D2D Data Privacy Protection Mechanism Based on Reliability and Homomorphic Encryption

Device-to-device (D2D) network utilizes various communication methods and regional resource sharing mechanisms, which, despite the efficient and effective communications, may lead to a variety of security threats. Therefore, without the assistance from the base station, effective managing regional resources and protecting private data on mobile devices have become a major challenge in D2D networks. In this paper, we propose a reliability-based central node election mechanism in a D2D network, where the attribute information is collected, normalized, and weight-summed to acquire the reliability of each mobile device. The central node can therefore be elected by sorting the reliability of all mobile devices. Furthermore, we propose a security protection mechanism of private data based on homomorphic encryption in a D2D network, where homomorphic encryption is employed to implement secure data aggregate of ciphertext in the elected central node. Finally, theoretical analyses and simulation experiment verify the superior effectiveness and efficiency of the proposed schemes.

SECTION I. Introduction

With the continuous development and advancement of mobile communication technologies, the number of mobile applications and volume of mobile data have grown explosively. In order to meet the requirements of high bandwidth and low latency in mobile communications, the fifth-generation (5G) mobile communication network has developed rapidly in recent years. Compared to its predecessor, the fourth generation (4G) mobile communication network, 5G networks are characterized by its significantly faster transmission rates (with theoretical peak transmission rate up to tens of Gbps), shorter latency and lower power consumption [1]. Device-to-device communication (D2D), as a key technology of 5G, has broad prospects in the areas of local communications, emergency communications, and the enhanced Internet of Things [1].

In D2D networks, data transmit and distribute directly between mobile devices, therefore significantly improving network throughput [2]. Given the fact that the device nodes are mostly mobile phones and tablet computers with very limited computing power in a D2D network, it is a core requirement to integrate network resources to improve the network performance. At present, D2D networks have three main operation modes [3], with the first one aiming to establish a connection between the mobile devices guided by a base station which also performs resource allocation [3]. In the second mode, where the mobile devices are outside the range of base station, mobile communication requires data routing through the mobile devices in the range and transmits the data to the base station. The third mode has a distinct difference from the other two modes in that the network implements communication through data forwarding among mobile devices in the area without a base station. The structure of the three modes is shown in Figure 1. In a traditional wireless network, all device nodes are connected to the base station, which therefore usually function as the core of the network. Since there is no base station serving as the core in the third mode of a D2D network, a central node needs to be elected taking charge of performing network resource management for improved network performance. How to properly elect and assign the central node then becomes the main challenge in this mode. In addition, in this mode, all nodes are directly linked to other nodes with unknown reliability which inevitably leads to mobile data traveling in the D2D network possibly through unknown or un-trusted nodes and subsequently putting data privacy protection on risk. Currently, there are four main data protection methods for D2D network communications [4], including access control, data obfuscation, data anonymization, and data encryption.

FIGURE 1.

Three operation modes of D2D network.

Data obfuscation is to protect data privacy based on summarized information or provides false information to reduce the accuracy of data. For example, when publishing location information of mobile device nodes, only the state or province where the mobile device node is located in would be advertised. Data anonymization is implemented in a way that the published private data contains a certain number of pseudonyms, so that the mobile device nodes receiving such data are not capable of identifying the private data owner. These two protection methods often have an adverse effect on data, so data encryption is most feasible for protecting private data in a D2D network. However, the traditional encryption algorithms have the challenge to data secure aggregation operations that cannot be performed without decrypting the ciphertext. Therefore, homomorphic encryption is introduced to perform data secure aggregation operation on the elected central node without decryption.

Combined with the central node election mechanism and homomorphic encryption algorithm, this paper proposes a D2D data privacy protection mechanism based on reliability and homomorphic encryption. The main contributions of this paper are as follows:

1. We propose a reliability-based central node election mechanism in D2D communication networks, where the information about a mobile device node is integrated to calculate its reliability, and a device node is elected as the central node of the D2D network based on the sorting of the reliability among all nodes.

2. The Paillier homomorphic encryption system is used to implement a private data protection model, where the integration and allocation of computing resource and data in a D2D network can also be achieved.

3. Simulating the real network environment of D2D communication verifies the reliability and security of the proposed model, establishing a D2D data privacy protection mechanism based on reliability and homomorphic encryption.

SECTION II. Related Work

There are four main methods of privacy protection: access control, data obfuscation, data anonymization, and data encryption, which have been widely used in wireless communications. The following section will discuss some existing privacy protection methods that can be used in D2D communication.

At present, there are mainly three models for access control, which are discretionary access control (DAC), mandatory access control (MAC), and role-based access control (RBAC). The DAC means that the mobile device node has full control over the object it belongs to and the program it runs. For example, a mobile device node owns its location information. The mobile device node only allows neighbor nodes to obtain this information, and the rest of the nodes have no right to reach it. The MAC refers to the administrator to define the relevant device to stipulate which device node can access the object. Because of its policy-based, any unauthorized operation cannot be performed, and this access control model can set the different security levels. The RBAC means that the administrator can define a series of roles and assign suitable mobile device nodes to them. Therefore, the mobile device nodes with corresponding roles can access the classified objects. Behrooz and Devlic [5] proposed a model for controlling context access through DAC systems, which uses context awareness and mobile device relationships to solve complex situations in communications, enabling mobile users to control who can access their context by specifying their context-aware privacy rules. Another SensorSafe access control model proposed by Chakraborty et al. [6] is designed to protect personal sensor data. The level of data access is determined by the administrator based on the credibility between users.

Data obfuscation is mainly to protect the private data by summarizing information and providing false information. Wishart obfuscates the data based on the content of the information and the surrounding environment, the users can set the obfuscation level applied to the data based on the current situation [7]. They can also control who is the information published to,and set the policy on publishing the information and the level of detail of the published information. Another mechanism is proposed by Franz to divide the devices in the network into different groups. The members of the group negotiate the private data policy to determine which private data can be published and the degree of obfuscation of the private data [8]. This mechanism is mainly applied to social networks.

In the aspect of data anonymization, since the D2D communication needs to be authenticated when it is established, it is necessary to consider the specialty of using data anonymization in the D2D network to protect data privacy. In the D2D network, it is necessary to combine anonymous technology and reliability mechanism to establish trust relationships and trust connections between device nodes anonymously [9]. In this way, even if the content is shared with strange device nodes, their identities will not be revealed, and more sensitive information can be shared. Christin et al. [10] proposed a method to protect the privacy of device sensor information in the network by giving the pseudonyms credibility. Any operation of device node using a pseudonym will affect its credibility.If the device changes a pseudonym, the credibility of the previous pseudonym is passed to the next pseudonym. In addition, Boutsis and Kalogeraki [11] share user’s location information on mobile devices. Each user only knows part of the location information, so they cannot identify the information source.

In the aspect of data encryption, the focus of related work is on lightweight encryption mechanisms because of the limitations of mobile device’s computing power and their energy consumption. Currently, the standard method widely used in wireless communication is the Public Key Infrastructure (PKI). Also, due to D2D communication, it should use multi-party and distributed cryptographic protocols. In 2007, Kate proposed an Identity-based Cryptography (IBC) system, where each device node can create a public key using locally available information [12]. They do not need to go to the central node to verify the public key, so the certificates can be directly exchanged with other mobile device nodes. In addition, Searchable Encryption (SE) can also be applied to D2D networks. It allows users to search for keywords in ciphertext.

SECTION III. Reliability-Based Central Node Election Mechanism

In the data security protection mechanism based on homomorphic encryption, a central node is required to collect the encrypted ciphertext of each mobile device, perform secure data aggregation operation and send it to the device that requests the data in the D2D network. At the same time, the network also needs the central node to manage the resources in the entire network. In a traditional communication network, this core device is usually a base station. However, in the D2D network, there is often no base station as a core device. Therefore, it is necessary to elect a device as a central node in the D2D network. Due to the complexity of the D2D network environment, the assessment of each mobile device in the network should consider its internal and external factors. In order to meet the above requirements, a reliability-based central node election mechanism is described in this section.

A. Related Terms

1. Term 1: Min-max Normalization [13]. The original data is transformed linearly so that the result is between 0 and 1. The conversion function is $X_{norm}=(X-X_{min})/(X_{max}-X_{min})$ , where $X_{norm}$ is a normalized data and $X_{min}$ and $X_{max}$ are a minimum and a maximum value of the original data set, respectively.

2. Term 2: Device Node Density [14]. One of the election factors of the reliability-based central node election mechanism refers to the density of device nodes around a device node in the D2D network, which is used to describe the degree of mobile device nodes located at the center of the D2D network.

3. Term 3: Service Success Rate. One of the election factors of the reliability-based central node election mechanism is the inspection of the device node providing services to the network and representing the ratio of the number of success service times and the total number of services provided by the device node historically.

4. Term 4: Service Hours. One of the election factors of the reliability-based central node election mechanism is the inspection of the device node providing services to the network and showing the time when the device node historically provides services.

B. Main Idea

To elect the central node for secure data aggregation and management of network resources in a D2D network, a reliability mechanism is proposed to evaluate and calculate the reliability of device nodes from different aspects of mobile device nodes. In this mechanism, we firstly collect information of each mobile device node in the D2D network, and the information is then normalized and weighted summation to get the reliability of each mobile device. Finally, according to this reliability, the central node of data aggregation and management network resources is elected in the current D2D network. The process of the election mechanism is shown in Figure 2. The detailed description is as follows.

1. The information of each mobile device node in the D2D network is collected in a quantized form, including: service hours, calculation performance, service success rate, device node density and transmission performance. The reliability of each mobile device node is obtained by using these quantized data.

2. According to the D2D network environment, the weight of each election factor is obtained through repeating experiments, and each factor is weighted to obtain the reliability of each mobile device.

3. The reliability of all devices in the D2D network is sorted, the most trusted mobile device node is elected to perform the central node of secure data aggregation.

FIGURE 2.

The process of the election mechanism.

C. Quantification of Device Node Information

To calculate the reliability of each mobile device node, all the factors need to be quantified to get the election factor. The above five factors are quantified as follows.

1. Quantification of service hours. The history service hour of each device in the D2D network is $m_{i}$ . The greater the value of $m_{i}$ , the longer the service hours of the mobile device.

2. Quantification of device node density. Broadcasting hello information to the entire network, the ratio $De$ of the number of neighboring mobile device nodes to the total number of mobile device nodes in the network is calculated. The specific process is as follows.

   1:

   sendHelloMessage(HelloBean)

   //Broadcast Hello messages to the entire network

   2:

   list<HelloBean>=receiveHelloMessage()

   //Receive Hello messages from the entire network

   3:

   $N \leftarrow list.length+1$

   //$N$ is the number of device nodes in the entire network

   4:

   $n_{i}$

   //$n_{i}$ is used to count the number of neighbor nodes

   5:

   for $i=1 \rightarrow list.length$ do

   6:

   if $list [i].jumb==1$ then

   7:

   $n_{i}+=1$

   //Device node with hop count 1

   8:

   end if

   9:

   end for

   10:

   $De_{i}=n_{i}/N$

   //Calculate the node density of the device,

   $De_{i} \in [0,1)$

3. Quantification of calculation performance. The number of milliseconds required for each mobile device to perform 1Kb data encryption in the D2D network is $x_{i}$ . The lower the value of $x_{i}$ , the better the computing performance of the mobile device node.

4. Quantification of transmission performance. Each mobile device in the D2D network requests 1Mb of data to each mobile device on the entire network. The number of mobile device nodes in the entire network is $N$ . The number of milliseconds between each device’s request and the time it takes to receive a total of $N$ Mb’s data is $y_{i}$ . The lower the value of $y_{i}$ , the better the transmission performance of the mobile device node.

5. Quantification of service success rate. The ratio of historical service times to successful service times of device nodes is $Su_{i}$ . The process is as follows.

   1:

   $Z \leftarrow getServiceTimes()$

   //$Z$ is the total number of services in the device node history

   2:

   $Z_{i} \leftarrow getSuccessServiceTimes()$

   //$Z_{i}$ is the number of service successes in the device node history

   3:

   $Su_{i}=Z_{i}/Z$

   //$Su_{i}$ is the service success rate of the device,

   $Su_{i} \in [0,1]$

Quantifying the above five participating factors, the abstract description of the network environment and device node performance can be visually displayed in a digital way. At the same time, it can quantitatively express the differences between node and node factors and provide basis for the calculation of reliability and the election of central nodes.

D. Election Factor Normalization

To make the value of the finally calculated reliability at [0,1], the quantified election factors need to be normalized. The node density and the service success rate of mobile device in the five election factors are already at [0,1], so only the service hours, calculation performance, and transmission performance need to be normalized. Using the min-max normalization algorithm to normalize service hours, as shown in Algorithm 1. However, the calculation performance and the transmission performance are inversely proportional to their quantified values. Therefore, the original min-max normalization algorithm needs to be simply modified so that the normalized value can react the mobile device node’s calculation and transmission performance, as shown in Algorithm 2 and Algorithm 3.

Algorithm 1 Normalize the Service Time

Input:

the serviceTime dataset $M$ , the number of serviceTime $m_{i}$

Output:

weight of serviceTime $Se_{i}, Se_{i}\in [{0,1}]$

1:

for $i = 0 \to M.length$ do

2:

if $m_{max}< m[i]$ then

3:

$m_{max}=m[i]$

4:

end if

5:

if $m_{min}>m[i]$ then

6:

$m_{min}=m[i]$

7:

end if

8:

end for

9:

return $Se_{i}=(m_{i}-m_{min})/(m_{max}-m_{min})$

Algorithm 2 Normalize the Calculation Performance

Input:

the calculationPerformance dataset $X$ , the number of calculationPerformance $x_{i}$

Output:

weight of calculationPerformance $Ca_{i}$ , $Ca_{i}\in [{0,1}]$

1:

for $i = 0 \to X.length$ do

2:

if $x_{max}< x[i]$ then

3:

$x_{max}=x[i]$

4:

end if

5:

if $x_{min}>x[i]$ then

6:

$x_{min}=x[i]$

7:

end if

8:

end for

9:

return $Ca_{i}=(x_{max}-x_{i})/(x_{max}-x_{min})$

Algorithm 3 Normalize the Transmission Performance

Input:

the transmissionPerformance dataset $Y$ , the number of transmissionPerformance $y_{i}$

Output:

weight of transmissionPerformance $Tr_{i}, Tr_{i}\in [{0,1}]$

1:

for $i = 0 \to Y.length$ do

2:

if $y_{max}< y[i]$ then

3:

$y_{max}=y[i]$

4:

end if

5:

if $y_{min}>y[i]$ then

6:

$y_{min}=y[i]$

7:

end if

8:

end for

9:

return $Tr_{i}=(y_{max}-y_{i})/(y_{max}-y_{min})$

The Algorithm 1 is to obtain the value of the node service hour election factor through the service hours of the node. Firstly, we obtain the dataset of historical service hours of all nodes in the entire D2D network, the maximum and minimum values in the dataset, that is, the longest service hours and the shortest service hours, and calculate the difference between them. Then, we calculate the difference between the service hours and the shortest service hours of each node. Finally, we find the ratio of two differences as the service hour election factor. As can be seen from the expression, the longer the service hours of the node, the greater the value of the service hour factor, and the better the node performing on the item.

The Algorithm 2 is to obtain the value of the node calculation performance election factor through the encryption time of the node. Firstly, we obtain the dataset of encryption time of all nodes in the entire network, the maximum and minimum values in the dataset, that is, the longest encryption time and the shortest encryption time, and calculate the difference between them. Then, we calculate the difference between the longest encryption time and the shortest encryption time of each node. Finally, we find the ratio of two differences as the calculation performance election factor. As can be seen from the expression, the shorter the encryption time of the node, the greater the value of the calculation performance factor, and the better the node performs on the item.

The algorithm 3 is to obtain the value of the node transmission performance election factor through the encryption time of the node. Firstly, we obtain the dataset of transmission time of all nodes in the entire network, the maximum and minimum values in the dataset, that is, the longest transmission time and the shortest transmission time, and calculate the difference between them. Then, we calculate the difference between the longest transmission time and the shortest transmission time of each node. Finally, we find the ratio of two differences as the transmission performance election factor. As can be seen from the expression, the shorter the transmission time of the node, the greater the value of the transmission performance factor, and the better the node performing on the item.

E. Election of a Central Node Through Reliability

After obtained the normalized election factors, the reliability of each mobile device node needs to be obtained through the weighted summation of the election factors. The weights of the above five factors are $\alpha$ , $\beta$ , $\gamma$ , $\delta$ , $\varepsilon$ and the weights satisfy the expression $\alpha +\beta +\gamma +\delta +\varepsilon =1$ . Due to the complexity of the D2D network, the weights of the factors are based on actual application scenarios and experiments. The reliability of the weighted summation obtained by $Re_{i}$ , as shown in Algorithm 4.

Algorithm 4 Calculate the Reliability

Input:

the node density $De_{i}$ , success rate of service $Su_{i}$ , weight of serviceTime $Se_{i}$ , weight of calculatedPerformance $Ca_{i}$ , weight of transmissionPerformance $Tr_{i}$ , weight of five factors $\alpha, \beta,\gamma,\delta,\varepsilon$

Output:

reliability $Re_{i}, Re_{i}\in [{0,1}]$

1:

return $Re_{i}=\alpha *De_{i}+\beta *Se_{i}+\gamma *Ca_{i}+\delta *Tr_{i}+\varepsilon *Su_{i}$

The higher the reliability value, the better the mobile device node is as the center node. After obtaining the reliability of each mobile device node in the D2D network, a fast sort algorithm is used to sort the reliability of each node in the D2D network. According to the result of quick sorting, the mobile device node with the highest reliability is selected as the central node, and the mobile device node with the second highest reliability is used as the backup central node. When the central node cannot provide services normally, the central node is replaced by the backup central node.

SECTION IV. Private Data Protection Mechanism Based on Paillier Homomorphic Encryption

In the D2D network, mobile device nodes often need to request network resources in the entire network, and these resources often contain some private information. If the relevant data is directly sent to the requesting device, it is easy to cause the privacy disclosure and the insecure D2D network. Therefore, the private data in the D2D network needs to be encrypted and then sent to the device that requested data. To achieve the above requirements, the Paillier homomorphic encryption model [15] is introduced into the D2D network, and the proposed private data protection mechanism based on homomorphic encryption is described as follows.

A. Related Terms

1. Term 1: Homomorphic Encryption. The multiple encrypted data is calculated directly to obtain an integrated ciphertext, which is the same as that obtained by processing the unencrypted original data in the same way.

2. Term 2: Additive homomorphism. There is an algorithm $\bigoplus$ that makes $E(x+y)=E(x)\bigoplus E(y)$ or $x+y=D(E(x)\bigoplus E(y))$ true without leaking $x$ and $y$ .

3. Term 3: Paillier encryption. Paillier encryption system, a probabilistic public key encryption system invented by Paillier [15] in 1999. The encryption algorithm is a homomorphic encryption that satisfies addition and multiplication homomorphism.

B. Main Idea

When a mobile device node in the D2D network requests resources from the entire network, a data privacy protection mechanism is needed to meet this requirement. Figure 3 employs homomorphic encryption algorithm [15] to encrypt the data of each mobile device node and sends it to the requesting device after secure data aggregation. Firstly, each mobile device node encrypts the data after receiving the request. Then the ciphertext is sent to the elected central node, which aggregates the received ciphertext and sends it to the requesting device. Finally, the requesting device node decrypts the ciphertext to obtain the resource.

FIGURE 3.

Architecture of private data protection mechanism based on homomorphic encryption.

C. Key Generation

In the key generation phase, each mobile device node needs to generate its own public key and private key and sends their public key to the elected central node. Each node in D2D network performs Paillier public and private key generation algorithm. $k$ mobile devices generate $k$ sets of public keys $(n,g)$ and private keys $(\lambda, \mu)$ . hey send the public keys to the central node, and the private keys are saved locally. Paillier public and private key generation algorithm, the process is as follows [15].

1:

Select prime numbers $p$ and $q$ randomly to satisfy $gcd(pq,(p-1)(q-1))=1$

2:

Let $n=pq$ , $\lambda =lcm(p-1,q-1)$

3:

Defining the function $L$ is $L(\mu)=(\mu -1)/n$

4:

Select random number $g(g\in Z_{n^{2}}^{*})$ and satisfy $gcd(L(g^{\lambda }mod n^{2}),n)=1$ , $\mu =(L(g^{\lambda }mod n^{2}))^{-1}mod n$

5:

The public key is $(n,g)$ and the private key is $(\lambda,\mu)$ .

Private data protection mechanism based on Paillier homomorphic encryption allows new mobile devices to join the D2D network. When a new mobile device requests to join the network, it firstly sends a request to the central node. After passed the authenticates of the central node device, the new mobile device node will join the D2D network. When the new node is added, the Paillier public and private key generation algorithm needs to be executed. The public key of the new device node generated by the algorithm is sent to the central node, and the private key is locally saved.

D. Encryption Process

When a mobile device node in the D2D network needs to request resources from the entire network, the requesting device firstly sends a request to the central node, which broadcasts a request to the entire network and publishes the public key of the requesting device. The other nodes in the D2D network need to encrypt the provided private data after receiving the request to prevent privacy leakage. Each node encrypts according to the obtained request and the public key, as shown in Algorithm 5 [15]. Finally, the encrypted ciphertext is sent to the central node, and the requesting device node needs to be shielded in the routing process to prevent the requesting device node from decrypting the ciphertext.

Algorithm 5 Paillier Encryption

Input:

the public key of request node $(n,g)$ , the clear-text $m(m\in Z_{n})$

Output:

the cipher-text $C_{i}$

1:

$r \gets randome()r\in Z_{n}^{*}$

2:

return $C_{i}=g^{m}*r^{n}mod n^{2}$

After collecting the ciphertext of each node in the D2D network, the central node prevents the data of each node from being leaked by the requesting device node, the secure aggregated operation needs to be performed on the received data, as shown in Algorithm 6. Finally, the processed ciphertext is sent to the requesting node, so the requesting node device can only obtain the requested data, not the data from a single node device, which protects the private data of each node device [16].

Algorithm 6 Paillier Homomorphic Process

Input:

the cipher-text dataset $C$

Output:

the cipher-text after homomorphic process $C^{*}$

1:

for $i = 0 \to C.length$ do

2:

$C^{*}*=C[i]$

3:

end for

4:

return $C^{*}$

After receiving the ciphertext processed by the homomorphic encryption, the requesting node device uses the private key saved locally to perform a decryption algorithm on the ciphertext to obtain the returned result, as shown in Algorithm 7.

Algorithm 7 Paillier Decryption

Input:

the cipher-text after homomorphic process $C^{*}$ , the public key of request node $(n,g)$ , the private key of request node $(\lambda,\mu)$

Output:

the return result $m^{*}$

1:

return $m^{*}=L(C^{\lambda }mod n^{2})*\mu mod n$

E. Homomorphic Verification

The Paillier encryption algorithm [15] is an encryption algorithm with additive homomorphism, which is processed as follows. Let message $m_{1}$ and $m_{2}$ exist, and encrypt them to get:

$$\begin{align*} E(m_{1})=&~g^{m_{1}}x_{1}^{n} mod n^{2} \tag{1}\\ E(m_{1})=&~g^{m_{2}}x_{2}^{n} mod n^{2} \tag{2}\end{align*}$$ (1)*(2) to get: $$\begin{equation*} E(m_{1})*E(m_{2})=g^{m_{1}}x_{1}^{n}*g^{m_{2}}x_{2}^{n} mod n^{2}=E(m_{1}+m_{2}) \tag{3}\end{equation*}$$

SECTION V. Theoretical Analysis and Performance Evaluation

The theoretical analysis section firstly verifies the D2D data privacy protection mechanism based on reliability and homomorphic encryption proposed in this paper satisfying the preset target. Based on this, the merits and rationality of the mechanism are discussed. The analysis is based on the comprehensiveness of the election mechanism, the complexity of the algorithm, and the efficiency of homomorphic encryption. The part of performance evaluation is based on the analysis of the experimental results to verify the reliability of the mechanism and the reasonableness of the reliability calculation method of the mobile device node.

A. Theoretical Analysis

Based on the D2D communication technology and network structure, this paper proposes a reliability-based central node election mechanism and employs Paillier homomorphic encryption system to construct a private data protection mechanism in the D2D networks. It ensures the security of private data in D2D networks while reasonably performs D2D network resource allocation. In the D2D network, mobile device node’s reliability calculation process, the historical service hours, historical service success rate, device node density, calculation performance and transmission performance of device nodes are taken into consideration, and the reliability of the device node is fully and reliably evaluated. The program firstly calculates the credibility of the internal factors of the mobile device node, including the node’s calculation performance, historical service hours, and historical service success rate, and evaluates whether the node device can allocate and manage D2D network resources and perform data aggregation. Secondly, the proposed mechanism evaluates the network environment of the mobile device node, including the density of nodes around the mobile device node and the surrounding network transmission performance. If the performance of the mobile device node is poor, the entire D2D network resource allocation may not be timely and the security of private data may be reduced. To ensure the credibility of the reliability, when calculating the reliability, it is not simply to sum up each election factor, but to use a weighted summation. In practice, the proportion of each factor’s weight may be adjusted according to the application of the network to select the most reasonable mobile device node as a central node in the D2D network.

In the same type of network node election mechanism, the paper [17] proposed a trusted center node selection algorithm, which firstly uses the history service success rate of the election node as a direct trust value. Then, after the election node processing the service provided by the specified node, the recommendation trust value of the election node to the specified node is obtained, and the recommendation trust values of all nodes in the network are added up and summed to obtain the recommendation trust value of the election node. Finally, according to the weights of the two parts, the comprehensive trust value is calculated. The mechanism considers only the node density of the mobile device node and the node service, and lacks the consideration of the calculation and transmission performance of the node.

The paper [18] proposed an algorithm that user requirements adapt to the node selection. Firstly, the user puts forward requirements for nodes, including factors such as calculation performance, bandwidth, and number of neighbor nodes. These parameters are formed into a matrix, which is processed to obtain a criteria level. Secondly, user-defined weights of various parts constitute a matrix, which is processed to be a decision level. Finally, according to the ratio of the importance of the decision and the criteria, a decision matrix is obtained, and the values in the matrix are sorted to obtain the decision goal which is the election node. This mechanism only considers the performance of the node itself and the surrounding network, and the user must give the quantized parameters, and lacks the calculation method of related parameters. The following is a summary of the above three mechanisms, as shown in Table.1.

TABLE 1 Node Selection Algorithms Comparison

In this paper, the algorithm complexity of the reliability-based central node election mechanism is $O(n^{3}\log _{2}n)$ . Because each node in the network only needs to fetch data from the other $n-1$ nodes, the number of execution times of the entire network is $n*(n-1)$ . Then, by multiplying the complexity of quick sorting ($O(n*\log _{2}n)$ ), we can obtain the complexity of our algorithm $O(n^{3}\log _{2}n)$ . As a comparison, the complexity of the other two algorithms is $O(n^{4}\log _{2}n)$ with the number of execution times of the entire network in [17] and [18] being $n^{3}$ . Therefore, the mechanism proposed in this paper has a lower algorithm complexity. The comparison of the algorithm complexity is shown in Table.2.

TABLE 2 Algorithm Complexity Comparison

B. Election Performance Evaluation

The feasibility of the reliability-based central node election mechanism is verified through repeated simulation experiments. Under the corresponding D2D communication network, the simulation experiment is used to verify that the nodes in the network can obtain the reliability under the actual situation, and elects the central node according to the reliability. The experimental environment is configured with an Intel Core i7–4710HQ CPU, 4G RAM, and Ubuntu 16.04 Linux operating system (64-bit) Network simulation software adopts NS2, and simulation data processing software adopts gawk. The reason for employing NS2 over NS3 is that NS2 has superior stability compared to Ns3.

Firstly, we employ NS2 to establish a D2D network topology as shown in Figure 4. The bandwidth between the node 2 and the node 3 is 100 Mbps and the delay is 10 ms, and the bandwidth between the other nodes is 500 Mbps and the delay is 2 ms.

FIGURE 4.

NS2 network topology.

Secondly, we set relevant parameters for each node as follows. Node 0, Node 1 and Node 2 are original nodes and the service hours are 100 hours, 110 hours and 120 hours respectively. The service success rates are 85%, 90% and 95% respectively. Node 3, Node 4 and Node 5 are newly added nodes and the service hours are 10 hours and the service success rate is 100%. Node 2 and Node 3 are portable computer, and the key length is 1024 bits. According to the encryption performance analysis, the encryption time is 12.94ms. Node 0, Node 1, Node 4, and Node 5 are Android devices. According to experiments, the encryption time is about 20.85ms. The parameters of the node are shown in Table.3.

TABLE 3 Node Parameters

Finally, the network simulation software is used to simulate the data transmission process to obtain the network performance parameters around each node. The network parameters of the nodes are shown in Table.4.

TABLE 4 Network Parameters

After all the election parameters of the node are obtained, five election factors are obtained by using the optimized normalization algorithm in the election mechanism. The obtained values of the reliability factors are shown in Table.5.

TABLE 5 The Reliability Values of Election Factors

Getting the weight matrix of each election factor, we here set the same for each election factor, $\{ \alpha,\beta,\gamma,\delta,\varepsilon \}=\{ 0.2,0.2,0.2,0.2,0.2 \}$ . Finally, according to the weights of each election factor, the weighted summation of the reliability to each node is obtained. The reliability of each node is shown in Table.6.

TABLE 6 Node Reliability

As shown in the table above, Node 2 and Node 3 are located in the network center. Compared with other nodes, these two nodes directly connect with more other device nodes, and the surrounding network bandwidth is consequently higher. Therefore, the central node should be selected from Node 2 and Node 3. Compared to Node 3, Node 2 has superior calculation performance, transmission performance and longer service hours. For these reasons, Node 2 is more suitable for serving as a network center node for managing network resources. The experimental results are in line with expectations.

C. Encryption Performance Evaluation

The feasibility of Paillier’s homomorphic encryption algorithm [15] is verified through repeated experiments. In the traditional network, the central node is often a base station with good performance. In the third mode of D2D network, there is no base station as the core of the network. Therefore, the elected central node may be a portable computer or even a smart phone. The experimental environment was configured with an Intel Core i7-4710HQ CPU, 8G memory, Windows 10 Ultimate operating system (64-bit), and encryption algorithm implementation software adopts MyEclipse.

Firstly, we verify the relationship between the data encryption time and the key length under the experimental environment. Generally, the longer the key length is, the lower the probability of private data leakage is, and the higher the security of data, while the encryption time of the unit data will also increase. Therefore, the key length needs to be adjusted according to the size and importance of the data. The experiment employs the Paillier encryption algorithm [15] to encrypt 8-byte-size bigInt data types for different lengths of keys and to calculate the time. Since the key generation process includes the generation of a random prime number, the time of the encryption process is related to the value of the prime number. Therefore, ten keys of the same data size are generated for encryption, and the average value of the encryption time is taken as the encryption time of the length key. The relationship between key length and encryption time is shown in Figure 5. From this figure, the encryption time of data increases exponentially with the increase of the key length, which is in line with expectations.

FIGURE 5.

Key length and encryption time.

Secondly, we verify the relationship between the data aggregation time and the number of aggregated data and key length under the experimental environment. Generally, the greater the number of data, the longer the key length, the longer the data aggregation time, and also the longer the network responds to the request. The data aggregation time is also related to the homomorphic algorithm. The experiment uses the Paillier homomorphism algorithm to perform data aggregation on the ciphertext of bigInt data when the keys are the same. The data aggregation time of different data amounts is counted. Then, we use different key lengths to aggregate data for 10 bigInt data types and calculate the time. Since the key generation process includes the generation of a random prime number, the time of the encryption process is related to the value of the prime number, so 10 groups of keys with the same data size are generated for encryption, and the time average time value of the data aggregation is taken as the aggregation time of the length key. The relationship between the number of aggregated data and aggregation time is shown in Figure 6, and the relationship between key length and aggregation time is shown in Figure 7. From these figures, we can see that the aggregation time of data grows linearly with the increase of aggregated data quantity, and exponentially increases with the increase of key length, which is in line with expectations.

FIGURE 6.

Data amounts and aggregation time.

FIGURE 7.

Key length and aggregation time.

Finally, we verify the relationship between data encryption time and key length under the experimental environment. Generally, the longer the key length, the longer the decryption time of the unit data. The experiment uses the Paillier decryption algorithm to decrypt the ciphertext aggregated by 10 times of 8-byte bigInt data types using different key lengths and to calculate the time. Since the key generation process includes the generation of a random prime number, the time of the decryption process is related to the value of the prime number, so ten keys of the same data size are generated for encryption, and the average value of the decryption time is taken as the decryption time of the length key. The relationship between key length and decryption time is shown in Figure 8. From this figure, the decryption time increases exponentially with the increase in key length, which is in line with expectations.

FIGURE 8.

Key length and decryption time.

SECTION VI. Conclusion

With the rapid development of the fifth-generation (5G) communication, D2D communication technology will deeply affect everyone’s daily work and life. For this reason, the data security and privacy of D2D communications becomes a critical issue and major challenge in D2D communications. In this paper, we introduce the homomorphic encrypted private data protection mode in D2D communication and solved the issue of lacking reliable devices to perform secure data aggregation operation without a base station. The proposed mechanism improves the security and anti-attack ability of D2D networks and optimizes the resources allocation in D2D network.

Our future work will be focused on the following aspects: 1) delving more deeply into the connection and influence factors between election factors and the impact of the dynamics of wireless devices on D2D networks; 2) the D2D network is dynamic in reality: a) the distance and bandwidth among nodes, the performance of nodes and other factors are continuously changing; b) new nodes can join the network while existing nodes can leave at arbitrary moments. How to elect central nodes in the real network environment needs to be further studied.