, TTA accepts the demands form data user to offer crowd sensing information for both type of dataas follows:

Step 1: At first, the data user made a demand as $D D = {{p s u d o}_{D_{i}^{o w n}}, t y p, T S, R g, S p}$ . Here, TS refers to time stamp, typ refers to type that it need special or regular data, Rg refers to region that the data come from, and Sp refers to specific operations particularly, which are need to be performed over the data. Also, Sp employs ${f o g}_{i}^{A}$ to rebuild the information in content manner.

Step 2: The data user sends request to TTA as $R_{1} = {E n c}_{{P k}_{T T A}} (γ_{D u}, {p s u d o}_{D u}, D D)$ , where refers to pseudonym of data user, DD refers to data demanding and $γ_{D u} = I d S i g . s i g n (D D, {p k}_{D u})$ constitutes a signature applied with data user over DD.

Step 3: Once the TTA receives, the TTA applies decryption with their private key and following that the signature is verified with exploring $I d S i g . v e r i f y (D D, {p s u d o}_{D u}, γ)$ . After successfully verifying signature, the TTA sends demand for data gathering and send to particular fog nodes or data owners. For the special data collection demand, the work is send to ${f o g}_{i}^{A}$ nodes. Conversely, the work to gather information is sent to $D_{i}^{o w n}$ for the demand of regular data collection.

A. Regular data collection notification

A regular data collection [7] is the process of collecting certain information by the data user within a given time. For illustration, when the data user requires performing survey over the behaviour of diving in a particular region within a given time, then TTA sends demand to $D_{i}^{o w n}$ and provide their information frequently until the given time. TTA performs the following steps while performing this specified case.

Step 1: The universe of parameters like $p^{U} = {1, 2, \dots N}$ is chosen by TTA and publishes common parameters as $⟨ M s k, P f k e . p a r a m s ⟩ \leftarrow P f k e . S e t u p ()$ that declares abeup.params to entire model.

Step 2: TTA selects relevant access framework $Δ_{i}$ that is relevant to $D_{i}^{o w n}$ and transmits it in a secure manner.

Step 3: Then TTA produces a set of attributes as $A_{S}$ such that $A_{S} = {δ_{i}, \dots δ_{n}}$ and creates relevant decryption key by processing $p k = P f k e . k e y g e n (P f k e . p a r a m s, M s k, δ_{i})$ .

Step 4: TTA transmits ${S e t}_{p k} = ⟨ {p k}_{i}, \dots, {p k}_{n} ⟩$ to nodes ${f o g}_{i}^{A}$ .

B. Special data collection notification

In the situation of urgent, the specified urgent relevant data will be gathered like natural disaster [7]. Consider that there is a landslide in particular region; the data user collects the relevant data from the particular region to overcome the congestion. In order to control congestion, the data user urgently gathers specified regions’ information to redirect the vehicles. In this specified case, TTA follows the following steps:

Step 2: TTA selects set of access structures as $A^{S e t} = {Δ_{{R g}_{1}}, Δ_{{R g}_{2}}, \dots, Δ_{{R g}_{n}}}$ that are relevant to particular region.

Step 3: Then TTA produces a set of attributes as $A_{S}$ such that $A_{S} = {δ_{j}, \dots δ_{n}}$ and creates relevant decryption key by processing $p k = P f k e . k e y g e n (P f k e . p a r a m s, M s k, δ_{j})$ .

Step 4: Also, TTA produces another set of attributes as $A_{S_{u p}}$ such that $A_{S_{u p}} = {δ_{U_{j}}, \dots δ_{U_{n}}}$ and creates relevant decryption key by processing $U p k = P f k e . k e y g e n (P f k e . p a r a m s, M s k, δ_{U_{j}})$ .

Step 5: TTA assigns $Q = {{s e t}_{p k}, A^{S e t}, A_{S_{u p}}}$ along with ${s e t}_{p k} = ⟨ {p k}_{j}, \dots {p k}_{n} ⟩$ , where ${s e t}_{p k}$ refers to number of decryption keys, $A^{S e t}$ refers to access structure and $A_{S_{u p}}$ refers to update key. Then TTA transmits $Q$ to ${f o g}_{i}^{A}$ securely.

Step 6: Then the access structure $A^{S e t} = {Δ_{{R g}_{1}}, Δ_{{R g}_{2}}, \dots, Δ_{{R g}_{n}}}$ is issued by ${f o g}_{i}^{A}$ nodes to $D_{i}^{o w n}$ via beacons. Also, the set of attribute $A_{S} = {δ_{j}, \dots δ_{n}}$ remains with ${f o g}_{i}^{A}$ nodes for modifying the access regulations and allocates the key for decryption as ${s e t}_{p k} = ⟨ {p k}_{j}, \dots {p k}_{n} ⟩$ .

Step 7: Further TTA transmits ${U p}_{s e t} = ⟨ {U p k}_{j}, \dots, {U p k}_{n} ⟩$ for ${f o g}_{i}^{B}$ nodes and the decryption key is employed to decrypt the modified ciphertext.

3.5.1.3. Data collection and aggregation

According to the demand of data user, the data collection and aggregation is performed. For regular demand, the ${f o g}_{i}^{A}$ nodes accept the data and aggregate them and transmit to TTA [7]. On the other hand, the ${f o g}_{i}^{A}$ nodes employ $P f k e . u p d a t e ()$ to broadcast the data directly to data user.

A. Modified regular data collection

After receiving the beacon, the crowd-sensing data is transmitted by performing the following steps with $D_{i}^{o w n}$ are as follows:

Step 1: Create a message as $M s g = {T S, l c, D t}$ . Here, TS refers to time stamp, Dt refers to data produced against the sensors and $l c$ refers to location of vehicle.

Step 2: According to access structure $Δ_{{R g}_{n}}$ , $D_{i}^{o w n}$ evaluates $R \leftarrow P f k e . E n c (M s g, P f k e . p a r a m s, Δ_{R g})$ for encrypting the sensor data. This mode of encrypting results in better encryption of data. However, to enhance the encryption process, we include an additional encryption process, i.e., $R \leftarrow P f k e . {E n c}_{2} [{E n c}_{1} (M s g, P f k e, p a r a m s, Δ_{R g})]$ . The newly added encryption process $({E n c}_{2})$ is carried out using Improved blowfish algorithm.

Here, $P f k e . p a r a m s = {u, v, h, G_{T}, G_{1}, G_{2}, τ, U, (h^{α γ^{i}}) i = 0, 1, \dots, n}$ where, $α, γ$ are random values and this conventional parameter $α, γ$ are commonly represented as $ϕ$ and this can be replaced with the proposed combination of chaotic maps namely modified logistic map and tent map. The modified logistic map [25] can be defined as in Eq. (3). Here,refers to control parameter.

φ_{k + 1} = [a . e^{φ_{k}} (1 - e^{φ_{k}})] \mod 1

(3)

Further, the tent map [26] that specifically takes the random chaotic series of values can be defined as in Eq. (4). Here, $p$ refers to transformation value in the range [0, 1].

\begin{aligned} φ_{k + 1} = {\begin{cases} \frac{φ_{k}}{p}, & 0 < φ_{k} ⩽ p \\ \frac{1 - φ_{k}}{p}, & p < φ_{k} < 1 \end{cases} \end{aligned}

(4)

Then the improved chaotic map (i.e. combined form of logistic and tent map) can be defined as in Eq. (5). Here, $T \in [0, 1]$ , $P \in [0, 1]$ and $a \in [0, 4]$ .

\begin{aligned} φ_{k + 1} = {\begin{cases} \frac{φ_{k}}{p}, & 0 < φ_{k} ⩽ p \\ \frac{1 - φ_{k}}{p}, & p < φ_{k} ⩽ T \\ [\frac{a . e^{φ_{k} - T}}{1 - T} (1 - \frac{e^{φ_{k} - T}}{1 - T}) \mod 1], & T ⩽ φ_{k} ⩽ 1 \end{cases} \end{aligned}

(5)

Step 3: Then $D_{i}^{o w n}$ processes $α = I d S i g . s i g n (R, {p k}_{D_{i}^{o w n}})$ and transmits $Q_{1} = {E n c}_{K e y} (α, R)$ , where key refers to shared symmetry key.

Step 4: After accepting $Q_{1}$ , the nodes ${f o g}_{i}^{A}$ decrypts $Q_{1}$ with deploying key and process $I d S i g . v e r i f y (R, {p k}_{_{D_{i}^{o w n}}}, α)$ . If the validation process is performed successfully, the ${f o g}_{i}^{A}$ nodes process $P f k e . d e c ({p k}_{_{{f o g}_{i}^{A}}}, R, P f k e . p a r a m s)$ to restore the original message. For better restoration of original data, we include an additional decryption process, which can be represented by $P f k e . {d e c}_{2} [{d e c}_{1} ({p k}_{{f o g}_{i}^{A}}, R, P f k e, p a r a m s)]$ . The newly added decryption process $({d e c}_{2})$ is carried out using Improved blowfish algorithm.

As we have includednewly added decryption process, if any malicious agent restores the message, it cannot show the vehicles’ identity. The malicious agent may trace the vehicles’ identity, can try to participate in crowd sensing without any authentication, or it may try to extract the contentof message within the crowd-sensing system.

Proposed Blowfish encryption: The key expansion continues to divide the 128-bit key length into many subkey arrays. The redesigned Blowfish decreased its size from 4168 bytes to 2128 bytes. These keys can also be produced and kept independently before any data is encrypted or decrypted. The $P$ -array include 20 32-bit subkeys. The 4 $S$ -Boxes will still have 256 unique entries of 32 bits each. 1.

The initialization of $P$ -array subsequent to 4 S-boxes was performed by means of constant strings consisting of hexadecimal digits of pi.

$P$ 1 is XORed with the initial 32 bits of key, then $P$ 2 is XORed with the subsequent 32 bits, and so on until all of the key’s bits are exhausted, which is up to $P$ 20. Continue the cycle till the entire $P$ -array gets XORed with the key bits.

The Blowfish technique is employed to encrypt an all-zero string with the subkeys defined in the preceding stages (1 and 2).

The result of step 3 replaced $P$ 1 and $P$ 2.

By means of the Blowfish process, step 3 output isencrypted with the revised subkeys.

Outcomesattained in step 5 substituted $P$ 3 and $P$ 4.

The conventional Blowfish method is shown in Eq. (6).

F = ((S_{1} (a) + S_{2} (b) \mod 2^{32}) \oplus S_{3} (c)) + S_{4} (d) \mod 2^{32}

(6)

Figure 2.

Architecture of improved blowfish algorithm.

Here, we have enhanced and proposed a new blowfish algorithm as shown in Fig. 2. As per improved blowfish, the F-function receives a 64-bit data stream and divides it into eight 8-bit segments, where $a$ represents the first eight bits, $b$ represents the second eight bits, and so on until the last eight bits. Convert every 8-bit data bit into a 32-bit value. The initial four 8-bit data streams use the first S-box, whereas the following four use the second S-box. The output from the S-boxes is then XORed or added to produce the final 32-bit value per S-box, which is then concatenated to yield the 64-bit output indicated in Eq. (7).

\begin{aligned} F & = (((S_{1} (a) \times 2 ≫ 1) \oplus [S_{1} (b) \oplus S_{1} (a)] + S_{1} (c) ≫ 1 \oplus S_{1} (d) ≫ 1)) \\ + (((S_{2} (a) \times 3 ≫ 1) \oplus [S_{2} (b) \oplus S_{2} (a)] + S_{2} (c) ≫ 1 \oplus S_{2} (d) ≫ 1)) \end{aligned}

(7)

B. Special data collection

In case of special data, the data user urgently requires data about the particular region [7]. According to TTA specifications, the ${f o g}_{i}^{A}$ nodes broadcast the received data with deploying vehicle to particular or transmit to data user. The nodes ${f o g}_{i}^{A}$ apply the following steps:

Step 1: The ${f o g}_{i}^{A}$ nodes create $U \leftarrow P f k e . U p d a t e (P f k e . p a r a m s, o p t, R, δ_{U_{j}})$ . Here, $δ_{U_{j}}$ refers to set of attribute for update functions.

Step 2: Then ${f o g}_{i}^{A}$ nodes returns $U = ⟨ R^{'} ⟩$ .

Step 3: Then ${f o g}_{i}^{A}$ processes $χ = I d S i g . s i g n (R^{'}, {p k}_{{f o g}_{i}^{A}})$ and transmits $Q_{2} = {E n c}_{K e y} (α, R)$ , where key refers to shared symmetry key.

Step 4: After accepting $Q_{2}$ , the nodedecrypts $Q_{2}$ with key and processes $I d S i g . v e r i f y (R^{'}, {p k}_{_{D_{i}^{o w n}}}, χ)$ . If the validation process is performed successfully, the ${f o g}_{i}^{B}$ nodes process $P f k e . d e c ({p k}_{_{{f o g}_{i}^{B}}}, R, P f k e . p a r a m s)$ to restore the original message.

3.5.1.4. MKI-KLMS technique: Data aggregation for both regular and special data

Data aggregation process is performed for both regular data and special data for eliminating the duplicated data. When the data ${D t}_{m}$ and ${D t}_{j}$ are similar, the transmission process is not performed. Yet, the data is transmitted, when the data are different. To aggregate the duplicate data, this paper proposes MKI-KLMS technique, in which it employs kernel least mean square [27] and hierarchical fractional bidirectional least mean square. Generally, the kernel least mean square utilizes two kernel function namely, Gaussian and polynomial kernel and this traditional kernel function is replaced with other kernel functions including exponential and hyperbolic kernel function. The exponential kernel function can be defined as in Eq. (8). Here, refers to standard deviation.

K_{E x p} (p, q) = \exp [\frac{- ∥ p - q ∥}{2 σ^{2}}]

(8)

Moreover, the hyperbolic kernel function can be defined as in Eq. (9). Here, $K_{H y p} (p, q) = \tanh (α p^{T} q + C)$ refers to constant and $α = \frac{1}{N}$ where, $N$ refers to data dimension.

K_{H y p} (p, q) = \tanh (α p^{T} q + C)

(9)

Then the combined form of proposed MKI-KLMS formulation determines the amalgamation of both exponential and hyperbolic kernel function that can be defined as in Eq. (10).

M K (p, q) = [\frac{K_{E x p} (p, q) + K_{H y p} (p, q)}{2}]

(10)

Thereby, the data is aggregated, encrypted with proposed fusion key and decrypted the original message securely with the same key.

4. Results and discussion

4.1. Simulation procedure

Table 2
Simulation parameters.

Parameter	Values
Parameter Value(s)	area = 100 * 100
Cell size	100 m
Optimal Election Probability of a node to become cluster head	p: float = 0.1
Traffic generation rate	Every 2 s
Physical layer	IEEE 801.1
Maximum number of rounds	rmax = 2000
Alpha_value	1
Imax	50 sec
Imin	1 sec

The proposed Secured fog-based vehicular crowd-sensing protocol was implemented in PYTHON. The simulation Parameters are tabulated in Table 2.

Table 3

Simulation set up.

Device specification
Processor	“AMD Ryzen 5 3450U with Radeon Vega Mobile Gfx 2.10 GHz”
Installed RAM	“16.0 GB (13.9 GB usable)”
Software specification
PYTHON version	3.7

Further, the system configuration was indicated in Table 3. In addition, the MAEM approach was compared with Tripple Data Encryption Standard (DES), DES, Blowfish, Location Authentication based Secure Participant Recruitment (LA-SPR) [3] and Advanced Encryption Standard-Galois Counter Mode (AES 256-GCM) [8], correspondingly. Moreover, it is examined with respect to Latency, Key Sensitivity, Encryption time, Decryption time and other metrics. Similarly, it was evaluated for varied types of attacks like Chosen-Ciphertext Attack (CCA), Chosen-Plaintext Attack (CPA), Known Ciphertext Attack (KCA) and Known Plaintext Attack (KPA).

4.2. Attack analysis on MAEM and conventional methods with regard to CCA, CPA, KCA, KPA, SCA and FIA

Figure 3.

Attack assessment on MAEM and traditional methods a) CCA b) CPA c) KCA and d) KPA for Secured fog-based vehicular crowd-sensing protocol.

The attack assessment on MAEM is contrasted with TRIPPLE DES, DES, ABE, BLOWFISH, LA-SPR [3] and AES 256-GCM [8] for Secured fog-based vehicular crowd-sensing protocol. Further, it is examined with respect to CCA, CPA, KCA and KPA attacks as well as the findings are displayed from Fig. 3(a) to 3(d). Also, the evaluation is carried out for varied key sizes (16, 32, 64 and 128). Moreover, the CCA attack is referred to as, “An attack strategy on cryptanalysis known as a chosen-ciphertext attack (CCA) allows the cryptanalyst to collect data via acquiring the decryptions of specified ciphertexts.” For the enhanced security, the model should acquire lesser attack ratings. Likewise, the MAEM accomplished minimal attack values and afford high security. At the key size 128, the CCA attack value of the MAEM is 0.1954, whilst the conventional methods acquired lesser CCA attack values, including, TRIPPLE DES = 0.4891, DES = 0.6583, ABE = 0.5275, BLOWFISH = 0.3738, LA-SPR = 0.3563 and AES 256-GCM = 0.4356, correspondingly. Furthermore, the CPA is said to as, “A chosen-plaintext attack (CPA) is indeed a type of cryptanalysis attack which assumes the attacker has access to the plaintext ciphers for every given plaintext.The attack’s objective is to obtain data that will render the encryption system less secure.” Here, the MAEM offered lesser CPA attack ratings with higher security to the data. Particularly, the CPA attack rate of the MAEM is much lower than the TRIPPLE DES, DES, ABE, BLOWFISH, LA-SPR and AES 256-GCM at the key size 16.

The KCA attack is defined as “The known ciphertext assault is a cryptanalysis attack type in which the attacker is presumptively limited to a particular set of ciphertexts.” Mainly, for the key size 32, the KCA attack rate of the MAEM scheme is 0.2082, though the TRIPPLE DES is 0.3857, DES is 0.4285, ABE is 0.2916, BLOWFISH is 0.4593, LA-SPR is 0.4981 and AES 256-GCM is 0.3774, correspondingly. The KPA attack analysis is determined as “The known-plaintext attack (KPA) is a type of cryptanalysis assault in which the attacker obtains accessibility to both the plaintext (also known as a crib) and its encrypted version (ciphertext). They are able to be utilized to reveal further hidden information, including secret keys and code books”. Additionally, the KPA attack value of the MAEM is minimal in the key size 64, this is superior to TRIPPLE DES, DES, ABE, BLOWFISH, LA-SPR and AES 256-GCM, respectively. The MAEM strategy exceeded standard techniques by offering lower attack ratings. Therefore, it is determined that the MAEM approach for fog based vehicular crowd might provide extremely high security.

Table 4

FIA and SCA attack analysis on MAEM and conventional methods.

Attacks	Tripple DES	DES	ABE	Blowfish	LA-SPR [3]	AES 256-GCM [8]	Proposed MAEM
FIA	0.524	0.213	0.643	0.447	0.558	0.235	0.1687
SCA	0.764	0.385	0.590	0.797	0.695	0.568	0.2023

Table 4 describes the FIA and SCA attack analysis on MAEM scheme over the conventional strategies for Secured fog-based vehicular crowd-sensing protocol.The SCA is defined as “A side-channel attack is vulnerability in security that affects the hardware or system as a whole instead of the program or its code directly in order to gain information from or modify the program execution of a system. “Mainly, the SCA attack rate of the proposed approach is 0.211, though the Tripple DES, DES, ABE, Blowfish, LA-SPR and AES 256-GCM gained greater SCA attack values. The FIA attack is referred to as, “The act of purposefully introducing flaws into a system is known as fault injection. Analyzing the system’s performance under stress is the aim of this technique. Engineers of hardware and software frequently cause flaws in their products for a variety of reasons.’ Further, the FIA attack rate of the proposed scheme is 0.168, whilst the Tripple DES is 0.524, DES is 0.213, ABE is 0.643, Blowfish is 0.447, LA-SPR is 0.558 and AES 256-GCM is 0.234, correspondingly.

4.3. Encryption and decryption time analysis on MAEM and conventional schemes

Figure 4.

Validation on MAEM and traditional strategies for Secured fog-based vehicular crowd-sensing protocol a) Decryption time and b) Encryption time.

Figure 4(a) and (b) represents the decryption and encryption time assessment on the proposed MAEM is compared with TRIPPLE DES, DES, ABE, BLOWFISH, LA-SPR [3] and AES 256-GCM [8]. For the effective performance of the model, it should accomplish minimal encryption and decryption time. Likewise, the MAEM recorded lesser values in all the key size. Particularly, the encryption time of the proposed approach is 0.00468s (Key Size = 32), even though the traditional techniques gained highest encryption time, notably, TRIPPLE DES = 0.0056s, DES = 0.0059s, ABE = 0.0058s, BLOWFISH = 0.0057s, LA-SPR = 0.0048s and AES 256-GCM = 0.0049s, correspondingly. Regarding the Fig. 4(a), the decryption time of the MAEM is minimal than TRIPPLE DES, DES, ABE, BLOWFISH, LA-SPR and AES 256-GCM. Mainly, for the key size 16, the proposed method yielded the decryption time of 0.00402s, which is minimal than TRIPPLE DES, DES, ABE, BLOWFISH, LA-SPR and AES 256-GCM. As a consequence, the MAEM exceeded other previous strategies with low encryption and decryption time, proving that it has a greater potential to protect the data in a fog-based vehicular crowd sensing protocol.

4.4. Key sensitivity and Latency analysis on MAEM and conventional schemes

Table 5
Computation time evaluation on MAEM and traditional methodologies.

Methods	Computation time
TRIPPLE DES	1.8451
DES	1.9675
ABE	3.1255
BLOWFISH	8.7533
LA-SPR [3]	5.6487
AES 256-GCM [8]	2.1549
Proposed	1.6885

Figure 5.

Validation on MAEM and traditional strategies for Secured fog-based vehicular crowd-sensing protocol a) Key Sensitivity and b) Latency.

The examination on the proposed MAEM over the TRIPPLE DES, DES, ABE, BLOWFISH, LA-SPR [3] and AES 256-GCM [8] with regard to key sensitivity and latency for secured fog-based vehicular crowd-sensing protocol is shown in Fig. 5(a) and 5(b). Moreover, the key sensitivity and latency must be lower for efficacious performance of the model. For the key size 64, the MAEM offered the key sensitivity of 0.1271, this is extremely lower than TRIPPLE DES (0.2476), DES (0.2819), ABE (0.2265), BLOWFISH (0.2658), LA-SPR (0.2597) and AES 256-GCM (0.2489), correspondingly. Considering the Fig. 5(b), the latency of the MAEM is higher over TRIPPLE DES, DES, ABE, BLOWFISH, LA-SPR and AES 256-GCM at the key size 128. Altogether, the MAEM has asserted its supremacy for secured fog-based vehicular crowd sensing protocol with lesser key sensitivity and latency and this is because of the improved data aggregation and modified attribute based encryption method.

4.5. Computation time analysis on MAEM and conventional strategies

Table 5 explains the computation time evaluation on proposed MAEM is contrasted over the TRIPPLE DES, DES, ABE, BLOWFISH, LA-SPR [3] and AES 256-GCM [8] for secured fog-based vehicular crowd-sensing protocol. The computation time need to be lower for secure data transmission in fog-based vehicular crowd sensing protocol. Further, the computation time of the MAEM methodology is 1.6885, mean while the TRIPPLE DES is 1.8451, DES is 1.9675, ABE is 3.1255, BLOWFISH is 8.7533, LA-SPR is 5.6487 and AES 256-GCM is 2.1549, correspondingly.

4.6. Efficiency and availability analysis on MAEM and conventional schemes

Table 6
Efficiency evaluation on MAEM and traditional strategies.

Methods	Efficiency(%)
TRIPPLE DES	56.6456
DES	61.6589
ABE	48.3554
BLOWFISH	51.6346
LA-SPR [3]	69.5525
AES 256-GCM [8]	78.5526
Proposed	90.9476

The efficiency evaluation on MAEM is examined over TRIPPLE DES, DES, ABE, BLOWFISH, LA-SPR [3] and AES 256-GCM [8] for Secured fog-based vehicular crowd-sensing protocol is illustrated in Table 6. Moreover, the greatest efficiency is acquired using the MAEM scheme is 90.9476, although the existing methods scored lower efficiency, notably, TRIPPLE DES = 56.6456, DES = 61.6589, ABE = 48.3554, BLOWFISH = 51.6345, LA-SPR = 69.5525 and AES 256-GCM = 78.5526, correspondingly.

Table 7

Availability evaluation on MAEM and traditional strategies.

Methods	Availability
TRIPPLE DES	0.4358
DES	0.4956
ABE	0.6235
BLOWFISH	0.7255
LA-SPR [3]	0.5235
AES 256-GCM [8]	0.7125
Proposed	0.9026

Additionally, the availability metric of the MAEM scheme is compared with Tripple DES, DES, ABE, Blowfish, LA-SPR and AES 256-GCM is exposed in Table 7. Here, the availability ought to be greater for the MAEM scheme. Moreover, the greatest availability is achieved using the MAEM methodology is 0.9026, though the Tripple DES, DES, ABE, Blowfish, LA-SPR and AES 256-GCM has attained minimal availability ratings.

4.7. Attack analysis on MAEM and conventional strategies for varied types of key size

Table 8
Attack analysis for key size 16.

Attacks	Triple DES	DES	ABE	Blowfish	LA-SPR [3]	AES 256-GCM [8]	Proposed (MAEM)
Eaves dropping	22.505	28.595	31.379	15.034	25.328	13.389	49.949
Man in the middle	19.336	20.970	26.468	17.055	28.583	22.505	36.199
Message tampering	20.340	19.237	24.489	10.955	18.560	26.884	33.418
Plain text attack	17.478	16.635	16.191	10.483	14.382	20.321	22.673
Brute force attack	18.560	19.202	22.151	15.177	14.382	16.011	29.253
False injection attack	0.524	0.293	0.643	0.447	0.558	0.235	0.1927
Side channel attack	0.764	0.385	0.590	0.797	0.695	0.568	0.2489

Table 9

Attack analysis for key size 32.

Attacks	Triple DES	DES	ABE	Blowfish	LA-SPR [3]	AES 256-GCM [8]	Proposed (MAEM)
Eaves dropping	22.732	28.884	15.186	26.646	25.584	13.524	50.348
Man in the middle	19.532	21.181	17.227	21.685	28.871	22.732	36.489
Message tampering	20.546	19.431	11.066	19.686	18.747	27.156	33.762
Plain text attack	17.655	16.803	10.589	17.365	14.528	20.527	22.836
Brute force attack	18.747	19.396	15.330	21.365	14.528	16.173	29.53
False injection attack	0.518	0.636	0.290	0.442	0.552	0.232	0.1903
Side channel attack	0.757	0.585	0.381	0.689	0.688	0.563	0.2433

Table 10

Attack analysis for key size 64.

Attacks	Triple DES	DES	ABE	Blowfish	LA-SPR[3]	AES 256-GCM [8]	Proposed (MAEM)
Eaves dropping	28.846	17.730	27.817	17.730	27.817	16.118	51.88
Man in the middle	24.034	19.710	31.005	19.710	31.005	25.050	38.426
Message tampering	22.095	13.734	21.185	13.734	21.185	29.341	45.803
Plain text attack	19.844	13.271	17.092	13.271	17.092	22.911	30.201
Brute force attack	23.724	17.870	17.092	17.870	17.092	18.688	33.676
False injection attack	0.491	0.623	0.484	0.433	0.541	0.228	0.1804
Side channel attack	0.542	0.473	0.374	0.475	0.574	0.351	0.199

Table 11

Attack analysis for key size 128.

Attacks	Triple DES	DES	ABE	Blowfish	LA-SPR [3]	AES 256-GCM [8]	Proposed (MAEM)
Eaves dropping	25.823	31.728	29.580	18.578	28.561	16.983	53.381
Man in the middle	22.750	24.334	24.817	20.538	31.717	25.823	39.088
Message tampering	23.724	22.654	22.898	14.623	21.997	30.070	47.419
Plain text attack	20.949	20.131	20.670	14.165	17.946	23.706	30.970
Brute force attack	21.997	22.620	24.510	18.717	17.946	19.526	35.40
False injection attack	0.449	0.417	0.379	0.429	0.336	0.225	0.1106
Side channel attack	0.536	0.468	0.370	0.370	0.568	0.248	0.1193

Attack evaluation on proposed method (MAEM) over the Tripple DES, DES, ABE, Blowfish, LA-SPR [3] and AES 256-GCM [8] for Secured fog-based vehicular crowd-sensing protocol. In this case, the attack analysis is done on two categorizes 1) The analysis of Eavesdropping, Man in the Middle, Message tampering, Plain Text Attacks, and Brute Force Attacks is based on the process of key breaking; and 2) The analysis of False Injection Attacks and Side Channel Attacks is based on correlation. Further, it is evaluated for distinctive number of key sizes (16, 32, 64 and 128) is summarized from Tables 8, 9, 10, and 11. Moreover, for a secured fog-based vehicular crowd sensing protocol, the attack rates should be higher for the key breaking procedure and lower for the correlation analysis.While assessing the Table 8, for the Message tampering attack, the MAEM offered the higher attack rate 33.418, mean while the conventional methodologies scored lower attack ratings, notably, Tripple DES = 20.340, DES = 19.237, ABE = 24.489, Blowfish = 10.955, LA-SPR = 18.560 and AES 256-GCM = 26.884, correspondingly. Additionally, for the key size 32, the MAEM gained the lesser attack value of 0.1903 in the false injection attack, whereas the Tripple DES, DES, ABE, Blowfish, LA-SPR and AES 256-GCM attained greater attack values. Likewise, for the other attacks, the MAEM accomplished superior values than the conventional strategies.

5. Discussion

This paper performs an encryption process using improved blowfish algorithm to encrypt the formerly encrypted data. The experimental findings in the simulation program demonstrate how competitive the suggested transmission mechanism is at thwarting attacks, enhancing message transmission efficiency, and providing stronger privacy assurances for users. Specifically, the Proposed (MAEM) produced a decryption time of 0.00402s for the key size 16, which is less than that of TRIPPLE DES, DES, ABE, BLOWFISH, LA-SPR, and AES 256-GCM.The time complexity depends on the number of attributes and the specific decryption policy. Efficient algorithms can handle this process effectively.This enhancement of the MAEM scheme with lower computation time is owing to the improved data aggregation with modified attribute based encryption method.

The primary drawbacks of the suggested security plan pertain to its ability to thwart internal attacks carried out by authorized users. Scalability becomes an issue when more vehicles participate in crowd-sensing. Careful planning is needed to ensure effective data aggregation and analysis over a huge fleet of vehicles. In the future research, we will concentrate on enhancing the suggested method to incorporate incentive-based participation with the least amount of cryptographic overhead possible while guaranteeing security and privacy.

6. Conclusion

This work suggested an attribute-based encryption strategy for a protected fog-based vehicle crowd sensing technique. Two stacked fog layers, referred to as fog layers A and B, were used in the creation of the suggested framework. Fog nodes and data owners (vehicles) make up fog layer A; fog nodes with RSU make up fog layer B. The TTA required regular or unique data based on the data user’s request. Additionally, the process for gathering data-driven information differs. Regular data was gathered from data owners and aggregated using the MKI-KLMS technique, which also used hierarchical fractional bidirectional least mean square and kernel least mean square for collecting redundancy data. Next, the data was encrypted using the PFKE approach, which involved inducing a fusion key to encrypt the message. Furthermore, the recently enhanced blowfish-based encryption was once more used for encryption. The original message was then retrieved by using the same key to complete the decryption procedure. Additionally, the proposed MAEM methodology’s computation time is 1.6885, whereas the corresponding values for TRIPPLE DES, DES, ABE, BLOWFISH, LA-SPR, and AES 256-GCM are 1.8451, 1.9675, and 3.1255, respectively.

References

Smahi

Xia

Sulemana

Fateh

Gao

Guizani

. A blockchainized privacy-preserving support vector machine classification on mobile crowd sensed data. Pervasive and Mobile Computing. 2020; 66: 101195.

Chen

Zhang

Zhong

. Differentially private double auction with reliability-aware in mobile crowd sensing. Ad Hoc Networks. 2021; 114: 102450.

Wang

Huang

Shen

Xiong

. A general location-authentication based secure participant recruitment scheme for vehicular crowdsensing. Computer Networks. 2020; 171: 107152.

Foschini

Martuscelli

Montanari

Solimando

. Edge-enabled mobile crowdsensing to support effective rewarding for data collection in pandemic events. Journal of Grid Computing. 2021; 19(3): 28.

Sun

Guizani

. Security and privacy preservation in fog-based crowd sensing on the internet of vehicles. Journal of Network and Computer Applications. 2019; 134: 89-99.

Ren

Wang

Zhang

. An intelligent big data collection technology based on micro mobile data centers for crowdsensing vehicular sensor network. Personal and Ubiquitous Computing. 2023; 27(3): 563-79.

Nkenyereye

Islam

Bilal

Abdullah-Al-Wadud

Alamri

Nayyar

. Secure crowd-sensing protocol for fog-based vehicular cloud. Future Generation Computer Systems. 2021; 120: 61-75.

Owoh

Singh

. Security analysis of mobile crowd sensing applications. Applied Computing and Informatics. 2022; 18(1/2): 2-1.

AP . Dynamic route scheduler in vehicular ad hoc network for smart crowd control. Personal and Ubiquitous Computing. 2020; 24: 441-9.

10.

Urra

Ilarri

. Spatial crowdsourcing with mobile agents in vehicular networks. Vehicular Communications. 2019; 17: 10-34.

11.

Pereira

Ricardo

Luís

Senna

Sargento

. Assessing the reliability of fog computing for smart mobility applications in VANETs. Future Generation Computer Systems. 2019; 94: 317-32.

12.

Yang

Xiong

Wang

Liu

. TPSense: a framework for event-reports trustworthiness evaluation in privacy-preserving vehicular crowdsensing systems. Journal of Signal Processing Systems. 2021; 93(2): 209-19.

13.

Rasheed

. Enhanced privacy preserving and truth discovery method for 5G and beyond vehicle crowd sensing systems. Vehicular Communications. 2021; 32: 100395.

14.

Nkenyereye

Liu

Song

. Towards secure and privacy preserving collision avoidance system in 5G fog based Internet of Vehicles. Future Generation Computer Systems. 2019; 95: 488-99.

15.

Ding

Zhang

Cai

Fang

. Smart cities on wheels: A newly emerging vehicular cognitive capability harvesting network for data transportation. IEEE Wireless Communications. 2017; 25(2): 160-9.

16.

Yassine

Hossain

Muhammad

Guizani

. Cloudlet-based intelligent auctioning agents for truthful autonomous electric vehicles energy crowdsourcing. IEEE Transactions on Vehicular Technology. 2020; 69(5): 5457-66.

17.

Bilal

Pack

. Secure distribution of protected content in information-centric networking. IEEE Systems Journal. 2019; 14(2): 1921-32.

18.

Alzubi

Dorgham

Alsayyed

. Cryptosystem design based on Hermitian curves for IoT security. The Journal of Supercomputing. 2020; 76(11): 8566-89.

19.

Bilal

Kang

. Network-coding approach for information-centric networking. IEEE Systems Journal. 2018; 13(2): 1376-85.

20.

Khelifi

Luo

Nour

Moungla

Ahmed

Guizani

. A blockchain-based architecture for secure vehicular Named Data Networks. Computers and Electrical Engineering. 2020; 86: 106715.

21.

Lai

Lin

Yang

Wang

. Efficient data request answering in vehicular ad-hoc networks based on fog nodes and filters. Future Generation Computer Systems. 2019; 93: 130-42.

22.

Peng

Zhong

Wang

Luo

Liu

Yang

Zhang

. Privacy-preserving truth discovery based on secure multi-party computation in vehicle-based mobile crowdsensing. IEEE Transactions on Intelligent Transportation Systems. 2024.

23.

Yin

Wei

Liu

Yang

Sun

Cheng

Mai

. VDCM: A data collection mechanism for crowd sensing in vehicular Ad Hoc networks. Big Data Mining and Analytics. 2023; 6(4): 391-403.

24.

Alzubi

Alazab

Alrabea

Awajan

Qiqieh

. Optimized machine learning-based intrusion detection system for fog and edge computing environment. Electronics. 2022; 11(19): 3007.

25.

Bekkouche

Diffellah

Yahi

. Some Improved Chaotic Maps Applied to Image Encryption. 1st International Conference on Engineering, Natural and Social Sciences. 2022.

26.

Lawnik

. Combined logistic and tent map. In Journal of Physics: Conference Series. 2018 Dec 1; 1141: 012132. IOP Publishing.

27.

Ninisha Nels

Amar Pratap Singh

. Hierarchical fractional quantized kernel least mean square filter in wireless sensor network for data aggregation. Wireless Personal Communications. 2021; 120: 1171-92.

Secured fog-based vehicular crowd-sensing protocol by Modified Attribute based Encryption Model

Abstract

Keywords

Nomenclature

1. Introduction

Table 1 Features and challenges on traditional techniques correspond to fog based vehicular crowd sensing protocol.

3. Proposed model of secured fog based vehicular crowd sensing protocol through PFKE technique

3.1. System model

3.3.1. Pfke.Setup()

3.3.2. Pfke.Keygen()

3.3.3. Proposed Pfke.Encrypt() steps

3.3.5. Proposed Pfke.dec()

3.4. Format of ID-based signature

3.4.1. IdSig.Setup()

3.4.2. IdSig.key()

3.4.3. IdSig.Sign()

3.4.4. IdSig.verf()

3.5. Description of proposed protocol

3.5.1. System initialization

3.5.1.1. Node registration

3.5.1.2. Crowd sensing request

3.5.1.3. Data collection and aggregation

4.1. Simulation procedure

Table 2 Simulation parameters.

Table 5 Computation time evaluation on MAEM and traditional methodologies.

4.6. Efficiency and availability analysis on MAEM and conventional schemes

Table 6 Efficiency evaluation on MAEM and traditional strategies.

Table 8 Attack analysis for key size 16.

6. Conclusion

References

Table 1
Features and challenges on traditional techniques correspond to fog based vehicular crowd sensing protocol.

Table 2
Simulation parameters.

Table 5
Computation time evaluation on MAEM and traditional methodologies.

Table 6
Efficiency evaluation on MAEM and traditional strategies.

Table 8
Attack analysis for key size 16.