Method for protecting privacy of sensitive data of a carrier in a smart logistics platform

Abstract

The present application relates to the technical field of electronic digital data processing, in particular to a private protection method for sensitive data of a carrier in a smart logistics platform, comprising designing a local differential privacy protection algorithm for multi-dimensional numerical data of the carrier and a local differential privacy protection algorithm for location data of the carrier; the multi-dimensional numerical data of the carrier is disturbed by using the local differential privacy protection algorithm for multi-dimensional numerical data of the carrier, and private protection numerical data is obtained; the location data of the carrier is disturbed by using the local differential privacy protection algorithm for location data of the carrier, and private protection location data is obtained; the multi-dimensional numerical data and the location data of the carrier are respectively protected by the two protection algorithms, and the problem that the existing smart logistics platform does not protect the logistics process data of the carrier and causes the personal privacy of the carrier to be leaked is solved.

Description

Technical Field

[0001] This invention relates to the field of electronic digital data processing technology, and in particular to a method for protecting the privacy of carrier sensitive data in a smart logistics platform. Background Technology

[0002] Smart logistics is a modern logistics model that integrates Internet of Things (IoT) technology and is one of the most important directions for the development of the logistics industry.

[0003] The intelligent logistics platform is the carrier and core of intelligent logistics. This platform breaks down information barriers between logistics resources and is built on an information sharing mechanism to provide information-based and intelligent logistics services to users at all levels.

[0004] In smart logistics platforms, the logistics information of human entities is highly private. However, current smart logistics platforms only protect the privacy of shippers and consignees, but do not protect the privacy of carrier logistics process data. This can lead to the leakage of carriers' personal privacy, thereby affecting the logistics efficiency of the entire smart logistics platform. Summary of the Invention

[0005] The purpose of this invention is to provide a method for protecting the privacy of carriers' sensitive data in a smart logistics platform, aiming to solve the problem that existing smart logistics platforms do not provide any privacy protection for carriers' logistics process data, which leads to the leakage of carriers' personal privacy.

[0006] To achieve the above objectives, the present invention provides a method for protecting the privacy of carrier sensitive data in a smart logistics platform, comprising the following steps:

[0007] Design local differential privacy protection algorithms for carrier multidimensional numerical data and carrier location data;

[0008] The carrier's multidimensional numerical data is perturbed using the local differential privacy protection algorithm to obtain privacy-protected numerical data;

[0009] The carrier's location data is perturbed using a local differential privacy protection algorithm to obtain privacy-preserving location data.

[0010] The carrier's multidimensional numerical data local differential privacy protection algorithm includes a local differential privacy budget method and an L-PM perturbation algorithm.

[0011] The specific method for perturbing the carrier's multidimensional numerical data using the carrier's local differential privacy protection algorithm to obtain privacy-protected numerical data is as follows:

[0012] The privacy budget for the carrier’s multidimensional numerical data is set using the local differential privacy budget method to obtain the first privacy budget.

[0013] Based on the first privacy budget, the multidimensional numerical data is perturbed using the L-PM perturbation algorithm to obtain privacy-protected numerical data.

[0014] The specific method for setting the privacy budget for the carrier's multidimensional numerical data using the local differential privacy budget method is as follows:

[0015] By using the safety threshold of attribute data, multidimensional numerical data is normalized to the first interval to obtain normalized data.

[0016] Random variables are generated based on the normalized data;

[0017] The random variable is compared with a preset value to obtain the comparison result;

[0018] A first privacy budget is generated based on the comparison results.

[0019] The method further includes, after obtaining privacy-preserving numerical data by perturbing the multidimensional numerical data using the L-PM perturbation algorithm based on the first privacy budget:

[0020] The privacy-protected numerical data is sent to the data server of the smart logistics platform;

[0021] The server collects the privacy-protected numerical data and statistically analyzes the average value of the carrier's attribute data;

[0022] Normalize the statistical mean to obtain the mean estimate.

[0023] The local differential privacy protection algorithm for carrier location data includes a local differential privacy budget method and a personalized perturbation mechanism algorithm for location data.

[0024] The specific method for perturbing the carrier's location data using the local differential privacy-preserving algorithm to obtain privacy-preserving location data is as follows:

[0025] Preprocess the location data to obtain location vectors;

[0026] A second privacy budget is obtained by setting a privacy budget for the carrier's location data using the local differential privacy budget method.

[0027] The location vector is perturbed using the personalized perturbation mechanism algorithm based on the second privacy budget to obtain privacy-preserving location data.

[0028] The specific method for preprocessing the location data to obtain the location vector is as follows:

[0029] Use a quadtree matrix to segment location-type data to obtain location regions;

[0030] The location region is indexed to obtain a location vector.

[0031] The specific method by which the personalized perturbation mechanism algorithm utilizing the location data perturbs the location vector based on the second privacy budget to obtain privacy-preserving location data is as follows:

[0032] The personalized perturbation mechanism algorithm for location data is based on the second privacy budget and iterates through each dimension of the location vector to obtain multiple iterated dimensions;

[0033] The multiple cyclic dimensions are normalized to obtain normalized dimensions;

[0034] By introducing Bernoulli variables into the normalized dimension, privacy-preserving location data is obtained.

[0035] Wherein, after introducing Bernoulli variables into the normalized dimension to obtain privacy-preserving location data, the method includes:

[0036] The region code in the location data and the privacy-protected location data are sent to the data server in the smart logistics platform.

[0037] This invention discloses a method for protecting the privacy of carrier sensitive data in a smart logistics platform. It designs local differential privacy protection algorithms for carrier multidimensional numerical data and carrier location data. The local differential privacy protection algorithm for carrier multidimensional numerical data is used to perturb the carrier's multidimensional numerical data to obtain privacy-protected numerical data. Similarly, the local differential privacy protection algorithm for carrier location data is used to perturb the carrier's location data to obtain privacy-protected location data. By employing these two algorithms to protect the privacy of both carrier multidimensional numerical data and location data, this method solves the problem of existing smart logistics platforms failing to provide any privacy protection for carrier logistics process data, which could lead to the leakage of carriers' personal privacy. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 It is a grid matrix partitioning and quadtree index graph.

[0040] Figure 2 The figures show the experimental results for the UNIFORM and GUASS datasets.

[0041] Figure 3 These are the experimental results for the MX and BR datasets.

[0042] Figure 4 This is a graph showing the experimental results of trajectory ratio estimation performance.

[0043] Figure 5 This is a flowchart of a method for protecting the privacy of carrier sensitive data in a smart logistics platform provided by the present invention. Detailed Implementation

[0044] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0045] Please see Figures 1 to 5 This invention provides a method for protecting the privacy of carrier sensitive data in a smart logistics platform, comprising the following steps:

[0046] S1 designs a local differential privacy protection algorithm for carrier multidimensional numerical data and a local differential privacy protection algorithm for carrier location data;

[0047] Specifically, the carrier's multidimensional numerical data local differential privacy protection algorithm ε-L_LDP (ε-LogisticLocal Differential Privacy, ε-L_LDP) includes a local differential privacy budget method and an L-PM perturbation algorithm.

[0048] The local differential privacy protection algorithm ε-LT_LDP (ε-Logistic Trajectory Location Differential Privacy) for the carrier location-based data includes a local differential privacy budget method and a personalized perturbation mechanism algorithm for location-based data.

[0049] S2 Perturb the multi-dimensional numerical data of the carrier using the local differential privacy protection algorithm for the carrier's multi-dimensional numerical data to obtain privacy-protected numerical data;

[0050] First, we define the symbols that appear in the subsequent algorithms.

[0051] Definition 1: Carrier. A total of n carrier users are registered in the intelligent logistics platform. The carrier set U = (u1, u2,... u n ), where u i (1 < i < n) can be either an individual in the transportation industry or an employee of a carrier company. Carrier u i can set its own privacy budget ε i .

[0052] Definition 2: Carrier multi-dimensional numerical data. The multi-dimensional numerical attribute set A = {A1, A2,..., A n}), where each attribute A j (1 ≤ j ≤ n) is independent of each other.

[0053] Definition 3: Attribute security domain set Γ = {τ1, τ2,..., τ 10}), τ j (1 ≤ j ≤ n) is the numerical security range of the attribute A i that the carrier u j (1 ≤ j ≤ n) can disclose. Among them, the maximum value of the security domain τ j interval is τ j_max and the minimum value is τ j_min .

[0054] Definition 4: A certain attribute of the carrier's numerical data. t i [A j is the value of the attribute A i (1 ≤ i ≤ n) of the user u j (1 ≤ j ≤ 10).

[0055] The specific method is as follows:

[0056] S21 Set the privacy budget of the carrier's multi-dimensional numerical data through the local differential privacy budget method to obtain the first privacy budget;

[0057] The specific method is as follows:

[0058] S211 uses the security threshold of attribute data to normalize multidimensional numerical data to the first interval, thus obtaining normalized data.

[0059] Specifically, a personalized local differential privacy budgeting method is introduced, allowing carriers to set privacy budgets for sensitive data based on their own or their carrier's privacy needs, utilizing the data security thresholds τ of various attributes. j The maximum value τ of the interval j_max With minimum value τ j_min Normalize the data to [-1, 1], and the normalization coefficient is calculated as k = [1 - (-1)] / (τ). j_max -τ j_min The data is normalized to the interval [-1, 1] by calculating the coefficients using normalization.

[0060] Nort i [A j ]=-1+k(t i [A j ]-τ j_min )=2(t i [A j ]-τ j_min ) / (τ j_max -τ j_min )-1.

[0061] S212 generates random variables based on the normalized data;

[0062] Specifically, we introduce a uniform random variable x in the range [0,1].

[0063] S213 compares the random variable with a preset value to obtain a comparison result;

[0064] Specifically, the uniform random variable x and Compare.

[0065] S214 generates a first privacy budget based on the comparison results.

[0066] Specifically, if x is greater than From [l(T i [A j ]),r(T i [A j Randomly select a value from ])] and copy it to Otherwise, it will be from [-C,l(T i [A j ]))∪(r(T i [A j Randomly select a value from []),C] and copy it to return

[0067] S22 Based on the first privacy budget, the multidimensional numerical data is perturbed using the L-PM perturbation algorithm to obtain privacy-protected numerical data.

[0068] Specifically, an L-PM algorithm was designed to personalize the perturbation mechanism for the carrier's multidimensional numerical data in the PM segmentation mechanism.

[0069] The L-PM perturbation algorithm has three input parameters t. i [A j ]、ε i and τ j , t i [A j For user u i Property A of (1≤i≤n) j The value of ε (1≤j≤10) i It restricts attackers in the security domain τ j The carrier lacks the ability to distinguish between any two values ​​within its range. i A personalized privacy budget of (1≤i≤n), τ j For attribute A j The safe range for the numerical value is (1≤j≤10). The output is the response value of the data perturbation, which is a continuous range of values ​​[-C, C].

[0070] Among them, the perturbation value Higher probabilities occur in the middle of the range, while lower probabilities occur at both ends. Equation (1) is its probability density function:

[0071]

[0072] in:

[0073]

[0074]

[0075] r(T i [A j ])=l(T i [A j ])+C-1.

[0076] S23 sends the privacy-protected numerical data to the data server of the smart logistics platform;

[0077] S24 The server collects the privacy protection numerical data and statistically analyzes the average value of the carrier's attribute data;

[0078] S25 performs a normalization operation on the statistical mean to obtain the mean estimate.

[0079] S3 uses a local differential privacy protection algorithm to perturb the carrier's location data to obtain privacy-protected location data.

[0080] The specific method is as follows:

[0081] S31 preprocesses the location data to obtain the location vector;

[0082] The specific method is as follows:

[0083] S311 uses a quadtree matrix to segment positional data and obtain positional regions;

[0084] S312 indexes the location region to obtain a location vector.

[0085] Specifically, initialize vectors M and H, and generate area code H and internal code C using the latitude and longitude of the carrier's location data, where H and C form unit code M. Vectorize. Convert C into a 2h-bit vector B (position vector) such that B[t+1] = 1, B[i] = 0 (i ≠ t+1), where t is the decimal number corresponding to the binary sequence C.

[0086] S32 sets the privacy budget for the carrier's location data using the local differential privacy budget method to obtain a second privacy budget;

[0087] Specifically, a personalized local differential privacy budgeting method is adopted to ensure that each carrier can modify the privacy protection budget for sensitive data as needed.

[0088] S33 uses the personalized perturbation mechanism algorithm of the location data to perturb the location vector based on the second privacy budget to obtain privacy-protected location data.

[0089] The specific method is as follows:

[0090] The personalized perturbation mechanism algorithm L-RR for location data described in S331 is based on the second privacy budget and iterates through each dimension of the location vector to obtain multiple iterated dimensions;

[0091] Specifically, the L-RR algorithm for personalized perturbation of location-based data has two input parameters: user u i A position vector B (1≤i≤n), privacy budget ε i The output parameter is the perturbed position vector S.

[0092] Initialize vector S. Each dimension S of the cyclic vector B. jAnd normalize the loop's dimension to Z.

[0093] S332 normalizes the multiple cyclic dimensions to obtain the normalized dimension Z;

[0094] S333 introduces a Bernoulli variable into the normalized dimension to obtain privacy-preserving location data.

[0095] Specifically, introduce Bernoulli variables. If u equals 1, then S j Assign a value of 1; otherwise, set S to 1. j Assign a value of 0. Return S.

[0096] S334 sends the area code in the location data and the privacy-protected location data to the data server in the smart logistics platform.

[0097] The feasibility of the proposed solution was verified in experiments, and the ε-L_LDP and ε-LT_LDP algorithms were implemented. The experimental machine parameters were: CPU, Intel Core™ i5-10500 CPU×6@3.1GHz, 32GB RAM, 500GB storage, Windows 10; the programming software used was MATLAB R2020b and PyCharm Community Edition 2021.2; and the algorithms were written in Python 3.6. The datasets used were: randomly generated simulation datasets, the GAUSS dataset and the UNIFORM dataset. The GAUSS dataset follows a Gaussian distribution with a mean of 60 and a standard deviation of 0.2. The UNIFORM dataset follows a uniform distribution between [30, 70]. Two public datasets, the BR dataset and the MX dataset, were extracted from IPUMS. The trajectory location dataset consists of GPS data from over 1,000 freight vehicles belonging to the smart logistics platform company from June 1, 2019 to June 30, 2019. A rectangular geographical area with high coverage trajectory records was selected, totaling 498,565 trajectory records. The time resolution was set to 300 seconds, resulting in a spatiotemporal dataset of 2858×64 GPS points.

[0098] The evaluation criterion used in the ε-L_LDP algorithm experiments is the mean square error (MSE). The definition of MSE is as follows:

[0099]

[0100] Among them, t i For real data, For estimated data.

[0101] The evaluation criteria used in the ε-LT_LDP algorithm experiments were Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE), defined as follows:

[0102]

[0103]

[0104] Where, p t It is the true proportion of the t-th trajectory (position). It is the estimated proportion of the t-th trajectory (position).

[0105] Experimental results are as follows Figure 2 , 3 As shown in Figure 4.

[0106] The above-disclosed embodiments are merely preferred embodiments of a method for protecting the privacy of carrier sensitive data in a smart logistics platform of the present invention. Of course, they should not be construed as limiting the scope of the present invention. Those skilled in the art can understand that implementing all or part of the above embodiments and making equivalent changes in accordance with the claims of the present invention are still within the scope of the invention.

Claims

1. A method for protecting the privacy of carrier sensitive data in a smart logistics platform, characterized in that, Includes the following steps: Design a local differential privacy protection algorithm for carrier multidimensional numerical data and a local differential privacy protection algorithm for carrier location data; the local differential privacy protection algorithm for carrier multidimensional numerical data includes a local differential privacy budget method and an L-PM perturbation algorithm; The L-PM perturbation algorithm is a personalized perturbation mechanism algorithm designed based on the PM segmentation mechanism for carrier multidimensional numerical data. The L-PM perturbation algorithm has three input parameters t. i [A j ]、ε i and τ j , t i [A j For the carrier u i Attribute A j The values ​​of ε, where 1≤i≤n, 1≤j≤10; where n is the number of carrier users, ε i It restricts attackers in the security domain τ j The carrier lacks the ability to distinguish between any two values ​​within its range. i Personalized privacy budget, τ j For attribute A j The safe range of the numerical value; the output is the response value of the data perturbation, which is a continuous range of values ​​[-C, C], where Among them, the perturbation value Higher probabilities occur in the middle of the range, while lower probabilities occur at both ends; equation (1) is its probability density function: in: r(T i [A j ])=l(T i [A j ])+C-1; The carrier's multidimensional numerical data is perturbed using the local differential privacy protection algorithm to obtain privacy-protected numerical data; The carrier's location data is perturbed using a local differential privacy-preserving algorithm to obtain privacy-preserving location data; Preprocessing of location data yields location vectors, where quadtree matrices are used to segment location data to obtain location regions; Index the location region to obtain the location vector; A second privacy budget is obtained by setting a privacy budget for the carrier's location data using the local differential privacy budget method. The location vector is perturbed using the personalized perturbation mechanism algorithm based on the second privacy budget to obtain privacy-preserving location data; The personalized perturbation mechanism algorithm for location data is based on the second privacy budget and iterates through each dimension of the location vector to obtain multiple iterated dimensions; The local differential privacy protection algorithm for carrier location data includes a local differential privacy budget method and a personalized perturbation mechanism algorithm for location data. The personalized perturbation mechanism algorithm for location data, L-RR, has two input parameters: carrier u. i Position vector B, privacy budget ε i The output parameter is the perturbed position vector S. Initialize vector S; each dimension S of the cyclic vector B j And normalize that dimension of the loop to Z; The multiple cyclic dimensions are normalized to obtain the normalized dimension Z; Introducing Bernoulli variables into the normalized dimension yields privacy-preserving location data; introducing Bernoulli variables... If u equals 1, then S j Assign a value of 1; otherwise, set S to 1. j If assigned a value of 0, return S.

2. The method for protecting the privacy of carrier sensitive data in a smart logistics platform as described in claim 1, characterized in that, The specific method for perturbing the carrier's multidimensional numerical data using the carrier's local differential privacy protection algorithm to obtain privacy-protected numerical data is as follows: The privacy budget for the carrier’s multidimensional numerical data is set using the local differential privacy budget method to obtain the first privacy budget. Based on the first privacy budget, the multidimensional numerical data is perturbed using the L-PM perturbation algorithm to obtain privacy-protected numerical data.

3. The method for protecting the privacy of carrier sensitive data in a smart logistics platform as described in claim 2, characterized in that, The specific method for setting the privacy budget for the carrier's multidimensional numerical data using the local differential privacy budget method is as follows: By using the safety threshold of attribute data, multidimensional numerical data is normalized to the first interval to obtain normalized data. Random variables are generated based on the normalized data; The random variable is compared with a preset value to obtain the comparison result; A first privacy budget is generated based on the comparison results.

4. The method for protecting the privacy of carrier sensitive data in a smart logistics platform as described in claim 3, characterized in that, After the first privacy budget is obtained by perturbing the multidimensional numerical data using the L-PM perturbation algorithm to obtain privacy-preserving numerical data, the method further includes: The privacy-protected numerical data is sent to the data server of the smart logistics platform; The server collects the privacy-protected numerical data and statistically analyzes the average value of the carrier's attribute data; Normalize the statistical mean to obtain the mean estimate.

5. The method for protecting the privacy of carrier sensitive data in a smart logistics platform as described in claim 1, characterized in that, After introducing Bernoulli variables into the normalized dimension to obtain privacy-preserving location data, the method further includes: The region code in the location data and the privacy-protected location data are sent to the data server in the smart logistics platform.

Images