Study of Network Security Based on Key Management System for In-Vehicle Ethernet

Abstract

With the rapid development of vehicle electronic communication technology, in-vehicle bus network system communicates with external electronic devices such as mobile phones and OBD II, causing in-vehicle bus networks to face severe network security threats. This study aims to explore the security scheme of in-vehicle bus networks based on a key management system to ensure the confidentiality, integrity, authenticity, and availability of vehicle communication, and innovatively propose a key management system. This key management system uses data encryption and signature algorithms based on the elliptic curve cryptographic domain, which is mainly composed of key generation and key distribution modules. By designing a key life cycle management strategy for In-Vehicle Ethernet and using the digital envelope technique, data encryption and digital signatures are combined to ensure the secure generation and distribution of keys. Experimental simulation results show that the session key negotiation speed of the proposed key management system for In-Vehicle Ethernet in this study is 1.533 ms, which improves the speed by 80.5% compared with the traditional key management system. The key management system proposed in this study improves the real-time information processing efficiency in In-Vehicle Ethernet and lays a solid foundation for the stable development of intelligent connected vehicles.

1. Introduction

With the rapid development of vehicle electronics, intelligence, and interconnected intelligent transportation, In-Vehicle Ethernet has become an indispensable communication system for vehicles. In-Vehicle Ethernet can replace traditional in-vehicle bus network systems such as LIN, CAN, FlexRay, etc., and become the backbone network of the next generation in-vehicle bus network system. In-Vehicle Ethernet establishes a high-speed communication network inside the vehicle to provide safe and stable data transmission. In-Vehicle Ethernet has the advantages of a high bandwidth up to 1 G/bps, a small number of ECUs, and reduced wiring harness costs [1,2]. However, with the popularization of smart vehicles, the network security issues of smart vehicles are becoming more and more prominent. An intelligent connected vehicle is a vehicle that integrates advanced communication technology, perception technology, control algorithms, and artificial intelligence. Such vehicles are able to collect and analyze a large amount of driving data, including vehicle status, driving behavior, road condition information, etc., and by connecting to the cloud, receive updates and maintenance information from manufacturers or service providers, and can even receive remote control commands in some cases, thus improving driving safety and enhancing traffic efficiency. During the vehicle information transmission process of In-Vehicle Ethernet, there will be data security risks such as data loss, data leakage, unauthorized access, and hacker intrusion. In order to further improve the active safety performance of vehicles, the data security performance of In-Vehicle Ethernet must be ensured. The core of vehicle communication data security technology lies in the high-efficiency encryption of data, so key management technology plays a key role in In-Vehicle Ethernet network security.

In-vehicle network security technology involves the integration of cross-field technologies such as data encryption and authentication, data compression, fault detection, and fault tolerant control. Since the communication protocols of in-vehicle bus networks such as In-Vehicle Ethernet and CAN bus do not have active protection functions, such as network intrusion detection and data security, it is difficult to ensure the security of in-vehicle networks in a multi-bus network reconstruction environment. Lee Y et al. [3] proposed to create a system for detecting attacks by using the cyber-kill chain method. The detection system shares network analysis results and develops security operation strategies by establishing a vehicle CVE website. Nichitelea T C et al. [4] designed an analog embedded controller based on two In-Vehicle Ethernet systems, and compared the control effect with the CAN bus. Woo S et al. [5] proposed a network attack model and a network protection security protocol for network attacks in the CAN bus network environment, and conducted an actual attack experiment on in-vehicle bus networks through smartphones. Although vehicles improve driving pleasure for drivers by connecting to network systems such as smartphones, Bluetooth, and the Internet, they also bring about in-vehicle network security issues such as hacker attacks and security vulnerabilities that cannot be ignored, seriously affecting the safe driving, personal privacy, and even endanger public safety. Therefore, an in-vehicle bus network system is no longer an independent and secure network system.

This paper is divided into five sections. In Section 2, the In-Vehicle Ethernet and key management techniques are analyzed. In Section 3, the design of a key management system for In-Vehicle Ethernet is proposed. In Section 4, the overhead performance of key generation and key distribution modules is evaluated. Finally, research conclusions are presented in Section 5.

2. In-Vehicle Ethernet and Key Management Technology

2.1. In-Vehicle Ethernet

In-Vehicle Ethernet is one of the core technologies of modern in-vehicle bus network systems, providing strong support for the intelligence, interconnection, and safety of vehicles [6]. The popularity of In-Vehicle Ethernet allows the vehicle entertainment system to efficiently interact with the external environment. For a long time, global vehicle manufacturers have used bus networks such as LIN, CAN, and FlexRay to independently build in-vehicle bus networks to ensure the security of in-vehicle bus networks [7]. The rapid development of technologies such as Internet of Vehicle (IoV), Advanced Driving Assistance System (ADAS), and V2X (Vehicle to Everything) requires more bandwidth to transmit information data [8]. Although in-vehicle bus networks play an important role in information transmission, they face challenges in handling high-speed data transmission. In addition, security incidents such as hacking not only threaten the driving safety of vehicles, but also may violate personal privacy and even pose a threat to public safety [9]. Therefore, the network security of In-Vehicle Ethernet has become a hot topic of current research.

As the main medium of vehicle communication, in-vehicle bus networks have a crucial security issue. Most attacks against intelligent connected vehicles rely on security vulnerabilities of in-vehicle bus networks [10]. In the field of data security, the data transmitted through an in-vehicle bus network system may contain sensitive information such as vehicle status and location data, so it is necessary to prevent unauthorized personnel from obtaining the information and ensure the confidentiality of the data. At the same time, methods such as restricting access rights and strengthening password protection are used to improve the management performance of in-vehicle bus networks, which can effectively prevent illegal intrusions and information leaks and ensure the security of in-vehicle bus networks [11]. In the field of counterattacks, In-Vehicle Ethernet faces DoS (denial of service) attacks, intrusion attacks, and other network attacks capability of the network needs to be strengthened. In order to further ensure the network security performance of In-Vehicle Ethernet, the key management performance based on key generation and key distribution technology must be improved.

2.2. Key Management Mechanism

Key management technology is a key core part of cryptography [12]. A key management system (KMS) includes key generation, storage, distribution, application, update, and destruction. A key management system reduces security risks caused by key leaks or improper management. A key management system ensures the security and validity of the encryption key, thereby ensuring the confidentiality caused by key leaks or improper management. A key management system ensures the security and validity of the encryption key, thereby ensuring the confidentiality and integrity of encrypted data. A key management system is composed of key generation and key distribution modules. Traditional key generation methods are designed through random number generators to ensure that the generated key is unpredictable. The random number generator is based on a complex algorithmic process where each key is unique and difficult to crack. Key distribution methods are designed by key exchange algorithms and are used in encryption algorithms and authentication algorithms.

Regarding key management technology, due to the high efficiency and low complexity of single-key key management solutions, attackers can easily detect key leaks throughout the network. Subhasish et al. [13] proposed various problems in password and key-based authentication and gave useful facts for authentication using biometric information. Using high-entropy-based information to demonstrate biometric authentication is superior to traditional authentication methods. Biometric keys cannot be easily copied, guessed, stolen, lost, distributed, or misremembered. Tian Y et al. [14] proposed a blockchain-based secure key management scheme. Compared with using traditional encryption algorithms, if an attacker obtains the key, it is possible to destroy the entire communication process and leak sensitive information. Therefore, designing a key management system to generate and distribute the key is particularly important in the entire in-vehicle bus network security system.

3. Results—Key Management System

3.1. Key Generation

Multimedia information (such as audio and video information) in In-Vehicle Ethernet has unpredictable characteristics. This study combines unpredictable characteristics of In-Vehicle Ethernet and designs a key generation system by using a random number generator. The process of key generation is shown in Figure 1.

The integer K is obtained by utilizing the remainder operation between the physical address Addr of the ECU and the finite field p, as shown in Formula (1).

K = Addr mod p

(1)

Select the latest K records from multimedia information records of In-Vehicle Ethernet to create a multimedia information list. Since this multimedia information list is obtained from dynamic and random multimedia information, it conforms to the dynamic characteristics of In-Vehicle Ethernet information and the design rules of the random number generator. In order to ensure the security of the generated random number generator, this study uses a hash algorithm to generate the random number key RAND_Key, as shown in Formula (2).

RAND_Key = HMAC (Addr + List)

(2)

In the random number key generation process, the dynamic entertainment information of an in-vehicle multimedia system is used, thereby increasing the randomness and unpredictability of the key. A multi-layered and highly secure key generation mechanism is constructed due to the residual operation of the physical address of the In-Vehicle Ethernet ECU and the application of encryption algorithms.

The remainder operation and the hash algorithm are fundamental operations in cryptography, and the implementation of these algorithms is not particularly complex, and they perform efficiently on modern processors. Due to the dynamic and random nature of the multimedia messages used, this approach is computationally more efficient than traditional key generation methods based on fixed algorithms, and because the algorithms involved are mainly mathematical, their power consumption is relatively low. In addition, key generation depends on the dynamic characteristics of multimedia information, which increases the difficulty of side-channel attacks. So, the algorithms in this section perform well in terms of computational power, and time required for operations, and have high efficiency. In terms of power consumption, it is expected to be low due to the high efficiency of the algorithm. As for the resistance to side-channel attacks, the method itself has some resistance.

3.2. Key Distribution

The role of key distribution is to ensure that both communicating parties can safely share the encryption key and the authentication key to ensure the confidentiality and integrity of communication. The encryption key is used to protect the privacy of data, while the authentication key is used to verify the identity of communicating parties, thereby preventing identity masquerading attacks. In the key distribution process, there are mainly key security, key management complexity, and security threats during the transmission. In order to further improve the confidentiality and reliability of the secret key, this paper innovatively introduces the idea of the digital envelope and proposes a key distribution method. This key distribution method divides the key into the encryption key and the authentication key, thereby improving the security of key distribution.

Table 1. Symbols and functional description.

Symbols	Functional Description
ECU_Tx	Sending node
ECU_Rx	Receiving node
${\mathrm{p}\mathrm{k}}_{\mathrm{T}\mathrm{x}}$	Public key of the sending node
${\mathrm{s}\mathrm{k}}_{\mathrm{T}\mathrm{x}}$	Private key of the sending node
${\mathrm{p}\mathrm{k}}_{\mathrm{R}\mathrm{x}}$	Public key of the receiving node
${\mathrm{s}\mathrm{k}}_{\mathrm{R}\mathrm{x}}$	Private key of the receiving node
C	Ciphertext
MIE	Plaintext
${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{{\mathrm{R}}_1}$$, {\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{{\mathrm{R}}_2}$	Using random number Key RAND_Key
${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{a}}$$, {\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{b}}$	A random number mapped to a finite field
S	Secret key
EK	Encryption key
AK	Authentication key
M1	Message sent by ECU_Tx to ECU_Rx
M2	Message sent by ECU_Rx to ECU_Tx

is a description of the symbols and functions used in the key distribution process.

Since the public key certificate is built into the vehicle electronic control unit (ECU), traditional key distribution uses public key cryptography to encrypt information. The public key is encrypted in the sending node (ECU_Tx), and the receiving node (ECU_Rx) receives the message and decrypts it through its own private key. In order to make up for the low efficiency of the secret key distribution in the public key cryptography method, this paper proposes a key distribution method based on the elliptic curve secp256k1 domain combined with the Schnorr algorithm, as shown in Figure 2.

The ECU_Tx node uses the random number key RAND_Key to obtain the random number ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{{\mathrm{R}}_1}$. The random number ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{{\mathrm{R}}_1}$ is mapped to the finite field of size q through modular operation to obtain the random number ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{a}}$, as shown in Formula (3).

$${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{a}}={\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{{\mathrm{R}}_1} \mathrm{m}\mathrm{o}\mathrm{d} \mathrm{q}$$

(3)

Through the dot multiplication operation on the elliptic curve, the elliptic curve base point G of the intermediate result ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{a}}$ times of the key is expressed as ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{a}}\times \mathrm{G}$. In the ECU_Tx node, use the private key ${\mathrm{s}\mathrm{k}}_{\mathrm{G}}$ to sign ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{a}}\times \mathrm{G}$. By calculating the message ${\mathrm{M}}_1$ sent by the ECU_Tx node to the ECU_Rx node, as shown in Formula (4).

$${\mathrm{M}}_1={\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{a}}\times \mathrm{G}\left|\right|\mathrm{s}\mathrm{i}\mathrm{g}\mathrm{n}({\mathrm{s}\mathrm{k}}_{\mathrm{T}\mathrm{x}},\mathrm{H}({\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{a}}\times \mathrm{G}\left)\right)$$

(4)

Among them, $\mathrm{H}({\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{a}}\times \mathrm{G})$ is the hash operation of ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{a}}\times \mathrm{G}$, which is used to generate a fixed-length digest $\mathrm{s}\mathrm{i}\mathrm{g}\mathrm{n}({\mathrm{s}\mathrm{k}}_{\mathrm{T}\mathrm{x}},\mathrm{H}({\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{a}}\times \mathrm{G})$) is the result of signing the hash value $\mathrm{H}({\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{a}}\times \mathrm{G})$ using the private key ${\mathrm{s}\mathrm{k}}_{\mathrm{T}\mathrm{x}}$ of the private key ${\mathrm{s}\mathrm{k}}_{\mathrm{T}\mathrm{x}}$ of the ECU_Tx node. The symbol $\left|\right|$ represents ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{a}}\times \mathrm{G}$ and $\mathrm{s}\mathrm{i}\mathrm{g}\mathrm{n}({\mathrm{s}\mathrm{k}}_{\mathrm{T}\mathrm{x}},\mathrm{H}({\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{a}}\times \mathrm{G})$) and is sent to the ECU_Rx node at the same time.

When the ECU_Rx node receives the message ${\mathrm{M}}_1$, it uses the private key ${\mathrm{s}\mathrm{k}}_{\mathrm{R}\mathrm{x}}$ of the ECU_Rx node and the public key ${\mathrm{p}\mathrm{k}}_{\mathrm{T}\mathrm{x}}$ of the ECU_Tx node to decrypt and verify the signature, respectively, thereby obtaining the session key intermediate result ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{a}}\times \mathrm{G}$. The ECU_Rx node uses the random number key RAND_Key to obtain the random number ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{{\mathrm{R}}_2}$, and maps the random number ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{{\mathrm{R}}_2}$ to a finite field of size q through modular operation to obtain the random number ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{b}}$, as shown in Formula (5).

$${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{b}}={\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{{\mathrm{R}}_2} \mathrm{m}\mathrm{o}\mathrm{d} \mathrm{q}$$

(5)

Calculate the session key s as a point on the elliptic curve, as shown in Formula (6). Among them, ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{a}}\times {\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{b}}\times \mathrm{G}$ is the result of the dot multiplication of two random numbers ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{a}}$ and ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{b}}$ and the base point G of the elliptic curve.

$$\mathrm{s}={\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{a}}\times {\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{b}}\times \mathrm{G}$$

(6)

Through the dot multiplication operation on the elliptic curve, the elliptic curve base point G, that is, ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{b}}$ times the intermediate result of the key, is expressed as ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{b}}\times \mathrm{G}$. By calculating the message ${\mathrm{M}}_2$ sent by ECU_Rx to ECU_Tx, as shown in Formula (7).

$${\mathrm{M}}_2={\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{b}}\times \mathrm{G}\left|\right|\mathrm{s}\mathrm{i}\mathrm{g}\mathrm{n}({\mathrm{s}\mathrm{k}}_{\mathrm{R}\mathrm{x}},\mathrm{H}\left({\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{b}}\times \mathrm{G}\right))$$

(7)

Among them, $\mathrm{H}({\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{b}}\times \mathrm{G})$ is the hash operation of ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{b}}\times \mathrm{G}$, which is used to generate a fixed-length digest. $\mathrm{s}\mathrm{i}\mathrm{g}\mathrm{n}({\mathrm{s}\mathrm{k}}_{\mathrm{R}\mathrm{x}},\mathrm{H}\left({\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{b}}\times \mathrm{G}\right))$ is the result of signing the hash value $\mathrm{H}\left({\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{b}}\times \mathrm{G}\right)$ using the private key ${\mathrm{s}\mathrm{k}}_{\mathrm{R}\mathrm{x}}$ of the ECU_Rx node. The symbol $\left|\right|$ represents ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{b}}\times \mathrm{G}$ and $\mathrm{s}\mathrm{i}\mathrm{g}\mathrm{n}({\mathrm{s}\mathrm{k}}_{\mathrm{R}\mathrm{x}},\mathrm{H}\left({\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{b}}\times \mathrm{G}\right))$ and is sent to the ECU_Tx node at the same time.

When the ECU_Tx node receives the message ${\mathrm{M}}_2$, it uses the private key ${\mathrm{s}\mathrm{k}}_{\mathrm{T}\mathrm{x}}$ of the ECU_Tx node and the public key ${\mathrm{p}\mathrm{k}}_{\mathrm{R}\mathrm{x}}$ of the ECU_Rx to perform decryption and signature verification, respectively, and obtains the session key intermediate result ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{b}}\times \mathrm{G}$. Calculate the session key s as the point ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{a}}\times {\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{b}}\times \mathrm{G}$ on the elliptic curve.

Both the ECU_Tx and ECU_Rx nodes obtain the key ${\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{a}}\times {\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{D}}_{\mathrm{b}}\times \mathrm{G}$ distributed in the key distribution process. Mark the obtained key as a point on the elliptic curve secp256k1, represented by the coordinate point (EK, AK). Among them, the abscissa EK is the encryption key, and the ordinate AK is the authentication key. According to the above key distribution scheme based on the elliptic curve secp256k1 domain combined with the Schnorr algorithm, it is described in pseudo-code, as shown in Algorithm 1.

Algorithm 1 Session Key Distribution Algorithm

Input: Prime modulus p, Generator point G, Private keys private key A, private     key B, Public keys public key A, public key B  Output: Session Key

1:$//\mathrm{A} \mathrm{selects} \mathrm{random} \mathrm{number} {RAND}_a$ 2: ${RAND}_a$ + Key Distribution.get random num(p) 3:$//\mathrm{Calculate} {RAND}_a$ $\times$ G 4: ${RAND}_a$$G\leftarrow {RAND}_a \times$ G 5://Sign with own private key 6: S A←Schnorr Sign.sign(private key A, ${RAND}_a$G.to bytes(), G, p) 7:$//\mathrm{Send} {RAND}_a$$G \left|\right|$S A to B 8://Verify signature S A 9: flag1←Schnorr Sign.verify(public key A, ${RAND}_a$G.to bytes(),S A,G) 10: if not flag1 then 11:   throw Exception(“Signature verification failed!”) 12: end if 13:$//\mathrm{B} \mathrm{selects} \mathrm{random} \mathrm{number} {RAND}_b$ 14: ${RAND}_b$←Key Distribution.get random num(p) 15:$//\mathrm{Calculate} {RAND}_b$ $\times$ G 16: ${RAND}_b$$G\leftarrow {RAND}_b \times$ G 17://Sign with own private key 18: S B←Schnorr Sign.sign(private key B, ${RAND}_b$G.to bytes(), G, p) 19:$//\mathrm{Send} {RAND}_b$$G \left|\right|$S B to A 20://Verify signature S B 21: flag2←Schnorr Sign.verify(public key B, ${RAND}_b$G.to bytes().S B,G) 22: if not flag2 then 23:   throw Exception(“Signature verification failed!”) 24: end if 25:$//\mathrm{Calculate} {RAND}_a$${RAND}_b$G 26: ${RAND}_a$ ${RAND}_{b}G$←${RAND}_a$ $\times$ ${RAND}_b$G 27://Use EK-coordinate as encryption key 28: EK←${RAND}_a$ ${RAND}_b$G. EK() 29://Convert int to bytes 30: enc key←EK.to bytes((EK.bit length() + 7)//8,byteorder =’big’) 31://Use y-coordinate as authentication key 32: AK←${RAND}_a$ ${RAND}_b$G. AK() 33://Convert int to bytes 34: auth key←AK.to bytes((AK.bit length()+ 7)//8,byteorder =’big’) 35://Convert to Session Key format 36: session key←Session Key(enc key,auth key) 37: return session key

3.3. Application of Key Management System

This study verified the versatility of a key management system by utilizing the commonly used AES-128 encryption algorithms in a key management system.

In the process of sending information from the ECU_Tx node to the ECU_Rx node, the plaintext of the vehicle-mounted Ethernet multimedia data that the two nodes need to interact with is MIE. The encryption key EK is a 256-bit key. In this study, the first 128 bits of the encryption key EK are used as the key of the AES-128 encryption algorithms (${\mathrm{E}\mathrm{K}}_{\mathrm{A}\mathrm{E}\mathrm{S}-128}$). The authentication key AK is a 256-bit key. In this study, the first 128 bits of the authentication key AK are used as the authentication key (${\mathrm{A}\mathrm{K}}_{\mathrm{H}\mathrm{M}\mathrm{A}\mathrm{C}}$). The ECU_Rx node uses the encryption key (${\mathrm{E}\mathrm{K}}_{\mathrm{A}\mathrm{E}\mathrm{S}-128}$) to encrypt the plaintext ${\mathrm{M}\mathrm{I}\mathrm{E}}_{\mathrm{T}\mathrm{x}}$ and obtain the ciphertext C, as shown in Formula (8).

$$\mathrm{C}=\mathrm{Enc} ({\mathrm{M}\mathrm{I}\mathrm{E}}_{\mathrm{T}\mathrm{x}}, {\mathrm{E}\mathrm{K}}_{\mathrm{A}\mathrm{E}\mathrm{S}-128})$$

(8)

The authentication key (${\mathrm{A}\mathrm{K}}_{\mathrm{H}\mathrm{M}\mathrm{A}\mathrm{C}}$) is calculated through the hash function to obtain the message authentication code MAC_Tx, as shown in Formula (9). This further prevents the message from being tampered with, and sends the ciphertext C and message authentication code MAC_Tx to the ECU_Tx node.

$$\mathrm{MAC}\_\mathrm{Tx}=\mathrm{H} \left({\mathrm{M}\mathrm{I}\mathrm{E}}_{\mathrm{T}\mathrm{x}}\right|\left|{\mathrm{A}\mathrm{K}}_{\mathrm{H}\mathrm{M}\mathrm{A}\mathrm{C}}\right)$$

(9)

When the node receives the ciphertext and message authentication code, it uses the encryption key (${\mathrm{E}\mathrm{K}}_{\mathrm{A}\mathrm{E}\mathrm{S}-128}$) to decrypt the ciphertext C to obtain the plaintext message ${\mathrm{M}\mathrm{I}\mathrm{E}}_{\mathrm{R}\mathrm{x}}$. The message authentication code ${\mathrm{M}\mathrm{I}\mathrm{E}}_{\mathrm{R}\mathrm{x}}$ is calculated through the hash function, as shown in Formula (10).

$$\mathrm{MAC}\_\mathrm{Rx}=\mathrm{H} \left({\mathrm{M}\mathrm{I}\mathrm{E}}_{\mathrm{R}\mathrm{x}}\right|\left|{\mathrm{A}\mathrm{K}}_{\mathrm{H}\mathrm{M}\mathrm{A}\mathrm{C}}\right)$$

(10)

If the message authentication code MAC_Rx of the receiving node is the same as the message authentication code MAC_Tx of the sending node, it proves that the plaintext message MIE is valid.

4. Model Validation

The experiments are simulated using CANoe. CANoe is a widely recognized and powerful simulation tool especially suited for the development, testing, and verification of automotive networks and communication systems. It supports a wide range of communication protocols, including Ethernet, which makes it ideal for simulating and testing Ethernet networks within vehicles. CANoe is able to accurately simulate and analyze the behavior of the network without involving the actual hardware, which is essential for the initial validation of algorithm effectiveness and performance.

CANoe can simulate individual nodes in a network, including sending and receiving network messages, to test the correctness of network communications. It provides diagnostic services simulation and testing capabilities to help developers verify the diagnostic response of the ECU (electronic control unit), and can perform real-time data collection and analysis, to assess network performance, such as communication delays, packet loss, and so on.

Since multimedia information has strong random characteristics, the random number key generated using multimedia information is unpredictable. At the same time, this article uses remainder operation and encryption algorithms to build a multi-level, highly secure key generation method, thus further improving the security of key information.

This study uses the order of key address, key, and ciphertext to form a key file, and generates parameters such as key address and the key length in the key management system, as shown in Figure 3.

In order to further verify the performance of the random number key generation method proposed in this study, the performance is compared with the GFSRG [15] and super prime number method [16] methods in the same experimental environment of vehicle multimedia information, as shown in Figure 4. Experimental simulation results show that as the number of samples increases, the chi-square test statistics of the three methods tend to move away from the rejection region, and the credibility gradually increases. For the same number of samples, the key generation random number algorithms proposed in this article are more evenly distributed in the area than other algorithms.

The key distribution method proposed in this article is implemented in combination with the Schnorr algorithm on the elliptic curve secp256k1 domain. In order to further verify the performance of the key distribution method proposed in this study, the experimental environment was set up with the key length in the integer domain being 2048 bits and the key length in the elliptic curve domain being 256 bits, and then combined on the elliptic curve secp256k1 domain. The overhead time performance of five signature algorithms including Okamoto [17], ElGamal [18], ECDSA [19], RSA [20], and Schnorr [21] was compared, as shown in Figure 5. Experimental results show that the key distribution overhead time based on the Schnorr signature algorithm is significantly lower than on other signature algorithms.

The storage overhead, storage space, and key negotiation time of the above five signature algorithms are shown in

Table 2. Storage Overhead, Space, and Key Negotiation Time.

	ECDSA	Okamoto	ElGamal	RSA	Schnorr
Memory overhead	$2\left|{\mathrm{Z}}_{\mathrm{P}}^{\mathrm{*}}\right|$	$3\left|{\mathrm{Z}}_{\mathrm{q}}^{\mathrm{*}}\right|$	$2\left|{\mathrm{Z}}_{\mathrm{q}}^{\mathrm{*}}\right|$	$\left|{\mathrm{Z}}_{\mathrm{q}}^{\mathrm{*}}\right|$	$\left|{\mathrm{G}}_1\right|+\left|{\mathrm{Z}}_{\mathrm{P}}^{\mathrm{*}}\right|$
Storage space (byte)	64	768	512	256	65
Key agreement time: (ms)	3.234	14.733	10.233	29.4	1.533

. Among them, $\left|{\mathrm{G}}_1\right|$ is the storage space occupied by a point on the elliptic curve, $\left|{\mathrm{Z}}_{\mathrm{P}}^{\mathrm{*}}\right|$ is the length of the element in the integer finite field composed of the order p of the elliptic curve, and $\left|{\mathrm{Z}}_{\mathrm{q}}^{\mathrm{*}}\right|$ is the size of the storage space occupied by the elements on the integer field. The key negotiation time is the time required by the key management system in the process of key generation and distribution.

Experimental results show that the efficiency of overhead time, storage overhead, storage space, and the secret key negotiation time based on the Schnorr signature algorithm is higher than other signature algorithms. The key negotiation time of the In-Vehicle Ethernet key management method proposed in this study is 1.533 ms, which is 80.5% higher than the traditional key management method. Therefore, the key management method proposed in this article is suitable for network security systems applied in In-Vehicle Ethernet systems with higher real-time characteristics.

5. Conclusions

In-Vehicle Ethernet has been widely used in Infotainment Systems, Advanced Driving Assistance Systems (ADASs), Vehicle Control Systems, etc., and has become the backbone of in-vehicle bus networks. This paper proposes a key management method based on key generation and key distribution to address the real-time and network security issues of In-Vehicle Ethernet. The key generation module utilizes the unpredictability of in-vehicle multimedia information to generate more secure and random keys by combining the residue operation of physical addresses and encryption algorithms. The key distribution module uses the elliptic curve secp256k1 domain combined with the Schnorr algorithm to divide the key into encryption key and authentication key, and ensures the secure distribution of the key through digital envelope technology. This method improves the efficiency and security of key distribution. Through experimental simulation, this key management system improves the key negotiation speed by 80.5% compared with the traditional key management system, which shows its improvement in real-time information processing efficiency and lays the foundation for the stable development of intelligent networked vehicles. The results show that this key management method is better than the traditional method in terms of computation overhead, storage overhead, storage space, and key negotiation time, and it can be used flexibly with traditional encryption algorithms such as AES-128. However, for the proposed key distribution scheme based on elliptic curve secp256k1 domains and the Schnorr signature algorithm, quantum computation affects the security of the algorithm, and while the Schnorr signature algorithm is considered secure in classical computation, quantum computers may find ways to break it quickly, as computations on elliptic curves may no longer be as difficult on quantum computers.