A car-road cloud homomorphic data aggregation method and system based on a guardian area segmentation encoding mechanism

Abstract

The application discloses a kind of vehicle-road cloud homomorphic data aggregation method and system based on guardian area segmentation encoding mechanism.The method includes generating multiple prime Paillier key;Sensor data is preprocessed and unified bit width;Guardian area segmentation encoding is used, and guardian area is set in each data segment high bit, and encryption after splicing;When ciphertext aggregation, by center offset strategy and ciphertext multiplication, modular inverse operation, addition and subtraction homomorphic aggregation are realized, and carry and borrow cross-section propagation is inhibited.Aggregation result is decrypted after the combination of multiple prime structure and Chinese remainder theorem and Hensel promotion, and segmented decoding is carried out.The application reduces communication and calculation overhead, supports addition and subtraction aggregation, avoids data segment crosstalk, and is suitable for vehicle-road cloud collaborative privacy protection scene.

Description

Technical Field

[0001] This invention relates to the fields of vehicle networking and privacy computing technology, and in particular to a method and system for vehicle-road-cloud homomorphic data aggregation based on a guardian segmentation coding mechanism. Background Technology

[0002] With the development of vehicle-road-cloud cooperative systems, frequent data exchanges are required between on-board units (OBU), roadside units (RSU), and the cloud. This type of data includes vehicle operating status information, environmental perception data, and traffic cooperation information, and is characterized by high real-time requirements, large data volume, and high privacy sensitivity.

[0003] In existing technologies, homomorphic encryption methods (such as the Paillier algorithm) are typically used to encrypt data to protect data privacy, allowing computation to be performed even in ciphertext form, thus preventing plaintext leakage. However, in vehicle-road-cloud multi-node collaborative scenarios, existing technologies still have the following problems in practical applications: (1) Traditional standalone encryption methods have high communication overhead. In some existing technologies, multi-source data usually adopts a single data independent encryption and transmission method. As the number of participating nodes increases, the amount of ciphertext increases significantly, resulting in increased communication load and encryption / decryption time, and affecting the real-time data transmission performance in vehicle-road-cloud collaborative scenarios. (2) Existing data packaging schemes are prone to cross-segment interference. To reduce the communication and encryption / decryption overhead of encrypting and transmitting individual data independently, some technologies employ data concatenation or packetization to map multiple data points into the same plaintext space for encryption. However, during homomorphic addition or subtraction operations, carry and borrow propagation occur in integer arithmetic. When multiple data points are concatenated in segments, carry or borrow from lower-order data may affect higher-order data, causing crosstalk between data segments and impacting the correctness of the decoding result. (3) High computational complexity The traditional Paillier decryption process involves large integer modular exponentiation. As the data size and key length increase, the computational overhead increases significantly, resulting in an efficiency bottleneck in vehicle-road-cloud real-time computing scenarios. (4) Lack of a unified mechanism that takes into account both structured representation and homomorphic computation Existing technologies lack a coordinated design between data organization and homomorphic computation. They either focus on data packaging to reduce communication overhead, but it is difficult to avoid the interference problem of homomorphic operations; or they keep data processed independently to ensure the correctness of computation, but this will bring higher communication and computation costs.

[0004] Therefore, there is an urgent need for a data aggregation method that can improve computational efficiency and accuracy while maintaining the advantages of low communication and computational overhead in data packaging and the security of homomorphic encryption. Summary of the Invention

[0005] To address the problems of high communication overhead, cross-segment interference, and insufficient computational efficiency in the existing homomorphic data aggregation process in vehicle-road-cloud collaborative scenarios, this invention proposes a vehicle-road-cloud homomorphic data aggregation method and system based on a guardian segmentation coding mechanism.

[0006] To improve the computational efficiency of homomorphic encryption algorithms, the proposed method introduces a multi-prime modulus structure based on the Paillier homomorphic encryption algorithm, and combines the Chinese Remainder Theorem (CRT) and Hensel lifting method to optimize the decryption process, thereby reducing the computational overhead caused by large integer modular exponentiation operations.

[0007] To address the issue of cross-segment interference that easily arises from homomorphic operations after multiple data packets are packaged, this invention further constructs a Guard-Band Segmented Encoding Mechanism (GBSE). By setting up guard zones between data segments and combining a center offset strategy, a segmented encoding structure, and supporting processing methods, a data representation model with boundary control capabilities is formed. This effectively suppresses data crosstalk caused by carry and borrow propagation while maintaining the ability to aggregate multiple data.

[0008] Through the above technical solution, this invention, while maintaining the advantages of multi-data packaging, achieves effective isolation between data segments during homomorphic computation, reduces the impact of cross-segment interference on decoding results, and improves the efficiency of homomorphic encryption and decryption, thereby enhancing the data aggregation performance and real-time processing capability in the vehicle-road-cloud collaborative environment.

[0009] This invention provides a method for vehicle-road-cloud homomorphic data aggregation based on a guardian segmentation coding mechanism, comprising:

[0010] Set the security parameter s, and generate the system key using a multi-prime Paillier structure, satisfying: , , , , in, Let N be distinct large prime numbers, lcm be the least common multiple, and {N, g} be the public key. For private key, .

[0011] The system operating parameters are determined, and then the system key and system operating parameters are distributed. The system operating parameters include: the maximum number of entities participating in the aggregation K, the data bit width L collected by the sensors, the number of sensor data collected T, and the guard zone bit width Q. According to the guard zone segmented coding mechanism, the guard zone bit width Q is constrained. The guard zone bit width Q is dynamically constrained based on the maximum number of entities participating in the aggregation K, the data bit width L, and the number of data T, as follows: , , The segmented encoding mechanism of the guardian adopts a center offset strategy. During the aggregation process, the data segment needs to reserve positive and negative fluctuation space at the same time. Therefore, the actual range of security that can be used for aggregation is constrained to half of the data segment capacity. At the same time, the bit width after encoding should be smaller than the bit width of the public key.

[0012] Data from multiple sensors is periodically collected, and the collected multi-source sensor data undergoes unified processing, including outlier removal and bit width standardization. Outlier data is removed according to a preset threshold rule, and all normal data is mapped to a fixed-length L-bit non-negative integer. For data with directional attributes, a representation method that separates amplitude and direction information is adopted, where direction information is preferably represented by angle values ​​or direction identifiers and participates in subsequent encoding and aggregation processing as an independent data segment. In subtraction aggregation scenarios, data can be represented in signed form and mapped to unsigned non-negative integers through offset mapping before participating in encoding. The offset mapping satisfies the following: , in, This is the mapped unsigned data with a width of L bits. Let be the i-th signed data with a width of L bits, where L is the data bit width. In the subtraction aggregation scenario, all data participating in the aggregation adopt a unified offset mapping method, ensuring that the number of data items participating in the operation on both the addition and subtraction sides is consistent. The offsets cancel each other out during the homomorphic operation and do not affect the final difference result.

[0013] Following a segmented encoding mechanism based on the guardian region, a single encrypted representation of multiple data points is achieved by concatenating multiple data points into a single integer representation using a segmented approach. Specifically, each data segment consists of a "guard region + data region," where the guardian region is set at the high bit of the data region and has a default value of 0. During multiple homomorphic aggregation processes, the guardian region only carries the offset introduced by the homomorphic operation. The data region is used to store various types of sensor data. The structure of each data segment is as follows: [Guardian Zone (Q position)][Data Zone (L position)].

[0014] Multiple data points are concatenated according to their displacement to form a unified encoded value. The expression is: , in, Let be the i-th unsigned data, T be the number of data items, Q be the guard block width, and L be the data block width. For the encoding result... Perform Paillier encryption, then encapsulate the ciphertext and send it.

[0015] The received encapsulated ciphertext is parsed to obtain the ciphertext message. Homomorphic computation is then performed on the received ciphertext message, using ciphertext multiplication to achieve plaintext addition and modular inverse operation to achieve subtraction. Specifically, following the guardian segmented coding mechanism, a center offset is introduced to suppress cross-segment interference that may occur during homomorphic operations. An offset is set for each data segment, providing positive and negative fault tolerance margins during aggregation. When homomorphic addition or subtraction is performed, numerical fluctuations generated by the operation are absorbed within the data segment, thus preventing propagation to adjacent data segments. The offset mechanism achieves the offset by multiplying the ciphertext of the data ciphertext with the ciphertext of the guardian offset G. The guardian offset G is: , Calculate the result before performing the aggregation operation. The value is then encrypted using the Paillier algorithm, and the encrypted value is... Homomorphic multiplication with the first data segment that needs to be aggregated is represented at the plaintext level as the guard zone of each data segment being added. Since the default value of the guard zone is 0, the final value of the guard zone is 0. The center offset of each data segment is then completed. Subsequently, addition or subtraction aggregation calculations are performed with other data, and the aggregation results are then encapsulated and sent.

[0016] The aggregation results are parsed and decrypted. During the decryption process, the Chinese Remainder Theorem is used in conjunction with the multi-prime number structure to convert the modulus... Operation decomposition to prime numbers Modulus space, and combined with the Hensel lifting method from low-mode The result gradually recovered to a high modulus. As a result, the computational complexity is reduced, and the decryption of each prime number modulo space can be performed in parallel to improve decryption efficiency. Then, the Chinese Remainder Theorem is used to restore the modulo results of each prime number to the plaintext result.

[0017] According to the segmented encoding mechanism of the guardian, the decryption result is subjected to segmented decoding operation, specifically including: extracting the current least significant bit data segment, removing the center offset, and right shifting. Repeat the aforementioned process until the aggregation results corresponding to all data segments are recovered.

[0018] This invention provides a vehicle-road-cloud homomorphic data aggregation system based on a guardian segmentation coding mechanism, which deploys the aforementioned vehicle-road-cloud homomorphic data aggregation method based on a guardian segmentation coding mechanism into a vehicle-road-cloud system, including:

[0019] The vehicle-road-cloud system comprises four entities: vehicle-mounted unit, roadside unit, cloud server, and key management center.

[0020] During the system initialization phase, security parameters and system operating parameters are determined. The key management center generates a multi-prime Paillier key and distributes the system key and system operating parameters to the vehicle-mounted unit, roadside unit, and cloud. The system operating parameters include the guard zone width Q, data width L, number of data items T, and the maximum number of aggregated entities K supported by the system. The Paillier private key is only distributed to the cloud server for storage and is used to perform decryption operations on the aggregation results. The vehicle-mounted unit and roadside unit only hold the public key parameters. The key distribution process is completed through a secure channel, and existing key protection mechanisms are used to store and protect the private key; this invention does not limit this aspect.

[0021] During the data acquisition phase, the on-board unit periodically collects vehicle sensor data, removes abnormal data according to preset threshold rules, unifies the bit width of normal data to L, and then encodes T data using the guardian segmented encoding mechanism and encrypts them using the public key parameter using Paillier encryption.

[0022] During the encrypted transmission phase, the on-board unit encapsulates the encrypted message, vehicle identification ID, timestamp, and integrity verification information into a data packet and sends it to the roadside unit. The integrity verification is preferably implemented using digital signatures, message authentication codes, or other integrity protection mechanisms; this invention does not limit this approach. The timestamp is used to suppress replay attacks, and the integrity verification information is used to verify the integrity and legitimacy of the message's origin during transmission.

[0023] During the data aggregation phase, the roadside unit parses the received data packets, discards erroneous data packets such as those with failed integrity checks or incorrect timestamps, caches normal data packets, and retains only the data corresponding to the latest timestamp of the same vehicle identification ID. Then, according to the guardian zone segmented encoding mechanism, it calculates the guardian zone offset, uses the public key parameter for Paillier encryption, and participates in the aggregation along with other data. It performs addition aggregation or subtraction aggregation as needed and generates the ciphertext of the aggregation result.

[0024] During the aggregation result distribution phase, the roadside unit encapsulates the aggregated encrypted message, timestamp, roadside unit identifier ID, and integrity verification information into a data packet and sends it to the cloud server.

[0025] During the decryption phase, the cloud server parses the received data packets, discards erroneous data packets such as those with integrity verification failures and timestamp errors, and performs cache management on normal data packets. Then, it uses the Paillier private key to perform multi-prime CRT-Hensel decryption on the aggregation result in the data packet, and completes segmented decoding according to the guardian segmented encoding mechanism to obtain the final aggregation result.

[0026] The beneficial effects of this invention are:

[0027] This invention employs a guardian-segmented encoding mechanism to map multiple data points to the same plaintext space for unified encryption, decryption, and aggregation, enabling multi-data single-ciphertext transmission and processing. This reduces communication overhead and improves data processing efficiency. The guardian-segmented encoding mechanism, by introducing guardian regions and center offsets into data segments, suppresses carry and borrow propagation across segments during homomorphic operations, providing support for addition and subtraction aggregation and ensuring the correctness of the aggregation results. Simultaneously, by combining the Paillier structure for multi-prime numbers, the Chinese Remainder Theorem, and the Hensel lifting method, the decryption process of the aggregation results is optimized, reducing the complexity of large integer operations and improving system real-time performance. Furthermore, this invention can dynamically adjust guardian parameters according to different aggregation scales, possessing good scalability and engineering adaptability. While maintaining the original security of Paillier homomorphic encryption, it achieves a synergistic improvement in aggregation correctness and computational efficiency. Attached Figure Description

[0028] Figure 1 This is a diagram of the vehicle-road-cloud data aggregation system architecture provided in an embodiment of the present invention;

[0029] Figure 2 A flowchart of a vehicle-road-cloud homomorphic data aggregation system based on a guardian segmentation coding mechanism is provided for an embodiment of the present invention. Detailed Implementation

[0030] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but the scope of protection of the present invention is not limited to the following embodiments.

[0031] The following embodiments are deployed in a vehicle-road-cloud cooperative environment. The system includes an on-board unit (OBU), a roadside unit (RSU), a cloud server, and a key management center (KMC). The overall architecture is described in [reference needed]. Figure 1 The vehicle-mounted unit is used to collect vehicle sensor data and perform data encoding and encryption; the roadside unit is used to receive the ciphertext and perform homomorphic aggregation calculations; the cloud server is used to decrypt the aggregation results and perform segmented decoding; and the key management center is used to generate and distribute system keys and system operating parameters.

[0032] In the following embodiment, the security parameter is set to s=1024, the number of prime numbers is k=4, and the key management center generates a multi-prime Paillier key that satisfies: , , , in Given distinct prime numbers, the public key is {N, g}, and the private key is { }

[0033] Set the data width to L=18 bits, the number of data items to T=6, and the maximum number of entities participating in the aggregation to K=8, based on the guardian width constraint: , , Given a guard zone width of Q=4, the structure of each data segment is as follows: [Guard area 4 bits][Data area 18 bits].

[0034] The system operating parameters include: number of data T=6, data bit width L=18, maximum number of entities participating in aggregation K=8, and guard zone bit width Q=4. The key management center distributes the system operating parameters to the vehicle unit, roadside unit, and cloud server through a secure channel, securely deploys the public key to the vehicle unit and roadside unit through a secure channel, and securely deploys the public key and private key to the cloud server through a secure channel.

[0035] The onboard unit collects data from the following six sensors: Vehicle speed value, in km / h; The angle between the vehicle's speed direction and due north, clockwise, is expressed in degrees. The distance of the vehicle from a reference point, in meters; The angle between the line connecting the vehicle and a reference point and due north at the vehicle's position, clockwise, is expressed in degrees. Vehicle acceleration value, in m / s² 2 ; The angle between the direction of vehicle acceleration and due north, expressed in degrees.

[0036] Example 1

[0037] This embodiment uses regional vehicle status statistics as an example to illustrate the additive aggregation process of the present invention. (See also...) Figure 2 .

[0038] The onboard unit first performs bit-width standardization on the sensor data that has been filtered out for anomalies. The distance and acceleration parameters are multiplied by 100, and the velocity is multiplied by 1000. After removing the decimal point, the distance and acceleration are converted to cm and cm / s respectively. 2 The speed is unitized in m / h and then converted into fixed-length 18-bit binary data.

[0039] Subsequently, the data is encoded in a segmented manner according to the above data order. Each data segment is concatenated in the form of "guard zone + data zone". The code value for vehicle ID J is M. J satisfy: , Specifically, the binary expansion of M1 is:

[0040] Each data segment has a guard zone at its high bits, separating the different data areas. The onboard unit performs Paillier encryption on the encoded value. , Where r is a random number satisfying gcd(r,N)=1. Subsequently, the on-board unit encapsulates the ciphertext, its own ID, timestamp, and integrity check into a data packet and sends it to the roadside unit.

[0041] The roadside unit receives and parses the received data packets, discards erroneous packets, obtains the ID and ciphertext from the normal packets, and then calculates the guard zone offset G: , Specifically, the binary expansion of G is:

[0042] The guardian zone for each data segment is set to 2. 4-1 -1=7, and then the roadside unit uses the Paillier algorithm to encrypt G: , Next, a homomorphic multiplication operation is performed with the first ciphertext C1 to be aggregated to achieve the center offset of each data segment: , Obtain the offset aggregated ciphertext C m C m The plaintext binary expansion is:

[0043] The roadside unit performs homomorphic multiplication aggregation on the ciphertext corresponding to multiple vehicles: , The aggregated ciphertext result, its own ID, timestamp, and integrity verification are encapsulated into a data packet and sent to the cloud server.

[0044] The cloud server receives and parses the received data packets, discarding erroneous packets and obtaining the ID and ciphertext from the normal packets. It then uses the private key to decrypt the ciphertext and employs a combination of multiple prime numbers, CRT, and Hensel lifting methods to recover the plaintext. Finally, it performs segmented decoding. Specifically, the decoded result is as follows:

[0045] Based on the aggregation results, further analysis can yield information on the average speed of vehicles in the region, the overall movement trend of the region, the distribution area of ​​vehicles, and road traffic conditions.

[0046] Example 2

[0047] This embodiment uses vehicle relative motion state analysis as an example to illustrate the subtractive aggregation process of the present invention. See [link to documentation]. Figure 2 .

[0048] To facilitate verification of the ability to analyze the relative motion state of vehicles during the subtractive aggregation process, this embodiment constructs a scenario in which two vehicles move along approximately the same lane direction. Therefore, their distance, speed, acceleration components, and differences are the same in the eastward and northward projections.

[0049] In the subtractive aggregation scenario, to represent the direction of motion, this embodiment converts velocity, distance, and acceleration into two directional components: east and north. The east component takes east as the positive direction, and the north component takes north as the positive direction.

[0050] The vehicle-mounted unit collects the sensor data above, after removing any anomalies. The velocity amplitude, distance, and acceleration are multiplied by 1000, 100, and 100 respectively, and the decimal point is removed. The velocity is then converted to m / h, and the distance and acceleration are converted to cm and cm / s respectively. 2 Then it is decomposed into two directions, due east and due north. Specifically, the decomposed data is as follows:

[0051] For data components in two directions, the offset mapping satisfies: , in This is the mapped unsigned data with a width of L bits. For the i-th signed data with a width of L bits (L=18), the offset data is then encoded in a guard-defined segmented manner according to the above order. Specifically, the encoded M1 and M2 are: M1 = 187725 × 2 110 +187725×2 88 +35304×2 66 +226840×2 44 +131505×2 22 +131505, M2 = 184558 × 2 110 +184558×2 88 +60965×2 66 +252501×2 44 +131377×2 22 +131377 The encoded fields of M1 are as follows:

[0052] The encoded fields of M2 are as follows:

[0053] The vehicle unit performs Paillier encryption on the encoded M to obtain ciphertext C1 and C2, and encapsulates the ciphertext, its own ID, timestamp, and integrity check into a data packet and sends it to the roadside unit.

[0054] The roadside unit receives and parses the received data packets, discarding invalid packets and obtaining the ID and ciphertext from the normal data packets. The roadside unit then calculates the guard zone offset G = 7 × 2. 128 +7×2 106 +7×2 84 +7×2 62 +7×2 40 +7×2 18 G is Paillier encrypted, and the encrypted G is homomorphically multiplied with the ciphertext C1 to obtain the ciphertext C. G Specifically, C G The fields are as follows:

[0055] Then take the modular inverse of C2, and multiply the modular inverse by C. G Perform homomorphic multiplication to obtain the aggregated result C. sub : , The aggregation results, ID, timestamp, and integrity verification are encapsulated into a data packet and sent to the cloud server.

[0056] The cloud server receives and parses the received data packets, discards erroneous packets, obtains the ID and ciphertext from the normal packets, then decrypts the ciphertext using the private key, recovers the plaintext using CRT and Hensel lifting methods, and finally performs segmented decoding. The resulting aggregated result is as follows:

[0057] The relative motion between vehicles can be obtained from the aggregation results. The relative motion can be used to analyze the relative motion trend, changes in proximity, and deviations in motion direction between vehicles.

[0058] The above embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the present invention. The segmented encoding mechanism for data protection is not limited to vehicle data in this embodiment, but can also be applied to other scenarios requiring homomorphic aggregation computation of multiple data. For those skilled in the art, any modifications, equivalent substitutions, and improvements made to this invention without departing from its spirit and substance should fall within the scope of protection of the present invention.

Claims

1. A method for aggregating homomorphic vehicle-road-cloud data based on a guardian-based segmented coding mechanism, characterized in that, include: Determine security parameters and system operating parameters, including the guard zone bit width Q, data bit width L, number of data T, and maximum number of entities participating in aggregation K; generate a multi-prime Paillier key and distribute the key and system parameters. Data from multiple sensors is collected, outliers are removed, and data bit width is standardized. According to the segmented coding mechanism of the guard zone, a guard zone is set in the high bit of each data and formed a data segment with the corresponding data. Multiple data segments are spliced ​​together by shifting to form the coded value. Perform Paillier encryption on the encoded value; The encrypted ciphertext is encapsulated, transmitted, received, and cached. According to the guardian segmented coding mechanism, homomorphic aggregation calculation is performed on the received multiple ciphertexts. The homomorphic aggregation calculation includes: introducing a guardian bias G, which satisfies: Where Q is the guard zone width, L is the data width, and T is the number of data items, G is Paillier encrypted and then used for homomorphic aggregation calculation. Each data segment is adjusted to the central region to achieve center offset of the data segment. Local constraints are applied to carry and borrow generated by homomorphic operation. Multiplication is performed on multiple ciphertexts to achieve homomorphic addition. Modulo inverse operation is performed on the ciphertext to be subtracted and then used for homomorphic multiplication to achieve plaintext subtraction. Under segmented structured coding, through guard zone isolation and center offset control, the carry or borrow generated by homomorphic aggregation calculation is restricted to the corresponding data segment, realizing parallel homomorphic calculation of multiple data items. The aggregated ciphertext data is encapsulated, transmitted, received, and cached. The aggregation result is decrypted, and the decryption process includes: using the Chinese Remainder Theorem combined with the multi-prime structure to decompose the modular operation into multiple modulo operations. The computation is optimized to reduce the complexity of modular arithmetic; parallel computation is performed in each sub-modular space; and the Hensel lifting method is used to optimize the modular arithmetic. The result is a step-by-step recovery model The result; the final plaintext is obtained by combining the sub-results; where The i-th prime number selected from the Paillier key for generating multi-prime numbers; According to the segmented encoding mechanism of the guardian, the decryption result is decoded in segments to obtain the corresponding data aggregation result.

2. The method according to claim 1, characterized in that, The multi-prime Paillier key generation includes: Select k distinct large prime numbers The prime numbers satisfy the following: Construct a modulus N that satisfies: The public key parameter g satisfies: Private key parameters satisfy: ,in, Let s be distinct large prime numbers, and s be a safety parameter. lcm(·) denotes the least common multiple function, {N,g} is the public key, { This is the private key.

3. The method according to claim 1, characterized in that, The guard zone width Q is dynamically determined based on the maximum number of entities participating in the aggregation K, the data width L, the number of data segments T, and the public key width N, satisfying the following: , The first constraint is used to limit the carry or borrow generated by each data segment during the homomorphic aggregation calculation process to not exceed the guard zone capacity; the second constraint is used to ensure that the overall data bit width after encoding is less than the bit width of the public key modulus N, so as to meet the data representation range requirements of Paillier encryption.

4. The method according to claim 1, characterized in that, The outlier removal and data bit width unification processing include: determining anomalies in sensor data and removing outlier data that exceeds the normal range; expanding all normal data into a fixed-length L-bit non-negative integer representation; normalizing data exceeding the length range; for data with directional attributes, representing amplitude and direction information separately, with direction information represented by angle values ​​or direction identifiers; in subtraction aggregation scenarios, data can be represented in signed form, and signed integers are mapped to non-negative unsigned integers through offset mapping, wherein the offset mapping satisfies: ,in, This is the mapped unsigned data with a width of L bits. Let be the i-th signed data with a width of L bits, where L is the data bit width. Under the conditions of uniform offset mapping and consistent number of addition and subtraction terms, the offsets cancel each other out during the homomorphic aggregation calculation process and will not affect the correctness of the final difference result.

5. The method according to claim 1, characterized in that, The guardian segmented coding mechanism is a data representation and processing mechanism that includes a guardian zone, a center offset, and a segmented coding structure. In the guardian segmented coding mechanism, T data points correspond to T data segments. Each data segment consists of a Q-bit guardian zone and an L-bit data area. The guardian zone does not carry business data, its initial value is 0, and it is located in the high-order bits of the data segment. The data area is used to store the corresponding data. Let be the i-th data point, and T be the number of data points. The encoded value is constructed by offset concatenation. as follows: In this process, different data are mapped to corresponding data segments through shift operations, enabling segmented storage of multiple data in the same plaintext space. The guard area is used to isolate each data segment in homomorphic aggregation computation and restrict carry and borrow propagation across data segments.

6. The method according to claim 1, characterized in that, The segmented decoding includes: Extract the current low-order data segment; remove the guard zone offset; shift the data to the right by L+Q bits; repeat the above steps until all data is recovered.

7. A vehicle-road-cloud homomorphic data aggregation system based on a guardian-based segmented coding mechanism, characterized in that, include: The on-board unit is used to collect sensor data and process data, encode data according to the segmented coding mechanism of the guard zone and complete encryption, and encapsulate the ciphertext, vehicle identification ID, timestamp and integrity verification information into a data packet and send it to the roadside unit. The roadside unit is used to receive data packets, perform data packet validity verification, perform homomorphic aggregation calculation according to the guardian segmented encoding mechanism, and encapsulate the aggregation result, roadside unit identifier ID, timestamp, and integrity verification information into a data packet and send it to the cloud server. The cloud server is used to receive data packets, perform data packet validity checks, perform multi-prime CRT-Hensel decryption, and decode segments according to the guardian segmented encoding mechanism. A key management center is used to generate multi-prime Paillier keys and system parameters and distribute the keys and system parameters to various entities through a secure channel; wherein the system is used to execute the data aggregation method according to any one of claims 1-6.

8. A computer-readable storage medium having a program stored thereon, characterized in that, When the program is executed by a processor, it implements the method according to any one of claims 1-6.

Images