A vehicular internet of things trusted data sharing method and system based on adaptive fragmentation and reputation mechanism

Abstract

The application discloses a kind of based on adaptive fragmentation and reputation mechanism's Internet of Vehicles trusted data sharing method and system, belong to communication technology and cross field of blockchain.It is three-layer physical architecture that the system adopts terminal vehicle layer, RSU edge layer and cloud service layer constitutes, and combines the logical double-chain structure of consensus chain and storage chain.This method is reconstructed dynamically and load balancing by spatiotemporal perception adaptive fragmentation mechanism, in combination with geographical location, node load and reputation information;Through distance, time and reputation triple attenuation factor and Bayesian fusion method, the authenticity of traffic event is dynamically evaluated;Adopt double-dimensional fusion reputation mechanism, decouple node single real-time performance and long-term behavior consistency, improve the robustness and anti-evil ability of reputation evaluation;And through Epoch-PoSH hybrid consensus mechanism, while improving system throughput and security, enhance the ability to resist long-range attack.The application is processed by on-chain and off-chain cooperation, effectively solve the single point failure in Internet of Vehicles data sharing, high concurrent processing capacity is insufficient and malicious data interference problem, realize data trusted, traceable, and give consideration to system performance, security and scalability.

Description

Technical Field

[0001] This invention relates to the intersection of communication technology and blockchain, specifically to a trusted data sharing method and system for vehicle networks based on adaptive sharding and reputation mechanisms. Background Technology

[0002] With the development of intelligent connected vehicles, the Internet of Vehicles (IoV) is gradually becoming an important infrastructure for improving traffic efficiency and ensuring traffic safety. Collaborative communication between vehicles, roadside units, and cloud platforms enables functions such as traffic incident sharing, road condition monitoring, and intelligent traffic scheduling, providing crucial support for the construction of intelligent transportation systems.

[0003] However, existing vehicle network data sharing relies heavily on centralized cloud platforms, which have problems such as high risk of single point of failure, insufficient high-concurrency processing capabilities, and difficulty in ensuring data credibility. This can easily lead to system interruptions, delayed event reporting, or malicious data affecting traffic decisions.

[0004] While blockchain technology has advantages such as decentralization, immutability, and traceability, it still faces challenges in the context of connected vehicles, including limited system throughput, high computational overhead of consensus mechanisms, high transaction confirmation delays, and insufficient ability to identify malicious nodes, making it difficult to meet the needs of real-time and highly reliable data sharing.

[0005] While existing sharding technologies can improve blockchain processing capabilities, they mostly only consider network topology or the number of nodes, lacking comprehensive optimization for dynamic changes in geographical location and traffic load. At the same time, existing reputation assessment mechanisms are inefficient in updating in dynamic environments and struggle to continuously identify malicious nodes.

[0006] Therefore, how to achieve an efficient, reliable, and scalable data sharing mechanism in the vehicle-to-everything (V2X) environment while ensuring system security has become a key technical problem that urgently needs to be solved. Summary of the Invention

[0007] To address the aforementioned issues, this invention provides a trusted data sharing method and system for vehicle-to-everything (V2X) networks based on adaptive sharding and reputation mechanisms. By employing a spatiotemporal-aware sharding strategy, a dynamic reputation evaluation model, and a hybrid consensus mechanism, the security, throughput, and scalability of the V2X data sharing system are achieved. Attached Figure Description

[0008] Figure 1 This is a system overall framework diagram provided for embodiments of the present invention. The diagram illustrates the system's three-layer physical architecture: the terminal vehicle layer, the edge layer with RSU as the geographical anchor point, and the cloud layer responsible for global governance; the core functional modules include an adaptive sharding strategy, a Bayesian dynamic reputation model, and an Epoch-PoSH hybrid consensus mechanism; and it also shows a logical dual-chain structure, including a consensus chain and a storage chain.

[0009] Figure 2 This diagram illustrates a multi-layered blockchain block structure provided in an embodiment of the present invention. The diagram shows the ledger construction based on the Epoch-Slot timeframe, highlighting the structural differences and field composition between the Head Block generated by the trusted administrator and the Common Block generated through competition between ordinary nodes.

[0010] Figure 3 This diagram illustrates the event reporting process provided in this embodiment of the invention. It depicts four stages from vehicle perception of the environment to reaching a consensus: event identification (classification of ordinary and emergency events), inter-node interaction, RSU information aggregation, and credibility feedback based on a Bayesian model.

[0011] Figure 4 This diagram illustrates the principle of a long-range attack as provided in an embodiment of the present invention. It shows how an attacker attempts to construct an alternative historical private chain and rewrite the ledger in a dynamic vehicular network environment, thus demonstrating the necessity of introducing a head-block checkpoint mechanism in this invention. Detailed Implementation

[0012] The technical solution provided by this invention includes the following steps:

[0013] 1. System Architecture Construction

[0014] The system comprises a terminal vehicle layer, a roadside unit edge layer, and a cloud service layer. It adopts a dual-chain structure with on-chain and off-chain collaboration.

[0015] (1) The consensus chain is used to record event indexes, node reputations, and system status information;

[0016] (2) The storage chain is used to encrypt and store the original traffic data, and the storage pressure on the chain is reduced by RSU sharding management.

[0017] 2. Spatiotemporal Aware Adaptive Slicing

[0018] The sharding algorithm is run periodically, and a machine learning-driven sharding optimization mechanism is introduced. By learning the spatiotemporal characteristics of nodes and load changes, adaptive optimization and dynamic reconstruction of the sharding structure are achieved. To this end, a formal representation of RSU and a corresponding objective function need to be constructed to quantify geographical distribution and load status. Let's assume... A set consisting of online RSU nodes:

[0019]

[0020] in Represents the RSU node set, nodes Recorded as , respectively representing two-dimensional coordinates and average load ( Sharding aims to balance geographical aggregation and load balancing by dividing nodes into... Each segment (center is) If the traditional partitioning optimization can be formalized into the following objective function:

[0021]

[0022] In the formula Represents a node Euclidean distance from the partition center This represents the average load of the node. This represents the balance coefficient between the two. The objective function aims to balance node geographical proximity and shard load balancing to improve the processing efficiency and resource utilization of the vehicle-to-everything (V2X) blockchain.

[0023] Traditional clustering algorithms (such as K-Means) have significant limitations in this scenario: they only consider spatial proximity while ignoring load constraints, making it difficult to adapt to dynamic changes in RSU traffic and lacking adaptive capabilities, easily leading to uneven resource allocation and performance bottlenecks. Therefore, achieving sharding load balancing is crucial for improving data sharing efficiency. Thus, a sharding algorithm for connected vehicle blockchains, integrating geographical location, node load, and node reputation, is used, with its loss function defined as:

[0024]

[0025]

[0026] in, For this fragment Current total load and total reputation; For fragmentation Corresponding to the target value, To balance the weights; For average load and reputation within a shard, according to The bias adaptively adjusts the penalty intensity to ensure balanced load across partitions and smooth penalty application. The loss function consists of two terms:

[0027] 1. Distance loss: through Minimize the Euclidean distance between nodes and the centroid to ensure compactness of the fragmentation space and low-latency communication.

[0028] 2. Load and Reputation Penalty: Introducing an exponential term to guide shard load convergence. Furthermore, the reputation is evenly distributed. The algorithm is based on a greedy strategy; when the metric exceeds the limit, the loss value increases exponentially to suppress nodes from joining overloaded shards.

[0029] This algorithm aims to adaptively optimize the partitioning structure. Each iteration integrates spatial distance, load skew, and reputation distribution, employing a three-stage greedy search: initially focusing on geographical distance, then balancing load and reputation constraints in the mid-to-late stages until convergence. Compared to traditional methods, it is more adaptive and fair in dynamic traffic, suppressing overload, preventing reputation monopolies, and maintaining a balance between safety and efficiency. The system should be run periodically to handle RSU load fluctuations and maintain optimal performance.

[0030] To verify the performance of the data sharing method of this invention in the Internet of Vehicles, 383 RSUs were deployed as data sharing anchors on key road sections in the central urban area of ​​Zhengzhou. Random traffic flow was generated using SUMO and imported into the Veins platform to simulate a high-concurrency mobile network environment. At the same time, the aggregation parameter mal-chaos was introduced to simulate a malicious node collusion attack. Experiments compared the Proposed (Geo+Load+Reputation) method proposed in this invention with Geo+Load, Geo Only, and Random schemes. The results show that the Proposed method can maintain low-latency communication from the node to the centroid, with low and stable load variance, and significantly reduce the number of compromised shards in scenarios with high concentration of malicious nodes (approximately 40.75% and 56.16% lower than Geo+Load and Geo-only, respectively). In the intra-shard consensus performance test, a block capacity of 800 transactions can achieve a throughput of approximately 610 TPS while balancing latency. The adaptive reconstruction cycle experiment shows that when the cycle is 10 minutes, the system throughput reaches a peak of approximately 4778 TPS, with a load variance of approximately 1.49, proving that this invention has good load balancing capabilities and anti-collusion attack performance while maintaining low latency.

[0031] 3. Credibility assessment of traffic incidents

[0032] To quantify message credibility, a spatiotemporal decay mechanism is introduced: the closer the node is to the event and the newer the message, the higher its authenticity.

[0033] First, construct the distance attenuation factor. Assume the vehicle... With the event Manhattan distance is Then the distance factor can be expressed in an exponentially decaying form. in This is the distance attenuation coefficient, used to control the rate at which confidence decreases with increasing distance.

[0034] Secondly, a time decay factor is constructed. A timestamp t is introduced to mark the timeliness of the event, assuming the event occurred at time t. The vehicle reporting time is Then the time factor can be defined as: This is the event decay coefficient. It reflects the decrease in the value of information due to delay.

[0035] Finally, a reputation decay factor is constructed. A node's historical performance determines its basic credibility, but reputation is time-sensitive. Let node... Previous moment The reputation of The reporting time is Define the reputation factor after time decay. for:

[0036]

[0037] in This is the reputation decay rate, used to prevent nodes from exploiting old, high-reputation nodes for malicious purposes.

[0038] Taking into account the independent effects of the spatiotemporal dimensions, a product-based approach is used to jointly model the three attenuation factors. Substituting the distance factor, time factor, and reputation factor, the message credibility function is obtained as follows:

[0039]

[0040] After a vehicle or RSU receives feedback, the credibility score can be combined with a Bayesian formula to fuse event priors and node feedback, dynamically assessing the event's authenticity. Introducing node historical reputation, distance, and time factors allows for the integration of multi-source information, improving identification accuracy and system robustness while maintaining manageable computational complexity. Event credibility is represented as a posterior probability:

[0041]

[0042] in, and For complementary events, the prior probability , Updated dynamically based on historical records. Conditional probability is determined by vehicle feedback:

[0043]

[0044] In the formula, Rate the credibility of the message. This expression represents the node's attitude towards the event. The feedback is mapped to a probability space, where the probability of unaware feedback is 0.5, reflecting its neutral characteristic.

[0045] Final aggregation result Must meet Treat it as a real event. The threshold value is influenced by a combination of event urgency, node reputation, and system status, and is adaptively adjusted by the RSU.

[0046] 4. Node reputation update

[0047] A two-dimensional fusion reputation mechanism was constructed. This mechanism decouples and merges a node's single real-time performance with its long-term behavior through dynamic weighting, ensuring the adaptability and accurate, smooth evolution of reputation assessment.

[0048] To evaluate the immediate behavior of a node in a single event, define the node's behavior in the event. Comprehensive Trustworthy Behavior Score and the final judgment of the incident Based on the continuous consistency model, node performance is characterized by the deviation between its feedback and the credibility of events, and its consistency score is defined as follows:

[0049]

[0050] in This is the node's feedback value for the event (corrected for factors such as distance and time delay). , .

[0051] To achieve dynamic evolution of node reputation, the system relies on Update the node's historical reputation. (Set up the node...) At any moment Reputation value Then its reputation will be updated after the event is completed.

[0052]

[0053] in To adjust the balance coefficient between historical reputation and current behavior weights, This is the upper limit of creditworthiness. Indicates the node in the most recent time window Internal behavioral consistency is defined as

[0054]

[0055] in For nodes in the window The number of events that are internally participated in and correctly judged. The total number of judged events in which it has participated.

[0056] The overall reputation update consists of two parts: reputation inertia items. And reputation correction items The inertia term ensures a smooth and stable evolution of node reputation, preventing sudden drops or jumps in reputation due to occasional misjudgments or abnormal behavior, and suppressing speculative behavior. The correction term reflects the node's performance in the current event, taking into account the influence of historical behavior. This constrains the increase in reputation based on short-term good performance, while also measuring node contribution. To align the dimensions, this design balances reputation stability and real-time performance, enabling dynamic and robust assessment of node credibility.

[0057] To verify the performance of the Bayesian-based trusted event verification method of this invention in vehicle-to-everything (V2X) environments (implementations 3 and 4), local simulations were conducted in a typical segmented environment in the central urban area of ​​Zhengzhou. The interactive subnet included RSUs and dynamically accessed vehicles, with a total of N=100 nodes and a ratio of honest nodes to malicious nodes of 7:3. Experimental results show that after multiple interaction cycles, the reputation value of honest nodes monotonically increases and remains stable at its upper limit, while the reputation value of malicious nodes gradually decreases, with a final reputation value difference of approximately 75 points, verifying the effective suppression of malicious behavior by this mechanism. In the event recognition accuracy test, when the proportion of malicious nodes reaches 30%, the method of this invention still maintains a recognition accuracy of over 95%, which is approximately 9.2% and 28% higher than linear weighting and majority voting strategies, respectively. It can also effectively identify abnormal behavior where the physical location is inconsistent with the observed data, proving that this invention is efficient in identifying high-quality nodes in dynamic V2X environments and possesses high accuracy and robustness in complex attack environments.

[0058] 5. Hybrid consensus mechanism

[0059] A hybrid PoS consensus mechanism, Epoch-PoSH (PoS+HotStuff), is proposed for partially synchronized networks with low computational overhead. This mechanism employs a two-tiered Epoch-Slot timeframe and introduces a Trusted Manager committee elected based on reputation ranking. In each Epoch, the first Slot generates a final HeadBlock through HotStuff consensus, recording a state snapshot and key parameters. Subsequent Slots generate Common Blocks based on stake competition between RSUs or vehicle nodes, recording vehicle events, access control, and authorized transactions. These Common Blocks are bound to the HeadBlock via hashes and state digests, ensuring ledger continuity and immutability. The HeadBlock serves as a global checkpoint, enabling nodes to quickly reconnect to the main chain upon re-entry, enhancing resistance to long-range attacks and improving synchronization efficiency.

[0060] In hybrid PoS blockchains, vehicle-to-everything (V2X) nodes determine their block production eligibility through a coordinate-based verifiable random function (VRF). This mechanism introduces geographical location factors into the PoS stake selection process, making block competition subject to the randomness of stake and spatial distribution, thereby maintaining fairness and attack resistance in a dynamic network.

[0061] In each slot of each Epoch, nodes independently calculate their block production eligibility based on their own geographical coordinates and a globally generated random number. Let node... The real-time coordinates are The system publishes a global random number in the current Epoch. Then the node executes the coin toss function locally:

[0062]

[0063] in, This indicates the node's private key signing operation, used to ensure the verifiability and unpredictability of the source of randomness. This represents the number of bits in the output of the hash function. When a node meets the condition...

[0064]

[0065] At that time, it is considered that the block production qualification has been obtained. Among them, For the current rights and interests of the node, This represents the reputation value of the current node. The proportional coefficient is adjusted by the system. This makes the expected number of block producers This is to ensure that typically only one node in each slot successfully produces a block.

[0066] In IoV networks, RSU nodes have fixed locations and strong communication capabilities, while vehicle nodes have dynamic locations and limited computational and bandwidth capabilities. To prevent centralization and enhance decentralization, a location compensation factor is introduced into the block production probability. RSU has a fixed value. The vehicle's speed and mobility are dynamically calculated.

[0067]

[0068] in and It is a normalization function. It refers to weighting. This compensation mechanism can moderately increase the probability of block publication by mobile nodes while ensuring overall security, thereby compensating for the disadvantages caused by insufficient communication conditions.

[0069] When node After determining that it meets the block production conditions in the current slot, generate a new block:

[0070]

[0071] This information is then broadcast to the entire network. Other nodes can verify a block producer's signature and the result of its coin-throwing function to confirm the node's production eligibility, thus preventing block forgery or selfish mining attacks. Furthermore, a time window penalty mechanism is introduced: if a node engages in dishonest behavior within a specified period, its stake will be reduced accordingly.

[0072]

[0073] Exponential reduction is implemented to ensure the long-term stability and fairness of the system. Assume the number of times dishonesty occurs is... Next, then corresponding Then it becomes:

[0074]

[0075] This causes the probability of block issuance to gradually decrease.

[0076] To verify the performance of the hybrid blockchain architecture of this invention in the Internet of Vehicles, experiments compared the traditional PoW mechanism with a hybrid PoS consensus scheme based on VRF. The results showed that the consensus time of traditional PoW increases exponentially with increasing computational difficulty, while the block generation time of the proposed solution remains stable and negligible, achieving low-latency consensus. Regarding storage scalability, through the "on-chain index + off-chain storage" architecture, the proportion of on-chain metadata quickly drops below 5% in high-density urban areas, verifying high scalability under large-scale applications. In the ledger synchronization efficiency test, using periodic header blocks as global checkpoints, nodes only need to verify the latest header block to quickly anchor to the main chain, effectively reducing communication overhead and improving efficiency. Improved synchronization efficiency; In a network of 383 nodes, compared with four schemes—PBFT, PBFT+PoS, HotStuff, and HotStuff+PoS (Epoch-PoSH)—HotStuff+PoS achieved a throughput of 7,700 TPS with a 32-shard configuration, the lowest and most stable consensus latency of approximately 400ms, and the least impact on latency when shards are added; Analysis of the total latency of data uploading to the chain shows that the latency of most message sharing is stable with no obvious long-tail phenomenon, proving that the present invention can provide uniform and efficient quality of service, meet the real-time requirements of the Internet of Vehicles, and outperform traditional schemes in terms of consensus speed, storage saving, synchronization efficiency, and high concurrency throughput.

[0077] Through the above implementation scheme, this invention proposes a trusted data sharing method and system for the Internet of Vehicles (IoV) based on adaptive sharding and a reputation mechanism. The system utilizes a three-layer physical architecture and an on-chain / off-chain dual-chain structure to achieve a closed-loop process of data acquisition, sharding management, credibility assessment, node reputation updates, and hybrid consensus. The spatiotemporal-aware sharding strategy improves the load balancing capability of nodes within shards, the Bayesian dynamic reputation model effectively identifies malicious nodes, and the Epoch-PoSH hybrid consensus mechanism ensures the security and high throughput of the blockchain. By storing raw data off-chain and recording indexes on-chain, high scalability and data integrity are guaranteed. The overall solution of this invention balances performance, security, and scalability, and can meet the needs of real-time trusted data sharing in the IoV environment.

[0078] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for trusted data sharing in the Internet of Vehicles, characterized in that, Includes the following steps: (1) Obtain the geographical location, node load status and node reputation information of the roadside unit (RSU) node and vehicle node in the vehicle network, and dynamically fragment the vehicle network node according to the geographical location, node load status and node reputation information, and divide the network node into multiple autonomous fragments. (2) Elect management nodes within each autonomous shard and construct a blockchain structure for data sharing. Process the event data reported by vehicle nodes, where event index information is recorded on the on-chain structure and shared data is stored in the off-chain storage structure. (3) Receive event information reported by vehicle nodes, evaluate the credibility of event information based on the spatial distance between the node and the event, the information reporting time and the node's historical reputation information, and update the reputation value of the vehicle node based on the evaluation results. (4) Based on the node reputation value and node rights information, block generation is performed within the shard, and the generated blocks are written into the blockchain to realize trusted sharing of vehicle network data and network status synchronization.

2. The method according to claim 1, characterized in that: The dynamic sharding in step (1) is achieved by constructing a sharding loss function. The sharding loss function comprehensively considers the distance between the node and the shard centroid, the node load status, and the node reputation status. By iteratively calculating the loss function value of the node, the node is divided into the shard with the smallest loss value to achieve network load balancing.

3. The method according to claim 2, characterized in that: The fragmentation loss function includes the Euclidean distance between the node and the fragment centroid, the node load impact term, and the node reputation impact term, and the term is weighted by setting load weight coefficient and reputation weight coefficient.

4. The method according to claim 1, characterized in that: The event information credibility assessment in step (3) is achieved by constructing a dynamic reputation model. The dynamic reputation model includes at least a spatial distance decay factor, a time decay factor, and a reputation decay factor. The credibility value of the event information is obtained by jointly calculating the decay factors.

5. The method according to claim 4, characterized in that: The node behavior score is calculated based on the difference between the feedback information from the vehicle node and the final judgment result of the event. The node reputation value is then updated by combining the consistency of the node's historical behavior, so as to achieve dynamic adjustment of the node reputation.

6. The method according to claim 1, characterized in that: The block generation in step (4) adopts a periodic block generation mechanism. At the beginning of each period, a management node is elected to generate a header block that records the global state. During the rest of the period, nodes compete to generate ordinary blocks based on reputation value and equity information.

7. A trusted data sharing system for vehicle networking, characterized in that, include: (1) A sharding management module is used to obtain the geographical location, node load status and node reputation information of RSU nodes and vehicle nodes, and to dynamically shard the network nodes according to the information. (2) Data management module, used to build on-chain structure and off-chain storage structure, and process the event data reported by vehicle nodes. The event index information is recorded in the on-chain structure and the shared data is stored in the off-chain storage structure. (3) Reputation assessment module, used to assess the credibility of event information based on the spatial distance between the node and the event, the information reporting time and the node's historical reputation information, and update the node's reputation value; (4) Block generation module, used to generate blocks based on node reputation value and node rights information, and write the blocks into the blockchain to achieve data sharing.

8. An electronic device, characterized in that, include: (1) At least one processor; (2) A memory that is communicatively connected to the processor; (3) The memory stores a computer program, which, when executed by the processor, causes the processor to perform the method described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program that, when executed by a processor, causes the computer to perform any one of claims 1 to 6.

Images