Cognitive internet of vehicles security spectrum auction sharing method based on idqn blockchain

Abstract

The present application relates to a cognitive vehicle networking security spectrum auction sharing method based on IDQN blockchain, comprising: constructing a blockchain-based vehicle networking; in the spectrum demand matching process of the vehicle networking, obtaining the spectrum idle amount and spectrum demand amount of authorized vehicles and cognitive vehicles, conditionally screening the spectrum idle amount and spectrum demand amount, and combining the optimal bandwidth and optimal bid to obtain an optimal scheme; in the spectrum matching process of the optimal scheme, obtaining the utility satisfaction and throughput, establishing an optimization target based on the utility satisfaction and throughput, taking the optimization target as a decision process, and using an IDQN learning model to solve the decision process through a state space, an action space, a reward function and a spectrum allocation strategy to obtain an optimal action, i.e. an optimal spectrum allocation decision. The present application realizes the optimal spectrum allocation strategy and improves the system throughput based on the utility function design, sorting mechanism and IDQN algorithm of the auction algorithm.

Description

Technical Field

[0001] This invention relates to the fields of blockchain and smart contract technology, and in particular to a cognitive vehicle-to-everything (V2X) security spectrum auction and sharing method based on the IDQN blockchain. Background Technology

[0002] Intelligent Transportation Systems (ITS) are increasingly attracting global attention as a transformative force in the transportation sector. By 2025, international mobile communication spectrum demand will reach 1340-1960 MHz, facing significant challenges such as insufficient spectrum resources and inefficient spectrum allocation. Addressing the mismatch between urgent spectrum demand and supply has become a major focus in reducing obstacles to the future development of Cognitive Vehicle-to-Everything (CIoV). Given the low spectrum utilization efficiency and significant limitations of static spectrum sharing, Cognitive Radio (CR) Dynamic Spectrum Sharing (DSS) is gradually becoming the mainstream application. It can further tap the potential of limited spectrum resources by using sensing technology to detect and utilize idle spectrum resources of licensed users. This plays a crucial role in 5G and above systems. Reallocating temporarily idle resources to improve spectrum utilization is an effective means of addressing the current problem of uneven spectrum resource allocation.

[0003] For dynamic spectrum sharing schemes in cognitive vehicle-to-everything (V2X) systems, traditional implementations tend towards centralized spectrum management based on databases. The stored data can be used to optimize spectrum allocation strategies, thereby reducing system management costs and improving spectrum utilization efficiency. However, centralized spectrum management suffers from drawbacks such as low flexibility and vulnerability to attacks. Addressing the issues of low base station flexibility and privacy breaches in cognitive V2X systems, blockchain, as a decentralized data distribution framework, has attracted significant interest in mitigating the impact of centralized storage frameworks. Its decentralized, immutable, and trustworthy characteristics allow blockchain, as a ledger, to update recorded information without a centralized coordinator or controller. Especially when combined with recent 5G and other technology-supported communication systems (such as edge computing, fog computing, and joint optimization), it can be used to maintain auditable and traceable data records for various applications. Therefore, this invention utilizes blockchain as an effective distributed database system to achieve dynamic allocation and data storage of spectrum resources, improving the security and utilization rate of the spectrum sharing system. Existing technical solutions also disclose a database for dynamic spectrum sharing based on the decentralized characteristics of blockchain. In non-cooperative spectrum sharing, a spectrum management scheme can be provided through a market mechanism, and the spectrum information of the entire system can be updated. Blockchain enables decentralized recording and management of idle spectrum resources, while also recording the spectrum access transaction process and spectrum usage in the blockchain, ensuring the security and trustworthiness of the dynamic spectrum access process and achieving transparency and traceability of spectrum transaction data. Existing technical solutions also disclose a blockchain spectrum sharing framework under a multi-operator spectrum sharing mechanism for dynamic spectrum sharing between primary and secondary users. A Multi-OPs Spectrum Sharing (MOSS) smart contract is proposed for spectrum transactions and payments between primary and secondary users, and an incentive and penalty mechanism is set up to curb malicious nodes. Existing technical solutions also propose a spectrum sensing service based on smart contracts. Nodes using smart contracts to describe sensor service data for spectrum sensing can only receive corresponding rewards according to the contract, thereby ensuring the security and accuracy of spectrum resource data and ensuring the subsequent implementation of spectrum sharing. Existing technical solutions also propose dynamic spectrum sharing systems based on blockchain technology, introducing spectrum tokens to verify and track spectrum resource usage, and utilizing smart contract-related algorithms to achieve dynamic sharing of spectrum resources. In summary, existing research also demonstrates that spectrum resource management and sharing have significant advantages in establishing trust, node consensus, and reducing transaction costs based on blockchain platforms.

[0004] With the rapid development of intelligent transportation technology and increasing emphasis on privacy and spectrum communication security, the aforementioned technologies face the urgent challenge of simultaneously improving communication performance and ensuring secure spectrum sharing. Blockchain, as a technology gradually being applied across various sectors, possesses the capability to address these challenges due to its ability to create a stable and reliable mutual trust environment in untrusted distributed systems. Existing technological solutions also propose a cryptocurrency-based incentive mechanism aimed at guiding demand-side users to share spectrum resources, thereby optimizing information transmission and allocation strategies in a decentralized environment and improving future channel revenue. Research presented in existing technological solutions also indicates that unorganized dynamic spectrum sharing can lead to severe interference problems, causing not only system chaos but also significant loss of spectrum capacity. To overcome these limitations, existing technological solutions propose constructing a blockchain-driven platform that introduces spectrum digital tokens to verify and track authorized frequency band usage and regulate the access order of secondary users to reduce interference. In existing technological research, interference and noise are unavoidable in both V2V and V2I communication, including interference from spectrum resource sharing and interference between vehicle devices, requiring consideration of a variety of factors. To address the shortcomings of current research that primarily utilizes spectrum sensing technology to obtain idle spectrum information, existing solutions propose a reverse auction mechanism to incentivize edge nodes to provide caching services and prevent selfish behavior by encouraging authorized users to share idle spectrum. To ensure security and enhance trust, existing solutions explore untrusted, secure, and distributed auction schemes based on the Ethereum blockchain framework to improve spectrum utilization in the absence of a centralized party. Furthermore, to alleviate privacy concerns among system participants and incentivize spectrum trading, existing solutions improve the proposed dynamic spectrum sharing dual-auction mechanism using a differential privacy protection method, preventing inference attacks on the auction process.

[0005] The aforementioned technical solutions, based on a secure and reliable blockchain system architecture, reduce privacy concerns among vehicle users participating in spectrum sharing and enhance their willingness to share. However, due to the scarcity of spectrum resources and their inherent competitive nature, designing a reasonable and effective spectrum allocation and incentive mechanism is crucial to addressing this shortage. Spectrum resource auctions primarily aim to resolve spectrum allocation issues among vehicles and effectively coordinate the use of different frequency bands within the spectrum, thereby reducing interference between vehicle communications and ensuring the reliability and security of the communication system. Secondly, combining a dynamic spectrum sharing method based on a reasonable reward mechanism with a blockchain-based resource sharing trading platform offers significant advantages in improving spectrum resource utilization and enhancing vehicle users' enthusiasm for participating in spectrum sharing. Furthermore, considering the consumption of computing resources in distributed systems and the intelligent dynamic allocation of spectrum resources, traditional methods struggle to adapt to complex and changing communication environments. To address these issues, deep reinforcement learning algorithms can be introduced, leveraging their distributed autonomous learning and rapid environmental adaptability to achieve decentralized intelligent spectrum management, reducing the computational and communication overhead of centralized methods while improving security.

[0006] The emergence of existing technologies demonstrates that DRL (Digital Relationship Management) can solve computationally intensive problems in large-scale state and action spaces. Agents obtain feature values ​​from high-dimensional raw data to approximate optimal action strategies. DRL is also applied in blockchain-based cognitive vehicular communication networks, used in the optimal decision-making problem of spectrum allocation. For example, existing technologies have researched and designed a DRL-based spectrum access and power control method. This method estimates cumulative rewards through neural networks and expands the state-action set of the Q-table, achieving the optimal strategy for spectrum sharing under power and interference constraints. Existing technologies propose a multi-task DQN (Distributed Quality Management) method to determine the most suitable channel and power level in sub-bands. This method effectively improves the QoS level of the SU (Sustainable Spectrum Array) and significantly shortens the training convergence time of DQN. Existing technologies provide a flexible resource-sharing market for V2V spectrum resource sharing to achieve low-latency communication services. In the process of resource sharing, security and privacy protection are also significant challenges. Existing technical solutions propose a blockchain-based scheme for secondary and primary users to collaborate in constructing anonymous regional privacy protection measures, while using integrity scores to measure user trustworthiness and establish a trustworthiness incentive mechanism. Furthermore, existing technical solutions not only consider the dynamic characteristics of resource availability and vehicle location, but also propose a lightweight blockchain-based resource sharing scheme that applies DRL-based resource pricing to smart contracts, improving the efficiency of spectrum management and allocation. Other existing technical solutions propose a distributed DSS communication scheme, treating each cell in a multi-user multiple-input multiple-output (MU-MIMO) licensed network as an intelligent agent and rewarding it with an average signal-to-noise ratio.

[0007] However, due to the limited computing resources of vehicles and the scarcity of spectrum resources, most of the aforementioned spectrum sharing incentive mechanisms consider auction-based algorithms and enhancing vehicle users' reputation scores to increase their enthusiasm for participating in spectrum sharing. However, this design does not consider the actual situation of vehicle communication, such as actual interference between vehicles. For spectrum sharing schemes, combining consideration of actual interference with utility functions based on auction algorithms can achieve a more reasonable spectrum allocation strategy, maximize utility satisfaction, and effectively improve system throughput. Summary of the Invention

[0008] To address the problems existing in the prior art, the present invention aims to propose a cognitive vehicle-to-everything (V2X) security spectrum auction and sharing method based on IDQN blockchain. It proposes a utility function design based on an auction algorithm, a sorting mechanism, and the IDQN algorithm to achieve optimal spectrum allocation and improve system throughput, considering interference between vehicle users. To achieve the above objective, the present invention provides the following solution:

[0009] A cognitive vehicle-to-everything (V2X) security spectrum auction and sharing method based on the IDQN blockchain includes:

[0010] To build a blockchain-based vehicle network, during the spectrum demand matching process of the vehicle network, the available spectrum and spectrum demand of authorized vehicles and cognitive vehicles are obtained. The available spectrum and spectrum demand are filtered by conditions, and the optimal solution is obtained by combining the optimal bandwidth and the optimal bid.

[0011] During the spectrum matching process of the optimal solution, the utility satisfaction and throughput are obtained. An optimization objective is established based on the utility satisfaction and throughput. The optimization objective is used as a decision-making process. The decision-making process is solved using the IDQN learning model through the state space, action space, reward function and spectrum allocation strategy to obtain the optimal action, i.e. the optimal spectrum allocation decision.

[0012] Optionally, constructing the vehicle network includes:

[0013] A vehicle network based on a consortium blockchain is formed by a number of authorized vehicles and cognitive vehicles, wherein the authorized vehicles and cognitive vehicles in the vehicle network are located within the area covered by the consortium blockchain composed of multiple base stations.

[0014] Before the spectrum demand matching stage of the vehicle network, the authorized vehicle and the cognitive vehicle are encrypted and initialized, and a public key and a private key pair are generated through the vehicle user's account address.

[0015] Optionally, obtaining the optimal solution includes:

[0016] Exclude licensed vehicles that do not meet the spectrum demand requirements, and select licensed vehicles that meet the signal-to-interference-plus-noise ratio (SINR) threshold from the licensed vehicles that meet the requirements. Exclude cognitive vehicles that do not meet the spectrum availability requirements, and select cognitive vehicles that meet the SINR threshold.

[0017] Obtain the optimal bandwidth and optimal bid of the authorized vehicle and the cognitive vehicle, compare them, obtain a bidding set, and based on the bidding set, send auction requests for the authorized vehicle and the cognitive vehicle respectively:

[0018] Sending the auction request for the authorized vehicle includes:

[0019] ;

[0020] in, For auction request information, A digital signature for the private key of the authorized vehicle PV; For the first The cognitive vehicle (SV) can meet the signal-to-interference-plus-noise ratio (SIR) of the licensed vehicle (PV), meaning that the cognitive vehicle does not interfere with the normal communication of the licensed vehicle (PV). The timestamp when the information was uploaded. This represents the set of possible optimal bids for the authorized vehicle set PVs;

[0021] Sending auction requests for each of the cognitive vehicles includes:

[0022] ;

[0023] in, For auction request information, Digital signature for the private key of the cognitive vehicle SV. For the first The PV of the authorized vehicle can meet the signal-to-interference-plus-noise ratio of the cognitive vehicle SV. This represents the set of possible optimal bids for the cognitive vehicle set SVs;

[0024] When the base station node receives the auction requests from the authorized vehicle and the cognitive vehicle respectively, it performs decryption and digital signature verification using the private keys of the authorized vehicle and the cognitive vehicle, and performs hash decryption of the information. If they match completely, it proves that the information has not been tampered with, and the optimal solution is generated.

[0025] Optionally, obtaining the utility satisfaction and the throughput includes:

[0026] When the authorized vehicle occupies the channel, the utility satisfaction of the authorized vehicle includes:

[0027] ;

[0028] in, For the first match Cognitive Vehicle SV for the first The signal-to-interference-plus-noise ratio (SIR) generated by the cognitive vehicle's SV represents the degree of interference between the cognitive vehicle's SV and the licensed vehicle's PV. Indicates the first The pricing of idle spectrum resources for each licensed vehicle (PV). This indicates the amount of free spectrum resources that authorized vehicles (PVs) can share. and Weight parameters used for PV calibration of authorized vehicles;

[0029] The throughput of the authorized vehicle for the cognitive vehicle without affecting the authorized vehicle includes:

[0030] ;

[0031] in, This represents the bandwidth occupied by the channel;

[0032] The perceived vehicle's utility satisfaction includes:

[0033] ;

[0034] in, For the first match PV to the first Signal-to-interference-plus-noise ratio (SIR) generated by a cognitive vehicle SV. Indicates the first Pricing of spectrum resources required by a cognitive vehicle (SV). Indicates the first The amount of free spectrum resources required by a single cognitive vehicle (SV). and Weighting parameters used for vehicle SV calibration;

[0035] The throughput of the cognitive vehicle without affecting the authorized vehicle includes:

[0036] ;

[0037] in, This is expressed as the bandwidth of the channel.

[0038] Optionally, obtaining the utility satisfaction and the throughput further includes:

[0039] When the authorized vehicle does not occupy the channel, the utility satisfaction of the authorized vehicle includes:

[0040] ;

[0041] in, Satisfactory parameters for PV calibration of authorized vehicles;

[0042] The perceived vehicle's utility satisfaction includes:

[0043] ;

[0044] in, Is it other cognitive vehicle SV to the first The signal-to-interference-plus-noise ratio (SIR) of a vehicle occupying a channel for cognitive vehicle SV. and Weighting parameters used for vehicle SV calibration.

[0045] Optionally, establishing the optimization objective includes:

[0046] Optimization of utility satisfaction includes:

[0047] ;

[0048] :

[0049] ;

[0050] ;

[0051] ;

[0052] ;

[0053] in, For the overall social welfare of auction algorithms, The utility formula for authorized vehicle users in the process of dynamic spectrum sharing is expressed as follows. To understand the utility formula for vehicle users in the process of dynamic spectrum sharing, This means that the amount of spectrum resources occupied cannot exceed the amount of spectrum resources available. This represents the channel with the least amount of available free spectrum resources. This indicates the channel with the most available free spectrum resources. To understand the spectrum requirements of vehicle SV. For the free spectrum resources of authorized vehicles, For the preset threshold time, Represents a hash. For ECDSA computation time, This is the maximum threshold for ECDSA computation time;

[0054] Optimizing system throughput includes:

[0055] ;

[0056]

[0057] ;

[0058] ;

[0059] ;

[0060] ;

[0061] in, For the total throughput of the system, Indicates the first Channel throughput of authorized vehicle users Indicates the first Channel throughput of a cognitive vehicle (SV). This indicates the transmit power of the cognitive vehicle SV. This indicates the maximum threshold for transmission power. =1 indicates the first The cognitive vehicle SV successfully occupied the matched first Authorized vehicle PV idle channel, =0 indicates that the attempt to use the device was unsuccessful. , The minimum and maximum thresholds for signal-to-interference-plus-noise ratio (SIR) are given. The total number of authorized vehicle PVs participating in spectrum sharing. The total number of cognitive vehicle SVs participating in spectrum sharing.

[0062] Optionally, the state space is:

[0063] ;

[0064] in, For a union group containing multiple elements, The status of the authorized vehicle PV. To recognize the status of the vehicle's SV. For spectrum resource prices, This refers to the signal-to-interference-plus-noise ratio (SINR) in a real-world communication environment.

[0065] The action space is:

[0066] ;

[0067] in, This indicates that the cognitive vehicle (SV) has selected to occupy the matching PV idle channel. This indicates that the cognitive vehicle SV selects an unused PV idle channel;

[0068] The reward function is:

[0069] The priority of subsequent sorting is determined based on the number of defaults in the history of the authorized vehicle and the recognized vehicle:

[0070] The sorting function for the authorized vehicles is:

[0071] ;

[0072] ;

[0073] in, This indicates the time based on the sorting function of authorized vehicles. The ranking This indicates that, based on the utility function of the authorized vehicle, The initial descending order of time. The number of times a vehicle has engaged in dishonest conduct. As a constant coefficient, when When, it indicates no dishonest behavior; when When this occurs, it indicates dishonest behavior;

[0074] The cognitive vehicle ranking function is:

[0075] ;

[0076] ;

[0077] in, This indicates that the sorting function based on the cognitive vehicle SV is in Ranking of moments This indicates that the utility function of the cognitive vehicle SV is in The initial descending order of time. To determine the frequency of dishonest behavior by the vehicle SV, It is a constant coefficient, consistent with the constant coefficient of PV, when When, it indicates no dishonest behavior; when When this occurs, it indicates dishonest behavior;

[0078] Based on the aforementioned utility satisfaction and throughput, and considering dishonest behavior in the vehicle ranking function, a reward function is constructed and expressed using weighted forms:

[0079] ;

[0080] ;

[0081] ;

[0082] in, This represents the reward setting for the utility function. , Indicates the reward setting for system throughput. , This indicates whether the SV has successfully occupied the idle channel of the successfully matched PV. This indicates the weight settings for the optimization objective;

[0083] The spectrum allocation strategy is as follows:

[0084] ;

[0085] in, For the desired long-term cumulative discount rewards, This is the discount parameter.

[0086] Optionally, obtaining the optimal action includes:

[0087] The state space is input into the IDQN learning model for iteration, and the corresponding action in the action space is selected according to different spectrum allocation strategies until the optimal action is obtained:

[0088] ;

[0089] in, This is the optimal strategy for spectrum matching and sharing. In the state The optimal action found below The optimal Q value.

[0090] Optionally, the action iteration of the IDQN learning model is represented as:

[0091] ;

[0092] in, Indicates the state Next action The reward A discount factor to measure the degree of future returns. Use neural network parameters for the target network The obtained Q-value estimate for the action in the next state, Use neural network parameters for the current network The obtained Q-value estimate for the action in the current state.

[0093] The beneficial effects of this invention are as follows:

[0094] This invention proposes a platform for trading spectrum resources in the cognitive vehicle network based on consortium blockchain. It is suitable for spectrum management between PV and SV. Users are responsible for uploading information communication and avoid communication interference between vehicles by publishing their needs, thereby reducing communication costs.

[0095] This invention proposes an auction utility formula design that considers the signal-to-interference-plus-noise ratio (SINR). The aim is to comprehensively address interference issues in the communication environment and the spectrum resource requirements of both the PV and SV, achieving satisfaction for both parties and ensuring stable and efficient spectrum matching. Furthermore, under the interference constraint that the SV does not affect the normal communication of the PV, a multi-objective joint optimization objective combining system throughput and utility satisfaction is designed.

[0096] This invention proposes an IDQN learning algorithm, which includes multi-objective optimization decisions on utility satisfaction and system throughput, sets a reward function that tracks a ranking mechanism for historical dishonest behavior, and improves utility satisfaction and system throughput by exploring the optimal spectrum allocation strategy. Attached Figure Description

[0097] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0098] Figure 1 This is a schematic diagram of a blockchain-based vehicle networking system model according to an embodiment of the present invention;

[0099] Figure 2 This is a flowchart illustrating the process of uploading spectrum transaction information to the blockchain according to an embodiment of the present invention.

[0100] Figure 3 This is a sorting diagram of PV and SV according to an embodiment of the present invention;

[0101] Figure 4 This is a flowchart of a neural network according to an embodiment of the present invention. Detailed Implementation

[0102] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0103] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0104] This embodiment discloses a method for secure spectrum auction and sharing in a cognitive vehicle network based on IDQN blockchain, including: constructing a blockchain-based vehicle network; during the spectrum demand matching process of the vehicle network, obtaining the available spectrum and spectrum demand of authorized vehicles and cognitive vehicles; filtering the available spectrum and spectrum demand based on conditions; and obtaining the optimal solution by combining the optimal bandwidth and the optimal bid; during the spectrum matching process of the optimal solution, obtaining the utility satisfaction and throughput; establishing an optimization objective based on the utility satisfaction and throughput; using the optimization objective as a decision-making process; and solving the decision-making process using the IDQN learning model through the state space, action space, reward function, and spectrum allocation strategy to obtain the optimal action, i.e., the optimal spectrum allocation decision.

[0105] Specifically, with the increasing number of mobile vehicles accessing the network and the growing demand for security applications, spectrum resources are becoming increasingly scarce. Securing and efficiently utilizing limited idle spectrum resources has become a significant challenge for future mobile communication systems. Currently, cognitive radio dynamic spectrum sharing technology provides an effective solution to the spectrum resource shortage problem in vehicle-to-everything (V2X) communication. Addressing the low willingness of Primary Vehicles (PVs) and Secondary Vehicles (SVs) to share spectrum in Cognitive Internet of Vehicles (CIoV) due to privacy concerns, and considering the low transmission rates caused by interference between vehicles, this invention designs a secure spectrum auction sharing mechanism based on a Consortium Blockchain (CB) for CIoV and an improved Deep Q-Network (IDQN) learning algorithm to improve the throughput and utility satisfaction of this communication system. Simultaneously, a consortium blockchain-based CIoV spectrum resource trading and auction platform utilizes its transparency, traceability, and consensus mechanism to construct a spectrum resource tracking and management scheme suitable for vehicle users. Furthermore, considering the low efficiency of spectrum resource matching between PVs and SVs due to conflicts over the occupation of high-quality idle spectrum resources, an auction utility formula considering signal-to-interference-plus-noise ratio (SNR) and spectrum resource transaction prices was designed. Under the constraints of spectrum resource demand from both parties and meeting interference thresholds, a fast matching method for PV and SV resources and spectrum resource pricing guarantee to maximize the satisfaction of both PVs and SVs, thereby achieving safe and efficient spectrum sharing and incentivizing a long-term honest competitive relationship between PVs and SVs. In addition, to address the issue of dishonesty by PVs and SVs during spectrum resource matching, a ranking mechanism for tracking historical dishonest behavior of PVs and SVs was designed. An IDQN learning algorithm based on a consortium blockchain dual auction was proposed. By exploring the optimal spectrum allocation strategy, utility satisfaction and system throughput are improved, achieving optimization and stability of the dynamic spectrum sharing strategy between PVS and SVS. Finally, the improved algorithm was compared with traditional DQN and Q-learning algorithms using the Python platform. Simulation results show that in the IDQN algorithm, the utility satisfaction of PVs and SVs increased by 18% and 22%, respectively; and the system throughput increased by 15%.

[0106] Furthermore, the construction of the Internet of Vehicles (IoV) includes: using a number of authorized vehicles and cognitive vehicles to form an IoV based on a consortium blockchain, wherein the authorized vehicles and cognitive vehicles in the IoV are located within the area covered by the consortium blockchain composed of multiple base stations; before the spectrum demand matching stage of the IoV, the authorized vehicles and cognitive vehicles are encrypted and initialized, and public and private key pairs are generated through the account addresses of vehicle users.

[0107] Specifically, this invention considers a communication network model of a vehicle-to-everything (V2X) network based on a consortium blockchain, consisting of multiple base stations and A vehicles (PVs) and B vehicles (SVs). In a cognitive V2X environment, a model is constructed considering that PVs and SVs are located within the area covered by the consortium blockchain composed of multiple base stations. Vehicle users include PVs and SVs, where PVs = {PV1, ..., PVA, ..., PVA} represents the set of all authorized vehicles, and SVs = {SV1, ..., SVb, ..., SVB} represents the set of all cognitive vehicles. In the system model, PVs and SVs communicate with the base station nodes that make up the consortium blockchain, and PVs and SVs perform spectrum resource matching and sharing, such as... Figure 1 As shown.

[0108] To address the needs of secure spectrum sharing and mobile vehicle users, a consortium blockchain consisting of multiple base stations was deployed. Each base station can be equipped with a server to support substantial computing resource requirements. This provides security and timeliness for real-time spectrum resource transactions, while also meeting the requirements for subsequent on-chain transaction information and data storage.

[0109] To achieve dynamic spectrum sharing within the proposed consortium blockchain transaction framework, in time slots An internal SVS can occupy and transmit data when the PVS's spectrum is idle. If an SVS and PVS cannot achieve a complete match within a time slot, a next round of spectrum matching will be performed until the SVS is completely matched. Based on this, we consider two cases: the first is that the number of PVs and SVs is equal; the second is that the number of PVs is greater than the number of SVs. In a time slot... At this time, an SV can only purchase spectrum resources sold by one PV, and similarly, a PV can only sell spectrum resources to one SV, thereby achieving the effect of quick matching transactions, improving vehicle utility satisfaction and system throughput.

[0110] In a spectrum resource trading architecture based on consortium blockchain, the main information in the auctioned spectrum transactions includes the timestamp recording the transaction. Address of authorized vehicles Recognizing the vehicle's address Transaction amount collection Contract address Matching results and historical transaction information And a unique hash value for the transaction information Information is quickly verified and recorded using hash encryption. It can be represented as:

[0111] ;

[0112] To ensure identity privacy, an asymmetric elliptic curve digital signature algorithm (ECDSA) is used to encrypt and initialize the identities and information of PVs and SVs in the consortium blockchain. Referencing Ethereum's public and private key generation algorithm, the encryption algorithm is not certified by a third-party institution. Instead, it generates a public key (BK) and private key (VK) pair using the vehicle user's account address, ensuring that PVs and SVs can legally share spectrum and conduct transactions within the consortium blockchain system. Furthermore, the encrypted signature result generated by the elliptic curve is represented by a pair of values... If we represent it in this way, then the ECDSA algorithm can be expressed as:

[0113] ;

[0114] ;

[0115] in, It represents Finite field This represents the point where the public key is generated. It is a prime number, and its value ranges from 1 to 2. , This indicates the modulo operation. It is less than An integer that defines the base point. multiples Cyclic period on the curve. This represents any parameter greater than 0. It represents the point at infinity.

[0116] For the cryptographic signature result It is the calculation point. of The coordinates are obtained by taking the modulus, and the purpose is to bind the signature result to specific elliptic curve points, which can be represented as:

[0117] ;

[0118] in The range of values ​​is within ,Pick coordinates The values ​​are distinct and random numbers, ranging from [value range missing]. .

[0119] Similarly, the signature encryption result It is by Private key Message hash and The calculation shows that... Specifically, it is expressed as follows:

[0120] ;

[0121] in Represents integers Perform a modulo operation. The random generation of VK can be represented as:

[0122] ;

[0123] in Modulus The set of non-negative integers, i.e. .

[0124] Additionally, BK is a base point in ECDSA. If the points obtained by scalar multiplication with VK are used for generation, then the generation of BK can be expressed as:

[0125] ;

[0126] In summary, the signature generation scheme It can be represented as:

[0127] ;

[0128] in, Indicates the encryption result. This indicates that the vehicle uses a digital signature with a private key. This represents the private key. It is a hash digest of the signature information.

[0129] In a consortium blockchain system, asymmetric encryption algorithms are used... and The signature verification process, which uses pairing to verify whether the information has been tampered with, can be represented as follows:

[0130] ;

[0131] in, This indicates that the vehicle user uses a public key for authentication. This indicates that the recipient has verified the signature information.

[0132] Since this invention involves hashing and encryption of multiple PVs and multiple SVs, the hash calculation is considered. Encrypted computing resources Consumption of communication resources They can be represented as:

[0133] ;

[0134] ;

[0135] ;

[0136] in, express The size of the data being processed This indicates the time consumed in each hash calculation; This indicates the computational cost of the ECDSA process. This indicates the time consumed in each signature calculation; Indicates the size of the signature data. This represents the amount of data transmitted per unit of spectrum bandwidth. Therefore, the total resource consumption can be calculated as follows: :

[0137] ;

[0138] Furthermore, obtaining the optimal solution includes: excluding licensed vehicles that do not meet the spectrum demand requirements, and selecting licensed vehicles that meet the signal-to-interference-plus-noise ratio (SINR) threshold from the licensed vehicles that meet the requirements; excluding cognitive vehicles that do not meet the spectrum availability requirements, and selecting cognitive vehicles that meet the SINR threshold; obtaining the optimal bandwidth and optimal bid of the licensed vehicles and cognitive vehicles, comparing them with each other to obtain a bidding set; and sending auction requests for the licensed vehicles and cognitive vehicles respectively based on the bidding set.

[0139] When a base station node receives auction requests from authorized vehicles for each cognitive vehicle, it decrypts and verifies the digital signature using the private keys of the authorized and cognitive vehicles, and performs hash decryption of the information. If the information matches completely, it proves that the information has not been tampered with, and the optimal solution is generated.

[0140] Specifically, in addition, during the spectrum demand matching phase, PV and SV will first provide their respective available spectrum on a spectrum trading platform based on a consortium blockchain. and spectrum demand And related content such as digital signatures. The node will adjust its frequency based on SV's spectrum requirements. Unsuitable PVs are eliminated, and then the PV with the best signal-to-interference-plus-noise ratio (SNR) is selected from the compliant PVs. The same process applies to PVs; unsuitable SVs are eliminated based on the SVs provided by the nodes, and the SVs that meet the SNR are selected. The base station nodes determine the optimal bandwidth for PVs and SVs based on the bidding mechanism of the auction algorithm. and best bid ,pass and By comparing the bids of PVs against SVs for spectrum resources, the optimal solution can be obtained. The PV adds its own digital signature to prevent information tampering. The PV filters out SVs that meet its requirements, and then, considering factors such as the optimal bid for the resource and the signal-to-interference-plus-noise ratio, selects the most satisfactory SV and sends an auction request to it. Therefore, the PV's auction request for SVs can be expressed as:

[0141] ;

[0142] in, It is an auction request message. This refers to the digital signature of the PV's private key; Indicates the first The SV can meet the signal-to-interference-plus-noise ratio of the PV, meaning that the SV does not interfere with the normal communication of the PV. The timestamp when the information was uploaded; This represents the set of possible optimal bids for PVs.

[0143] Then for all PVs Specifically, it can be expressed as:

[0144] ;

[0145] Meanwhile, the alliance base station nodes select eligible PVs based on the SV's spectrum requirements and send auction request information. The content of the sent information is encrypted using the SV's private key to prevent malicious attacks and tampering. Therefore, the SV's auction request for PVs can be represented as:

[0146] ;

[0147] in, It is an auction request message. This refers to the digital signature of the SV's private key; Indicates the first The PV vehicle can meet the signal-to-interference-plus-noise ratio (SIR) of the SV vehicle. The timestamp when the information was uploaded; This represents the set of possible optimal bids for SVs.

[0148] Then for all SVs Specifically, it can be expressed as:

[0149] ;

[0150] After the base station node receives the auction request information for PVs and SVs, it then... and The process involves decryption and digital signature verification, as well as hash decryption of the information. If there is a complete match, it proves that the information has not been tampered with. In particular, the base station generates the optimal matching scheme for PVs and SVs according to the auction bidding rules, and then feeds back the matching result to the corresponding PVs and SVs. The feedback information from the BS can be expressed as follows:

[0151] ;

[0152] in, It uses the base station node's private key for encryption. It's a timestamp. It represents the matching result of PVs and SVs for a transaction amount that satisfies both parties.

[0153] The combination of the properties of consortium blockchains and encryption algorithms enhances system security and reliability, providing a trusted spectrum trading environment for PV and SV participants in spectrum sharing. Its key feature is allowing participants to control access to data, meaning sensitive data can be encrypted or accessed only by authorized participants, ensuring the authenticity and immutability of transactions while verifying the sender's identity, thus improving data privacy and security. Considering the characteristics of pre-selected node management in consortium blockchains, and to achieve consensus while conserving computing resources, this invention employs a practical Byzantine Fault-Tolerant (PBFT) consensus algorithm. When consortium blockchain nodes reach an agreement, the blockchain achieves consensus, and both parties successfully complete the spectrum transaction, ensuring the trustworthiness and security of the transaction. The specific process of uploading spectrum transaction information to the blockchain is as follows: Figure 2 As shown.

[0154] Furthermore, considering the specific scenario of cognitive vehicle-to-everything (V2X) communication networks under the dynamic topology of rapidly moving vehicle equipment, assuming there are A PVs and B SVs within a time slot t, based on the spectrum resource requirements and the limitation of interference between vehicle equipment, SVs can purchase idle spectrum resources from PVs. Through the constraints of spectrum resource demand and interference limitation, utility function formulas for PVs and SVs are given respectively, to satisfy the satisfaction of PVs and SVs with the dynamic spectrum process.

[0155] Furthermore, obtaining utility satisfaction and throughput includes: the utility satisfaction of the authorized vehicle when the authorized vehicle occupies the channel, the throughput of the authorized vehicle without affecting the authorized vehicle, and the throughput of the cognitive vehicle without affecting the authorized vehicle; the utility satisfaction of the authorized vehicle and the utility satisfaction of the cognitive vehicle when the authorized vehicle does not occupy the channel.

[0156] Specifically, this invention considers two scenarios where PV occupies the channel and the impact of noise interference from PV and SV on the spectrum matching process.

[0157] The first scenario involves PV occupying the channel. A utility satisfaction formula for PV and SV is designed, which includes the signal-to-interference-plus-noise ratio (SINR), the idle spectrum resources of PV, the required spectrum resources of SV, and their respective pricing for their respective resources. The utility satisfaction formula for PV is... It can be represented as:

[0158] ;

[0159] in, It is the first match SV vehicle against the first The signal-to-interference-plus-noise ratio (SIR) generated by a PV vehicle represents the degree of interference from the SV to the PV. Indicates the first The pricing of idle spectrum resources by a vehicle PV. This represents the amount of free spectrum resources that PVs can share. and The weight parameters used for PV calibration, and For the first SV vehicle against the first Formula for calculating the signal-to-interference-plus-noise ratio generated by a PV vehicle It can be represented as:

[0160] ;

[0161] in, It is the first The transmission power of the PV vehicle It is the first Channel gain of PV Expressed as noise power density, Indicates the first The channel bandwidth occupied by the PV vehicle This represents the total power of the background noise. This indicates the interference of SV to PV. Regarding the channel gain of PV... Represented as:

[0162] ;

[0163] in, This represents the complex gain of the Rice channel.

[0164] The designed cognitive vehicle-to-everything (V2X) environment based on consortium blockchain uses the Ricean channel model to calculate the channel gain. The Ricean channel complex gain formula is as follows: It can be represented as:

[0165] ;

[0166] in:

[0167] ;

[0168] ;

[0169] in, The shape parameter representing the Ricean channel determines the Ricean distribution. and Decide, The signal amplitude represents the main direct path. This represents the total signal standard deviation, including signals from the direct path and those affected by multipath effects. This indicates that the mean is zero and the variance is . Complex Gaussian noise. Indicates path loss. It is the transmission distance between vehicles. It is the frequency of the spectral signal.

[0170] Then path loss It can be represented as:

[0171] ;

[0172] in , , The parameter settings represent path loss. Distance can be represented as:

[0173] ;

[0174] , They represent the first PV and the first Location information of vehicle SV.

[0175] Similarly, for the interference formula of SVs on PV The consideration can be expressed as:

[0176] ;

[0177] in, For SV users' transmit power, The channel gain is expressed as SV. The subsequent considerations for channel state and path loss are basically the same as for PV. The considerations for SV and the second case also follow the same approach.

[0178] When considering the signal-to-interference-plus-noise ratio (SINR) during communication, the PV throughput can also be calculated using the channel capacity formula without affecting the PV. The throughput formula is... It can be represented as:

[0179] ;

[0180] in, This represents the bandwidth occupied by the channel. This indicates the signal-to-interference-plus-noise ratio (SIR) when SV occupies the channel.

[0181] Furthermore, in this model, utility-based function design needs to satisfy not only the communication needs of the PV, but also the needs satisfaction of the SV; therefore, the formula for SV is... It can be represented as:

[0182] ;

[0183] in It is the first match PV to the first The signal-to-interference-plus-noise ratio generated by the SV vehicle. Indicates the first The pricing of spectrum resources required by SV vehicles. Indicates the first The amount of free spectrum resources required by a single SV. and The weight parameters used for SV calibration, and For the first PV to the first Formula for signal-to-interference-plus-noise ratio generated by SV It can be represented as:

[0184] ;

[0185] in, It is the first The transmit power of the SV vehicle It is the first The channel gain of the vehicle SV Expressed as noise power density, Indicates the first The channel bandwidth occupied by the SV vehicle. This represents the total power of the background noise. Indicates the first The interference of a vehicle's PV to its SV; then the interference formula. It can be represented as:

[0186] ;

[0187] in, For PV users' transmit power, Indicates channel gain.

[0188] Similarly, the SV throughput is calculated using the channel capacity formula without affecting the PV. It can be represented as:

[0189] ;

[0190] in, This is expressed as the bandwidth of the channel. This represents the signal-to-interference-plus-noise ratio (SIR) generated by PV on SV when SV occupies the channel.

[0191] The second scenario involves the PV not occupying the channel. Similarly, the idle and required spectrum resources of the PV and SV, along with their respective pricing for their resources, are expressed as utility formulas for the PV and SV. Since the PV is not occupying the channel at this time, its spectrum resources are idle, so only the interference problem of other SVs competing for idle spectrum resources needs to be considered.

[0192] Then the PV calibration formula can be considered. Represented as:

[0193] ;

[0194] in, Satisfactory parameters for PV calibration, and .

[0195] Similarly, based on the first case, the present invention considers the SV calibration formula. Represented as:

[0196] ;

[0197] in, Is it other SVs against the first The signal-to-interference-plus-noise ratio of the vehicle occupying channel SV. and The weight parameters used for SV calibration, and , For other SV pairs Signal-to-interference-plus-noise ratio of vehicles occupying channel SV It can be represented as:

[0198] ;

[0199] in, It is the first The transmit power of the SV vehicle It is the first The channel gain of the vehicle SV Expressed as noise power density, Indicates the first The channel bandwidth occupied by the SV vehicle. This represents the total power of the background noise. This indicates interference from other SVs. It can be represented as:

[0200] ;

[0201] in, That is the transmit power of other SVs. Indicates channel gain.

[0202] Similarly, the throughput of SV can be calculated using the channel capacity formula when PV does not occupy the channel. It can be represented as:

[0203] ;

[0204] in, This indicates the bandwidth occupied by the channel. This indicates that when PV does not occupy the channel, other SVs affect the first PV. The signal-to-interference-plus-noise ratio of the vehicle occupying channel SV.

[0205] To maximize the satisfaction of the algorithm proposed in this invention With system throughput This paper proposes a secure spectrum auction and sharing method based on IDQN blockchain in cognitive vehicle-to-everything (V2X) systems. During spectrum allocation, base stations first initialize using hash encryption and ESDSA. To maximize the effectiveness of PV and SV while minimizing resource consumption and ensuring maximum user satisfaction, the paper addresses these issues. The entire hashing and encryption process was taken into account. It also provides constraints on time and resource consumption.

[0206] Optimized formula for maximizing satisfaction It can be represented as:

[0207] ;

[0208] :

[0209] ;

[0210] ;

[0211] ;

[0212] ;

[0213] in, This refers to the overall social welfare of the auction algorithm. This refers to the utility formula for PV users in the process of dynamic spectrum sharing. This refers to the utility formula for SV users during the dynamic spectrum sharing process. (Constraints) middle This means that the amount of spectrum resources occupied cannot exceed the amount of spectrum resources available. This represents the channel with the least amount of available free spectrum resources. This represents the channel with the largest amount of available free spectrum resources. (Constraint) Indicates the spectrum resource requirements of SV. Cannot exceed the matched PV idle spectrum resources .constraint Represents hash Computation time of ECDSA To Within the range. Constraints This indicates that the time required for all SV matching to complete must be within one... Inside.

[0214] Since this invention addresses the problem of finding the optimal strategy through multi-objective joint optimization, and combines the interference limitation of SV on PV, it provides the system throughput. The representation is:

[0215] ;

[0216] :

[0217] ;

[0218] ;

[0219] ;

[0220] ;

[0221] in, This refers to the total throughput of the system. Indicates the first Channel throughput of PV users Indicates the first The channel throughput of the vehicle SV. Constraints The text indicates that the SV's transmit power has been limited. Cannot exceed the maximum threshold .constraint This indicates the SINR threshold set for the PV, ensuring that any interference caused by it will not affect the normal communication of the PV. (Constraint) This means that the SINR caused by PV to the channel must also meet the actual requirements of SV. (Constraint) This indicates whether the SV is occupying an idle channel of the matching PV. =1 indicates the first The SV successfully occupied the first matched vehicle. PV idle channel, =0 indicates that the attempt to use the device was unsuccessful.

[0222] Furthermore, Markov decision process problems include: state space, action space, reward function, and spectrum allocation strategy.

[0223] Specifically, this invention designs the optimization objective of jointly satisfying the system's performance and throughput as a continuous-state Markov Decision Process (MDP) problem, which mainly consists of a state space. Action space Reward function and spectrum allocation strategies The design of the reward function is mainly based on the objective of the multi-objective joint optimization proposed in this invention.

[0224] During the communication process between vehicle users Within a time slot, in the cognitive vehicle network environment, the SV will occupy the idle spectrum of the PV that has been successfully matched with it, and it will determine whether the SV has successfully occupied the spectrum. The SV and PV can continuously improve and optimize according to the strategy of the deep reinforcement learning algorithm, select the optimal strategy, and finally complete the cognitive vehicle network dynamic spectrum sharing process. The composition design of the MDP is as follows.

[0225] (1) State space The establishment

[0226] For cognitive vehicle-to-everything (V2X) dynamic spectrum resource sharing, PV and SV need to interact and trade spectrum resources based on their respective resource needs within a consortium blockchain architecture, taking into account various real-world scenarios to generate their state space. . Specifically, it can be expressed as:

[0227] ;

[0228] state It is a union group containing multiple elements, including the states of PVs. SV's state and the price of spectrum resources to be considered. Signal-to-interference-plus-noise ratio in actual communication environments .

[0229] (2) Action space The establishment

[0230] When performing spectrum matching and occupying an idle channel, whether the SV (Single Stream) occupies the idle channel and successfully achieves communication is crucial. The SV first observes the generated state space and selects its action according to a strategy. The main decision is whether the SV successfully occupies the idle channel of the successfully matched PV (Performance Module). Therefore, the action space... It can be represented as:

[0231] ;

[0232] in, This indicates that SV has chosen to occupy the matching PV idle channel. This indicates that SV has selected an unused PV free channel that is not currently in use.

[0233] (3) State transition probability

[0234] In complex vehicle-to-everything (V2X) communication environments, state transition probabilities are determined in different state spaces. Different actions should be chosen based on different strategies. The long-term expected reward of the resulting future returns directly affects the spectrum resource allocation decision, thus influencing the probability of the system transitioning to the next state. It can be represented as:

[0235] ;

[0236] in, This indicates that after taking an action in the current state, the state transitions to the next state. The probability of.

[0237] (4) Reward function Design

[0238] In Markov decision-making, the ultimate goal is to maximize the long-term cumulative reward. To ensure the timeliness and fairness of subsequent spectrum matching and sharing, this invention designs a ranking mechanism for tracking historical dishonest behavior using a joint reward function. The aim is to influence the priority of PV and SV in subsequent rankings based on the number of defaults in their historical records. The PV ranking function can be expressed as:

[0239] ;

[0240] ;

[0241] in, This indicates the sorting function based on PV in time. The ranking This indicates that, according to the utility function of PV, The initial descending order of time. The number of times PV engaged in dishonest behavior. As a constant coefficient, when When this occurs, it indicates that there has been no dishonest behavior. When it is a certain number of times dishonest behavior has occurred.

[0242] Similarly, the SV sorting function can be expressed as:

[0243] ;

[0244] ;

[0245] in, This indicates that the sorting function is based on SV. Ranking of moments This indicates that the utility function based on SV is in The initial descending order of time. The number of times SV engaged in dishonest behavior. It is a constant coefficient, consistent with the constant coefficient of PV, when When this occurs, it indicates that there has been no dishonest behavior. When PVs and SVs are in a certain number, it indicates a specific number of dishonest acts. The order of PVs and SVs is as follows: Figure 3 As shown.

[0246] Both PV and SV receive the sorting formula and, after establishing their respective descending sequence lists, communicate with each other. The base station broadcasts the matching results to both PV and SV. The broadcasted matching content... It can be represented as:

[0247] ;

[0248] in, After establishing sorted lists for SVs and PVs, the base station The content of communication and broadcasting The matching results of spectrum resources and historical transaction information after both parties have established a sorting list.

[0249] After spectrum matching is completed, PV and SV will continuously monitor each other for dishonest behavior through a designed smart contract. If either party engages in dishonest behavior, the vehicle will record it locally and upload it to the consortium blockchain to form an immutable transaction record. The dishonest behavior of PV and SV will accumulate, thus serving as a penalty to lower their ranking in the next round of ranking. and The number of dishonest acts will be combined with the reward function in the proposed algorithm, which will help create an honest and fair spectrum trading environment. It should be noted that dishonest behavior by PVs includes sending false channel occupancy information and failing to allow SVs to use idle channels after a match is completed. Dishonest behavior by SVs includes occupying channels but not using them and refusing to pay currency after using them.

[0250] Furthermore, to meet the requirements of multi-objective joint optimization, improve the satisfaction of both PV and SV and the system throughput, the reward function will be designed to combine these two objectives and be expressed in a weighted manner:

[0251] ;

[0252] ;

[0253] ;

[0254] in, express This refers to the reward setting for the utility function. express It is a reward setting for system throughput. This indicates whether the SV has successfully occupied the idle channel of the matched PV. Only then will rewards be considered, if Not considering cumulative rewards, in addition, the following were also considered: and Regarding the impact of the reward function on the two scenarios, the rewards for PV and SV decrease as dishonest behavior increases. This indicates the weight settings for the optimization objective. .

[0255] (5) Strategy optimization:

[0256] For a multi-objective optimization design that comprehensively considers the utility satisfaction and system throughput of PVs and SVs in spectrum resource sharing matching, this invention considers the long-term cumulative rewards obtained by PVs and SVs in spectrum resource sharing. The reward function incorporates a weighted joint design of utility satisfaction and system throughput. The external environment parameter design of PVs and SVs in cognitive vehicle-to-everything (V2X) communication is simulated using a Python platform, and the optimal Q-value is sought based on the reward function. That is, in different state spaces... Different actions should be chosen based on different strategies. The present invention will Defined as a strategy to achieve spectrum sharing, given an initial state. and spectrum sharing strategy To find the optimal action, the expected long-term cumulative discount reward is... It can be represented as:

[0257] ;

[0258] in, These are discount parameters, and the range is... .

[0259] The ultimate goal is to find the optimal strategy for spectrum sharing in a cognitive vehicle-to-everything (V2X) environment, that is, to find the optimal action to maximize the global benefits of both PVs and SVs. Therefore, the set of all strategies is set as follows: , Then the optimal strategy It can be represented as:

[0260] ;

[0261] in, The goal is to find the optimal strategy for spectrum matching and sharing, taking into account factors such as the total amount of spectrum resources required and the unified pricing of the occupied idle spectrum resources, based on the designed PV and SV actions and state spaces.

[0262] Furthermore, obtaining the optimal action involves iterating through the state space input into the IDQN learning model, selecting the corresponding action in the action space according to different spectrum allocation strategies, until the optimal action is obtained.

[0263] The optimal action is found through the design of an optimal strategy to achieve efficient spectrum resource sharing between PV and SV. In IDQN, the optimal action refers to the action taken in a given state. In the improved DQN output, the action corresponding to the maximum Q value is the optimal spectrum allocation decision that the PV and SV should make after learning the optimal policy. PV and SV can use the learning of the optimal Q value to solve the problems of spectrum matching and channel selection. However, in a cognitive vehicular network environment with multiple PVs and multiple SVs, the state space or action space becomes very large and needs to maintain continuous states, making the traditional Q-learning table method infeasible and inefficient for finding the optimal policy. Therefore, this invention uses a deep neural network instead of Q-learning for the Q value.

[0264] First, let's discuss the loss function for the improved algorithm. Design:

[0265] ;

[0266] in Indicates the state Next action The reward It is a discount factor. These are the parameters of the target Q-network. Used to calculate the error between the target Q value and the current Q value.

[0267] At the same time, a greedy algorithm is used to control the balance between exploration and exploitation, selecting the optimal action with a certain probability. Specifically, it can be expressed as:

[0268] ;

[0269] in This indicates an action selected based on random probability. .

[0270] In summary, the iterative formula for the Q-value of the improved algorithm can be expressed as:

[0271] ;

[0272] The updated optimal policy function can be expressed as: :

[0273] ;

[0274] These functions together constitute the core mechanism for solving the optimal policy in IDQN, among which, For learning rate, , refers to state The value lies in choosing the optimal action in the current state. The strategy that yields the maximum return also takes into account the long-term discounted returns in the system.

[0275] In IDQN learning, the neural network consists of an input layer, hidden layers, and an output layer. Generally, the state space is considered as the input to the neural network. First, the discrete input data is reduced in dimensionality using embedding methods to decrease the computational burden and improve learning efficiency. Then, batch data sampling is performed using a replay buffer, followed by concatenation before being input into the IDQN neural network. After nonlinear processing using the ReLU activation function, the Q-value for each action is finally output from the output layer. This allows the neural network to approximate the Q-values ​​of the traditional Q-table, thus replacing the traditional Q-table. The flowchart of the neural network is as follows: Figure 4 As shown:

[0276] In summary, the IDQN algorithm proposed in this invention can solve the dynamic spectrum sharing problem in cognitive vehicle-to-everything (V2X) communication scenarios with multiple PVs and multiple SVs by designing the utility function and solving the optimal strategy.

[0277] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A cognitive vehicle-to-everything (V2X) security spectrum auction and sharing method based on IDQN blockchain, characterized in that, include: To build a blockchain-based vehicle network, during the spectrum demand matching process of the vehicle network, the available spectrum and spectrum demand of authorized vehicles and cognitive vehicles are obtained. The available spectrum and spectrum demand are filtered by conditions, and the optimal solution is obtained by combining the optimal bandwidth and the optimal bid. Obtaining the optimal solution includes: Exclude licensed vehicles that do not meet the spectrum demand requirements, and select licensed vehicles that meet the signal-to-interference-plus-noise ratio (SINR) threshold from the licensed vehicles that meet the requirements. Exclude cognitive vehicles that do not meet the spectrum availability requirements, and select cognitive vehicles that meet the SINR threshold. Obtain the optimal bandwidth and optimal bid of the authorized vehicle and the cognitive vehicle, compare them, obtain a bidding set, and based on the bidding set, send auction requests for the authorized vehicle and the cognitive vehicle respectively: Sending the auction request for the authorized vehicle includes: ； in, For auction request information, A digital signature for the private key of the authorized vehicle PV; For the first The cognitive vehicle (SV) can meet the signal-to-interference-plus-noise ratio (SIR) of the licensed vehicle (PV), meaning that the cognitive vehicle does not interfere with the normal communication of the licensed vehicle (PV). The timestamp when the information was uploaded. This represents the set of possible optimal bids for the authorized vehicle set PVs; Sending auction requests for each of the cognitive vehicles includes: ； in, For auction request information, Digital signature for the private key of the cognitive vehicle SV. For the first The PV of the authorized vehicle can meet the signal-to-interference-plus-noise ratio of the cognitive vehicle SV. This represents the set of possible optimal bids for the cognitive vehicle set SVs; When the base station node receives the auction requests from the authorized vehicle to the cognitive vehicle, it performs decryption and digital signature verification using the private keys of the authorized vehicle and the cognitive vehicle, and performs hash decryption of the information. If they match completely, it proves that the information has not been tampered with, and the optimal solution is generated. In the process of spectrum matching of the optimal solution, the utility satisfaction and throughput are obtained. An optimization objective is established based on the utility satisfaction and throughput. The optimization objective is used as a decision-making process. The decision-making process is solved using the IDQN learning model through the state space, action space, reward function and spectrum allocation strategy to obtain the optimal action, i.e. the optimal spectrum allocation decision. Obtaining the optimal action includes: The state space is input into the IDQN learning model for iteration, and the corresponding action in the action space is selected according to different spectrum allocation strategies until the optimal action is obtained: ； in, This is the optimal strategy for spectrum matching and sharing. In the state The optimal action found below The optimal Q value; The action iteration of the IDQN learning model is represented as follows: ； in, Indicates the state Next action The reward A discount factor to measure the degree of future returns. Use neural network parameters for the target network The obtained Q-value estimate for the action in the next state, Use neural network parameters for the current network The obtained Q-value estimate for the action in the current state.

2. The cognitive vehicle-to-everything (V2X) security spectrum auction and sharing method based on IDQN blockchain according to claim 1, characterized in that, Building the aforementioned vehicle network includes: A vehicle network based on a consortium blockchain is formed by a number of authorized vehicles and cognitive vehicles, wherein the authorized vehicles and cognitive vehicles in the vehicle network are located within the area covered by the consortium blockchain composed of multiple base stations. Before the spectrum demand matching stage of the vehicle network, the authorized vehicle and the cognitive vehicle are encrypted and initialized, and a public key and a private key pair are generated through the vehicle user's account address.

3. The cognitive vehicle-to-everything (V2X) security spectrum auction and sharing method based on IDQN blockchain according to claim 1, characterized in that, Obtaining the utility satisfaction and the throughput includes: When the authorized vehicle occupies the channel, the utility satisfaction of the authorized vehicle includes: ； in, For the first match Cognitive Vehicle SV for the first The signal-to-interference-plus-noise ratio (SIR) generated by the cognitive vehicle's SV represents the degree of interference between the cognitive vehicle's SV and the licensed vehicle's PV. Indicates the first The pricing of idle spectrum resources for each licensed vehicle (PV). This indicates the amount of free spectrum resources that authorized vehicles (PVs) can share. and Weight parameters used for PV calibration of authorized vehicles; The throughput of the authorized vehicle for the cognitive vehicle without affecting the authorized vehicle includes: ； in, This represents the bandwidth occupied by the channel; The perceived vehicle's utility satisfaction includes: ； in, For the first match PV to the first Signal-to-interference-plus-noise ratio (SIR) generated by a cognitive vehicle SV. Indicates the first Pricing of spectrum resources required by a cognitive vehicle (SV). Indicates the first The amount of free spectrum resources required by a single cognitive vehicle (SV). and Weighting parameters used for vehicle SV calibration; The throughput of the cognitive vehicle without affecting the authorized vehicle includes: ； in, This is expressed as the bandwidth of the channel.

4. The cognitive vehicle-to-everything (V2X) security spectrum auction and sharing method based on IDQN blockchain according to claim 3, characterized in that, Obtaining the utility satisfaction and the throughput also includes: When the authorized vehicle does not occupy the channel, the utility satisfaction of the authorized vehicle includes: ； in, Satisfactory parameters for PV calibration of authorized vehicles; The perceived vehicle's utility satisfaction includes: ； in, Is it other cognitive vehicle SV to the first The signal-to-interference-plus-noise ratio (SIR) of a vehicle occupying a channel for cognitive vehicle SV. and Weighting parameters used for vehicle SV calibration.

5. The cognitive vehicle-to-everything (V2X) security spectrum auction and sharing method based on IDQN blockchain according to claim 1, characterized in that, The optimization objectives include: Optimization of utility satisfaction includes: ； ： ； ； ； ； in, For the overall social welfare of auction algorithms, The utility formula for authorized vehicle users in the process of dynamic spectrum sharing is expressed as follows. To understand the utility formula for vehicle users in the process of dynamic spectrum sharing, This means that the amount of spectrum resources occupied cannot exceed the amount of spectrum resources available. This represents the channel with the least amount of available free spectrum resources. This indicates the channel with the most available free spectrum resources. To understand the spectrum requirements of vehicle SV. For the free spectrum resources of authorized vehicles, For the preset threshold time, Represents a hash. For ECDSA computation time, This is the maximum threshold for ECDSA computation time; Optimizing system throughput includes: ； ； ； ； ； in, This represents the total throughput of the system. Indicates the first Channel throughput of authorized vehicle users Indicates the first Channel throughput of a cognitive vehicle (SV). This indicates the transmit power of the cognitive vehicle SV. This indicates the maximum threshold for transmission power. =1 indicates the first The cognitive vehicle SV successfully occupied the matched first Authorized vehicle PV idle channel, =0 indicates that the attempt to use the device was unsuccessful. , The minimum and maximum thresholds for signal-to-interference-plus-noise ratio (SIR) are given. The total number of authorized vehicle PVs participating in spectrum sharing. The total number of cognitive vehicle SVs participating in spectrum sharing.

6. The cognitive vehicle-to-everything (V2X) security spectrum auction and sharing method based on IDQN blockchain according to claim 1, characterized in that, The state space is as follows: ； in, For a union group containing multiple elements, The status of the authorized vehicle PV. To recognize the status of the vehicle's SV. For spectrum resource prices, This refers to the signal-to-interference-plus-noise ratio (SINR) in a real-world communication environment. The action space is: ； in, This indicates that the cognitive vehicle (SV) has selected to occupy the matching PV idle channel. This indicates that the cognitive vehicle SV selects an unused PV idle channel; The reward function is: The priority of subsequent sorting is determined based on the number of defaults in the history of the authorized vehicle and the recognized vehicle: The sorting function for the authorized vehicles is: ； ； in, This indicates the time based on the sorting function of authorized vehicles. The ranking This indicates that, based on the utility function of the authorized vehicle, The initial descending order of time. The number of times a vehicle has engaged in dishonest conduct. As a constant coefficient, when When, it indicates no dishonest behavior; when When this occurs, it indicates dishonest behavior; The cognitive vehicle ranking function is: ； ； in, This indicates that the sorting function based on the cognitive vehicle SV is in Ranking of moments This indicates that the utility function of the cognitive vehicle SV is in The initial descending order of time. To determine the frequency of dishonest behavior by the vehicle SV, It is a constant coefficient, consistent with the constant coefficient of PV, when When, it indicates no dishonest behavior; when When this occurs, it indicates dishonest behavior; Based on the aforementioned utility satisfaction and throughput, and considering dishonest behavior in the vehicle ranking function, a reward function is constructed and expressed using weighted forms: ； ； ； in, This represents the reward setting for the utility function. , Indicates the reward setting for system throughput. , This indicates whether the SV has successfully occupied the idle channel of the successfully matched PV. This indicates the weight settings for the optimization objective; The spectrum allocation strategy is as follows: ； in, For the desired long-term cumulative discount rewards, This is the discount parameter.

Images