Homomorphic encryption traffic flow prediction method based on location privacy protection

Abstract

The invention belongs to the technical field of intelligent traffic, and particularly relates to a homomorphic encryption traffic flow prediction method based on location privacy protection, which comprises the following steps: a service requester generates a key pair, a central server generates a key pair, a service node generates a key pair, and a vehicle generates a key pair. Determining coordinates of a task area, performing first-layer encryption on the public key of the central server, and then performing second-layer encryption on the public key of the service node; the two-layer encrypted task area coordinates and task contents are broadcasted to related service nodes; the first layer of encrypted task area coordinates are obtained through decryption of a private key, and then task content is broadcasted to all vehicles in the task area coordinates; determining traffic data in a vehicle acquisition task area serving as a task executor, and analyzing and processing the acquired data by the central server to generate a final report; and carrying out traffic data prediction on the collected traffic data, and predicting traffic data in a future time period.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent transportation, and particularly relates to a homomorphic encryption traffic flow prediction method based on location privacy protection. Background Art

[0002] Intelligent Transportation Systems (ITS) rely on accurate traffic flow prediction to optimize traffic dynamics, alleviate congestion, and improve efficiency. Mobile Crowdsensing (MCS) uses vehicle networks to collaborate in data collection, which is an important way to achieve this goal. Existing MCS systems mostly collect data in the following ways:

[0003] In the case of using federated learning, although the sharing of raw data is avoided, there is still a risk of privacy leakage;

[0004] In the case of using differential privacy, privacy is protected by adding noise, but the prediction accuracy will be reduced;

[0005] In the case of using graph neural networks, traffic data can be effectively processed, but it does not have the ability to protect privacy itself.

[0006] It can be seen that existing MCS systems have problems in data privacy protection. Therefore, a new mechanism is needed that can comprehensively consider privacy protection, data availability, and computational efficiency to achieve more secure and accurate traffic flow prediction. For this purpose, we propose a homomorphic encryption traffic flow prediction method based on location privacy protection. Summary of the Invention

[0007] The purpose of the present invention is to provide a homomorphic encryption traffic flow prediction method based on location privacy protection to solve the problems existing in the background art.

[0008] To achieve the above technical purpose, the technical solution adopted by the present invention is as follows:

[0009] A homomorphic encryption traffic flow prediction method based on location privacy protection includes the following steps:

[0010] S100: Generate key pairs (pk, sk) for each entity in the system, where pk is the public key and sk is the private key. Among them, the service requester generates the key pair (pk r , sk r ), the central server generates the key pair (pk c , sk c ), the service node generates the key pair (pk e , sk e ), and the vehicle generates the key pair (pk v , sk v );

[0011] Each entity discloses its public key while keeping the private key confidential;

[0012] S200: The service requester determines the task area coordinates [(lx, rx), (ly, ry)], where (lx, rx) represents the left - right longitude range of the task area, and (ly, ry) represents the lower - upper latitude range of the task area;

[0013] The service requester uses the public key pk of the central server c to perform the first - layer encryption on the task area coordinates [(lx, rx), (ly, ry)] to obtain the first - layer encrypted task area coordinates

[0014] Then, the service requester uses the public key pk of the service node e to perform the second - layer encryption on the first - layer encrypted task area coordinates to obtain the two - layer encrypted task area coordinates

[0015] S300: The service requester sends the two - layer encrypted task area coordinates and the task content T to the central server, and then the central server broadcasts the received information to the relevant service nodes;

[0016] where the task content T represents the traffic data to be sensed, the reward policy, and the sensing time;

[0017] S400: After the service node receives and the task content T, it first uses the private key sk e to decrypt and obtain the first - layer encrypted task area coordinates

[0018] Then the service node broadcasts the task content T to all vehicles within the task area coordinates ;

[0019] S500: Determine the vehicle as the task executor. This vehicle collects the traffic data within the task area, uploads the collected traffic data to the corresponding service node. The service node processes and stores the received data, and sends the data to the central server. The central server and the service node cooperate to analyze and process the collected data to generate the final report;

[0020] S600: Use a neural network model based on gated recurrent units to predict the collected traffic data, so as to predict the traffic data in the future time period.

[0021] The specific first - layer encryption in step S200 is as follows:

[0022] It means that in the first - layer encryption, the left - hand longitude coordinate is encrypted by the public key of the central server;

[0023] It means that in the first - layer encryption, the right - hand longitude coordinate is encrypted by the public key of the central server;

[0024] It means that in the first - layer encryption, the lower - side latitude coordinate is encrypted by the public key of the central server;

[0025] It means that in the first - layer encryption, the upper - side latitude coordinate is encrypted by the public key of the central server;

[0026] The second - layer encryption in step S200 is specifically as follows:

[0027] It means that in the second - layer encryption, the left - hand longitude coordinate is encrypted twice by the public key of the service node;

[0028] It means that in the second - layer encryption, the right - hand longitude coordinate is encrypted twice by the public key of the service node;

[0029] It means that in the second - layer encryption, the lower - side latitude coordinate is encrypted twice by the public key of the service node;

[0030] It means that in the second - layer encryption, the upper - side latitude coordinate is encrypted twice by the public key of the service node.

[0031] The decryption in step S400 is specifically as follows:

[0032] It means the left - hand longitude coordinate in the first - layer encryption;

[0033] It means the right - hand longitude coordinate in the first - layer encryption;

[0034] It means the lower - side latitude coordinate in the first - layer encryption;

[0035] It means the upper - side latitude coordinate in the first - layer encryption.

[0036] "Determining the vehicle as the task executor" in step S500 specifically includes the following steps:

[0037] S510: The vehicle responding to the task content T uses the public key pk of the central server c to encrypt its own position coordinates (x0, y0), where x0 represents the longitude position of the vehicle and y0 represents the latitude position of the vehicle, and obtains the encrypted position information and will Send it to the corresponding service node;

[0038] S520: The service node uses the homomorphic encryption algorithm to calculate the encrypted task area coordinates in the first layer and the encrypted position information of the vehicle to obtain

[0039] Then send the calculation result to the central server;

[0040] S530: The central server decrypts the calculation result with the private key sk c to obtain [(Δ lx , Δ rx ), (Δ ly , Δ ry )], and determines the calculation result [(Δ lx , Δ rx ), (Δ ly , Δ ry )] to determine whether the vehicle responding to the task is within the task area. If the vehicle is within the task area, the central server designates the vehicle as the task executor.

[0041] The encryption in step S510 is specifically as follows:

[0042] represents the position coordinates encrypted with the public key of the central server;

[0043] represents the position coordinates encrypted with the public key of the central server.

[0044] The calculation in step S520 is specifically as follows:

[0045] represents and homomorphic encryption;

[0046] represents and homomorphic encryption;

[0047] represents and homomorphic encryption;

[0048] represents and homomorphic encryption.

[0049] The decryption in step S530 is specifically as follows:

[0050]

[0051] The result determination condition in step S530 is as follows:

[0052] If Δ lx ≤ 0 ∧ Δ rx ≥ 0 ∧ Δ ly ≤ 0 ∧ Δ ry ≥ 0, it is necessary to simultaneously satisfy Δ lx ≤ 0, Δ rx ≥ 0, Δ ly ≤ 0, and Δ ry ≥ 0, then it is determined that the vehicle is within the task area.

[0053] The present invention has the following advantages:

[0054] ① Balance between privacy protection and data availability: Devote to maintaining the balance between privacy protection and data availability during the data collection and processing process to ensure the accuracy of traffic flow prediction.

[0055] ② Vehicle location privacy protection: Adopt innovative methods to protect the vehicle location information. Encrypt the location data by using the public key of the central server and encryption technology to generate encrypted location information. Even if the edge nodes can access the encrypted location information, they cannot restore the original vehicle location data through decryption operations.

[0056] ③ Task area location privacy protection: When the service requester determines the location area of the task, the public key of the central server will be used to encrypt the location of the task area to generate ciphertext. The edge nodes can access the encrypted task area location information, but they also cannot decrypt the original task area location. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] The present invention can be further illustrated by the non-limiting embodiments given in the drawings.

[0058] Figure 1 is a schematic flow chart of the present invention;

[0059] Figure 2 is a schematic flow chart of the present invention for determining the task execution vehicle. DETAILED DESCRIPTION OF THE INVENTION

[0060] In order to enable those skilled in the art to better understand the present invention, the technical solution of the present invention will be further described below in conjunction with the drawings and embodiments.

[0061] Embodiment:

[0062] A homomorphic encryption traffic flow prediction method based on location privacy protection, characterized in that it includes the following steps:

[0063] S100: Generate key pairs (pk, sk) for each entity in the system, where pk is the public key and sk is the private key. Among them, the service requester generates the key pair (pk r , sk r ), the central server generates the key pair (pk c , sk c ), the service node generates the key pair (pk e , sk e ), and the vehicle generates the key pair (pk v , sk v );

[0064] Each entity publishes its public key and keeps the private key secret.

[0065] The service node refers to a base station with computing and encryption capabilities;

[0066] In this step, key generation is represented as KeyGen(ε) → (pk, sk);

[0067] For a given security parameter ε, a pair of public and private keys (pk, sk) can be generated. Specifically, we select two prime numbers p and q that satisfy the relationship p > 2 ε , q > 2 ε ;

[0068] Set n = pq, and λ = θ(n) = lcm(p - 1, q - 1), where θ represents the Carmichael function;

[0069] Randomly select an integer between 0 and n 2 -1 that is relatively prime to n 2 such that gcd(L(g λ mod n 2 ), n) = 1, where mod is the modulo operation and the function L(u) = (u - 1) / 1;

[0070] Finally, take pk = (n, g) and sk = λ as the key pair.

[0071] S200: The service requester determines the task area coordinates [(lx, rx), (ly, ry)], where (lx, rx) represents the left - right longitude range of the task area, and (ly, ry) represents the lower - upper latitude range of the task area;

[0072] The service requester performs the first - layer encryption on the task area coordinates [(lx, rx), (ly, ry)] using the public key pk c of the central server to obtain the first - layer encrypted task area coordinates

[0073] The first - layer encryption is specifically as follows:

[0074] It means that in the first - layer encryption, the left - hand longitude coordinate is encrypted by the public key of the central server;

[0075] It means that in the first - layer encryption, the right - hand longitude coordinate is encrypted by the public key of the central server;

[0076] It means that in the first - layer encryption, the lower - side latitude coordinate is encrypted by the public key of the central server;

[0077] It means that in the first - layer encryption, the upper - side latitude coordinate is encrypted by the public key of the central server;

[0078] Then, the service requester performs a second - layer encryption on the task - area coordinates after the first - layer encryption through the public key pk e of the service node to obtain the task - area coordinates after two - layer encryption

[0079] The second - layer encryption is specifically as follows:

[0080] It means that in the second - layer encryption, the left - hand longitude coordinate is double - encrypted by the public key of the service node;

[0081] It means that in the second - layer encryption, the right - hand longitude coordinate is double - encrypted by the public key of the service node;

[0082] It means that in the second - layer encryption, the lower - side latitude coordinate is double - encrypted by the public key of the service node;

[0083] It means that in the second - layer encryption, the upper - side latitude coordinate is double - encrypted by the public key of the service node;

[0084] In this step, the encryption operation is expressed as Enc(m, pk)→c, that is, for the known plaintext m, the public key pk is used to generate the ciphertext c for the plaintext m. Specifically, a random selection is made Calculate c = g m r n mod n 2 , and c is returned as the ciphertext of m.

[0085] S300: The service requester sends the task - area coordinates after two - layer encryption and the task content T to the central server, and then the central server broadcasts the received information to the relevant service nodes;

[0086] Among them, the task content T represents the traffic data to be sensed, the reward strategy, and the sensing time;

[0087] S400: The service node receives and the task content T, and first uses the private key sk e to decrypt and obtain the task area coordinates encrypted at the first layer

[0088] Specifically:

[0089] represents the left longitude coordinate in the first layer of encryption;

[0090] represents the right longitude coordinate in the first layer of encryption;

[0091] represents the lower latitude coordinate in the first layer of encryption;

[0092] represents the upper latitude coordinate in the first layer of encryption;

[0093] Then the service node broadcasts the task content T to all vehicles within the task area coordinates inside.

[0094] In this step, the decryption operation is represented as Dec(c, sk) → m, that is, for the known ciphertext c, the private key sk is used to restore its plaintext m, following the following relationship:

[0095] m = L(c λ mod n 2 ) / L(g λ mod n 2 ) mod n.

[0096] S500: Determine the vehicle as the task executor. This vehicle collects traffic data (such as traffic flow, driving speed, etc.) within the task area. To protect privacy, the vehicle can perform preliminary processing on the collected data and then upload it. For example, it can perform blurring processing on the precise location, and upload the collected traffic data to the corresponding service node. The service node processes and stores the received data, and sends the data to the central server. The central server and the service node cooperate to analyze and process the collected data to generate a final report;

[0097] In this step, "determine the vehicle as the task executor" specifically includes the following steps:

[0098] S510: The vehicle responding to the task content T uses the public key pk of the central server cEncrypt its own position coordinates (x0, y0), where x0 represents the longitude position of the vehicle and y0 represents the latitude position of the vehicle, to obtain the encrypted position information and send it to the corresponding service node;

[0099] Specifically:

[0100] represents the longitude position coordinates encrypted by the public key of the central server;

[0101] represents the latitude position coordinates encrypted by the public key of the central server;

[0102] S520: The service node uses the homomorphic encryption algorithm to calculate the first-layer encrypted task area coordinates and the encrypted position information of the vehicle to obtain

[0103] Specifically:

[0104] represents and 's homomorphic encryption;

[0105] represents and 's homomorphic encryption;

[0106] represents and 's homomorphic encryption;

[0107] represents and 's homomorphic encryption;

[0108] Then send the calculation result to the central server;

[0109] S530: The central server decrypts the calculation result with the private key sk c to obtain [(Δ lx , Δ rx ), (Δ ly , Δ ry ), and for the calculation result [(Δ lx , Δ rx ), (Δ ly , Δ ry)]Make a determination to determine whether the vehicle responding to the task is within the task area. If the vehicle is within the task area, the central server will use this vehicle as the task executor;

[0110] The decryption calculation is specifically as follows:

[0111]

[0112] In this step, the result determination condition is:

[0113] If Δ lx ≤ 0 ∧ Δ rx ≥ 0 ∧ Δ ly ≤ 0 ∧ Δ ry ≥ 0, it is necessary to simultaneously satisfy Δ lx ≤ 0, Δ rx ≥ 0, Δ ly ≤ 0, and Δ ry ≥ 0, then it is determined that the vehicle is within the task area.

[0114] S600: Use a neural network model based on a gated recurrent unit to predict the collected traffic data, so as to predict the traffic data in the future time period;

[0115] In this step, specifically:

[0116] S610: Construct the model input time series data x represents the data input at each successive time step;

[0117] Output time series data y represents the predicted data output at each successive future time step;

[0118] Unit hidden state h represents the hidden state vector at each successive time step, containing the time information and state information of historical traffic data;

[0119] S620: Update the gate calculation to determine how much memory from the previous time step should be retained and how much should be updated from the current time step:

[0120]

[0121] Among them, σ is the Sigmoid activation function, represents the input at the current time step t, and respectively represent the weight matrix parameters of the update gate, represents the unit state at the previous time step t - 1;

[0122] In this step, the result is mapped to the range (0, 1) through the Sigmoid activation function;

[0123] S630: Reset gate calculation, determining how to combine the current input with past memories:

[0124]

[0125] where σ is the Sigmoid activation function, represents the reset gate value at the current time step t, and represent the weight matrix parameters of the reset gate respectively, represents the cell state at the previous time step t - 1;

[0126] In this step, the result is mapped to the range (0, 1) through the Sigmoid activation function;

[0127] S640: Candidate activation, including the combined vector of the current input and the memory of the previous time step:

[0128]

[0129] where tanh represents the hyperbolic tangent activation function, mapping the data to the range (-1, 1), W and U represent the weight matrix parameters in the candidate activation calculation respectively, and ⊙ represents the element-wise multiplication Hadamard product;

[0130] S650: Hidden state update, calculating the hidden state at the current time step based on the value of the update gate:

[0131]

[0132] where ⊙ represents the element-wise multiplication Hadamard product;

[0133] Through the above model, the traffic flow in the future time period can be predicted based on historical traffic data.

[0134] The present invention has the following advantages:

[0135] ① Balance between privacy protection and data availability: Committed to maintaining the balance between privacy protection and data availability during the data collection and processing process to ensure the accuracy of traffic flow prediction.

[0136] ② Vehicle location privacy protection: An innovative method is adopted to protect the vehicle location information. By using the public key of the central server and encryption technology to encrypt the location data, encrypted location information is generated. Even if the edge nodes can access the encrypted location information, they cannot restore the original vehicle location data through decryption operations.

[0137] ③Location privacy protection for the task area: When determining the location area of the task, the service requester will use the public key of the central server to encrypt the location of the task area to generate ciphertext. The edge node can access the encrypted location information of the task area but cannot decrypt the original location of the task area.

[0138] The above embodiments are only used to exemplarily illustrate the principles and effects of the present invention, rather than to limit the present invention. Any person familiar with this technology can modify or change the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or changes made by those with ordinary knowledge in the technical field without departing from the spirit and technical ideas disclosed by the present invention should still be covered by the claims of the present invention.

Claims

1. A homomorphic encryption traffic flow prediction method based on location privacy protection, characterized in that: It includes the following steps: S100: Generate key pairs (pk, sk) for each entity in the system, where pk is the public key and sk is the private key. Among them, the service requester generates the key pair (pk r , sk r ), the central server generates the key pair (pk c , sk c ), the service node generates the key pair (pk e , sk e ), and the vehicle generates the key pair (pk v , sk v ); Each entity discloses its public key and keeps its private key confidential; S200: The service requester determines the task area coordinates [(lx, rx), (ly, ry)], where (lx, rx) represents the left and right longitude ranges of the task area, and (ly, ry) represents the lower and upper latitude ranges of the task area; The service requester uses the public key pk of the central server c to perform the first-layer encryption on the task area coordinates [(lx, rx), (ly, ry)] to obtain the first-layer encrypted task area coordinates Then, the service requester performs a second-layer encryption on the task area coordinates encrypted in the first layer using the public key pk of the service node e to obtain the task area coordinates encrypted in two layers S300: The service requester sends the task area coordinates encrypted twice and the task content T to the central server, and then the central server broadcasts the received information to relevant service nodes; Among them, the task content T represents the traffic data to be sensed, the reward policy, and the sensing time; S400: After the service node receives and the task content T, it first uses the private key sk e to decrypt and obtain the first-layer encrypted task area coordinates Then the service node broadcasts the task content T to all vehicles within the task area coordinates ; S500: Determine the vehicle as the task executor. The vehicle collects the traffic data within the task area, uploads the collected traffic data to the corresponding service node. The service node processes and stores the received data, and sends the data to the central server. The central server collaborates with the service node to analyze and process the collected data to generate a final report; S600: Use a neural network model based on a gated recurrent unit to predict the collected traffic data, so as to predict the traffic data in the future time period.

2. The homomorphic encryption traffic flow prediction method based on location privacy protection according to claim 1, wherein: The first-layer encryption in step S200 is specifically: Indicates that in the first layer of encryption, the left longitude coordinate is encrypted by the public key of the central server; It means that in the first-layer encryption, the longitude coordinate on the right is encrypted with the public key of the central server; It means that in the first - layer encryption, the lower - side latitude coordinates are encrypted with the public key of the central server; In the first layer of encryption, the upper latitude coordinates are encrypted using the public key of the central server; The second-layer encryption in step S200 is specifically: It means that in the second-layer encryption, the longitude coordinate on the left is encrypted twice with the public key of the service node; It means that in the second-layer encryption, the longitude coordinate on the right is encrypted twice with the public key of the service node; It means that in the second-layer encryption, the lower latitude coordinates are encrypted twice with the public key of the service node; It means that in the second-layer encryption, the upper latitude coordinates are encrypted twice with the public key of the service node.

3. A homomorphic encryption traffic flow prediction method based on location privacy protection according to claim 2, characterized in that: The decryption in step S400 is specifically: Indicates the left longitude coordinate in the first - layer encryption; Represents the right longitude coordinate in the first layer of encryption; Indicates the lower latitude coordinate in the first-layer encryption; Indicates the upper latitude coordinate in the first layer of encryption.

4. A homomorphic encryption traffic flow prediction method based on location privacy protection according to claim 3, characterized in that: The "determining the vehicle as the task executor" in step S500 specifically includes the following steps: S510: The vehicle in response to the task content T uses the public key pk of the central server c to encrypt its own position coordinates (x0, y0), where x0 represents the longitude position of the vehicle and y0 represents the latitude position of the vehicle, to obtain the encrypted position information and then send it to the corresponding service node; S520: The service node uses a homomorphic encryption algorithm to calculate the encrypted task area coordinates in the first layer and the encrypted location information of the vehicle to obtain Then send the calculation result to the central server; S530: The central server decrypts the calculation result using the private key sk c to obtain [(Δ lx , Δ rx ), (Δ ly , Δ ry )], and determines the calculation result [(Δ lx , Δ rx ), (Δ ly , Δ ry )] to determine whether the vehicle responding to the task is within the task area. If the vehicle is within the task area, the central server designates the vehicle as the task executor.

5. A homomorphic encryption traffic flow prediction method based on location privacy protection according to claim 4, characterized in that: The encryption in step S510 is specifically: Indicates the location coordinates encrypted by the public key of the central server; Indicates the location coordinates encrypted by the public key of the central server.

6. The homomorphic encryption traffic flow prediction method based on location privacy protection according to claim 5, characterized in that: The calculation in step S520 is specifically: representation and homomorphic encryption of representation and homomorphic encryption of representation and homomorphic encryption of representation and homomorphic encryption of 7. A homomorphic encryption traffic flow prediction method based on location privacy protection according to claim 6, characterized in that: The decryption in step S530 is specifically:

8. A homomorphic encryption traffic flow prediction method based on location privacy protection according to claim 7, characterized in that: The result determination condition in step S530 is: If Δ lx ≤ 0 ∧ Δ rx ≥ 0 ∧ Δ ly ≤ 0 ∧ Δ ry ≥ 0, it is necessary to simultaneously satisfy Δ lx ≤ 0, Δ rx ≥ 0, Δ ly ≤ 0, and Δ ry ≥ 0, then it is determined that the vehicle is within the mission area.

Images