CN109508559A - Multidimensional data local method for secret protection in intelligent perception system based on contiguous function - Google Patents
Multidimensional data local method for secret protection in intelligent perception system based on contiguous function Download PDFInfo
- Publication number
- CN109508559A CN109508559A CN201811301086.6A CN201811301086A CN109508559A CN 109508559 A CN109508559 A CN 109508559A CN 201811301086 A CN201811301086 A CN 201811301086A CN 109508559 A CN109508559 A CN 109508559A
- Authority
- CN
- China
- Prior art keywords
- data
- secret protection
- multidimensional
- local
- dimension
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/60—Protecting data
- G06F21/62—Protecting access to data via a platform, e.g. using keys or access control rules
- G06F21/6218—Protecting access to data via a platform, e.g. using keys or access control rules to a system of files or objects, e.g. local or distributed file system or database
- G06F21/6245—Protecting personal data, e.g. for financial or medical purposes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/60—Protecting data
- G06F21/602—Providing cryptographic facilities or services
Abstract
The invention discloses the multidimensional data local method for secret protection in a kind of intelligent perception system based on contiguous function.The method that method proposed by the present invention is primarily based on local privacy conversion; local secret protection is carried out in user terminal to the perception data of participating user; it is then based on contiguous function and data after secret protection is sampled and synthesized; finally obtain the data set with local secret protection that can directly issue; compared to general method for secret protection; the present invention is not only able to provide local difference secret protection in user terminal, and publication data utility is improved by reducing Multidimensional Awareness data value field space.The present invention realizes the dimension-reduction treatment of multidimensional data during realization, and expense is only limited in the part of local conversion and dimensional probability distribution estimation, is not necessarily to complex calculation or encrypting and decrypting.In intelligent perception system, the present invention is simple, is easily achieved, and scalability is strong, and practical value is high.
Description
Technical field
The invention belongs to secret protection fields, and in particular to the multidimensional number based on contiguous function in a kind of intelligent perception system
According to local method for secret protection.
Background technique
With the arriving and rapid development of information age, extensive data source and various data fusion are formd true
Big data intelligent perception system, the perception data from a large amount of participating users will converge in server end, finally service
Multidimensional intelligent perception data after convergence are sent to third party's tissue and analyzed and researched by device, raw with the production for facilitating people
It is living, a more efficient social environment is provided.But directly publication higher-dimension perception data can expose the privacy letter of participating user
Breath, and due to incidence relation between higher-dimension gunz data attribute, unprecedented privacy threats are even more resulted in, are made simultaneously
Secret protection is obtained to be faced with formidable challenges.The existing article largely based on difference privacy technology is suggested at present, is estimated from statistics
The angle of meter protects the privacy of user.But when facing the multidimensional gunz data of Attribute Association relationship complexity, these methods
Computation complexity and in terms of all existing defects, and be difficult to avoid that privacy threats brought by insincere server.
The privacy publication for multidimensional data in intelligent perception system mainly faces the severe challenge of two aspects at present.On the one hand, existing
Most methods assume that server is believable, i.e., non-local secret protection, to be subject to internal attack;Another party
Face, the output of Multidimensional Awareness data is huge, and when perception data dimension increases, existing method is subject to " dimension disaster ",
And signal-to-noise ratio can be reduced seriously.
Summary of the invention
Present invention aim to address privacy of user protection problems in intelligent perception system, provide a kind of Intellisense system
Multidimensional data local method for secret protection in system based on contiguous function.The present invention can be to join in effective protection intelligent perception system
With the data-privacy of user, while guarantee efficient computing cost and improve publication data utility.
The present invention adopts the following technical scheme that realize:
Multidimensional data local method for secret protection in intelligent perception system based on contiguous function, first to each in system
The data of user are all based on Randomized response technology and carry out local privacy conversion, and the Bit String after being then based on conversion calculates two dimension
Probability distribution and calculating multidimensional property dependency structure, finally according to probability distribution and the synthesis of dependence function and publication Multidimensional Awareness number
According to, specifically includes the following steps:
1) perception data local privacy is converted: giving the d dimension perception data D of N number of user, the data of each user are in user
End is directly hashed into Bit String, is then overturn at random to Bit String, the Bit String after obtaining secret protection, and be sent to
Server end, specifically includes the following steps:
1-1) Hash mapping: to the d dimension data of each userIn each attribute value breathed out
Uncommon conversion, thus by each data valueBeing converted to length is mjBit StringWherein i=1,2 ... N, j=1,2,
...d;
1-2) random overturning: based on Randomized response technology to each Bit StringEach bit turned at random
Turn, to provide local secret protection;
1-3) Bit String connects: it carries out after overturning at random, all Bit Strings of each dimension is attached, thus
To a dmjThe bit vectors of position, and it is sent to server end;
2) it calculates dimensional probability distribution: after the Bit String after getting the secret protection of server end, being returned using Lasso
The dimensional probability distribution of reduction method calculating initial data;
3) it calculates multidimensional property dependency structure: being based on dimensional probability distribution, calculate the Pearson correlation coefficients of multidimensional property
The dependence of multidimensional property is described, to establish the dependency structure model between multidimensional property;
4) it is sampled and is synthesized based on contiguous function: based on inverse cumulative distribution and dependency structure, being connected using multivariate Gaussian
Function sampling and synthesis Multidimensional Awareness data are connect, the publication data set with secret protection is obtained
A further improvement of the present invention lies in that in step 1) after user terminal carries out local privacy conversion, as gunz
Each participating user provides local difference secret protection in sensory perceptual system, and guaranteeing subsequent step all is based on secret protection
It is carried out in data afterwards, the sensitive data of user will not be revealed, and for d- dimension data, step 1) obtains difference privacy
The expression formula of protection is
Wherein, ε indicates difference secret protection degree, and h is hash function number, and f is overturning probability.
A further improvement of the present invention lies in that step 1-1) in Hash mapping concrete operations are as follows: to every dimension attribute AjMake
Use hash functionBy raw valueBeing converted to length is mjBit StringExpression formula be
A further improvement of the present invention lies in that step 1-2) in Bit StringCarry out the expression of random overturning rule
Formula is
Wherein, f ∈ (0,1) is used to control the probability overturn at random, thus specified local secret protection degree.
A further improvement of the present invention lies in that step 1-3) in carry out Bit String connection concrete operations be: by all categories
The Bit String of property is connected in turn, and obtains a dmjThe bit vectors of position, the expression formula of the vector are
A further improvement of the present invention lies in that calculating the concrete operations of dimensional probability distribution in step 2) are as follows: set step 1)
In to two dimensional attributes (Ak,Av) hash function that uses is respectivelyWithThe expression formula for then calculating Bit String Candidate Set is
Wherein, ΩkAnd ΩvIt is attribute (Ak,Av) codomain;
It is returned based on Lasso and calculates two dimensional attributes (Ak,Av) the expression formula of dimensional probability distribution be
Wherein,It is attribute (Ak,Av) true bit count and group
At vector.
A further improvement of the present invention lies in that the concrete operations of step 3) are, attribute (A two-by-two is calculatedk,Av) (k, v=
1 ... Pearson's coefficient d)Finally obtain the dependency structure R of d- dimension attribute.
A further improvement of the present invention lies in that the concrete operations of step 4) are, successively d- dimension attribute is sampled and closed
At, and then obtain the generated data collection that can externally issueFor
Wherein, (X1,X2,...,Xd)∈[0,1]N×dIt is to meet multivariate Gaussian contiguous function d- dimension random vector, implies
Dependence between d- dimension attribute, (F1 -1,F2 -1,...,Fd -1) be d- dimension attribute inverse cumulative distribution function.
The present invention has following beneficial technical effect:
Multidimensional data local method for secret protection in intelligent perception system of the present invention based on contiguous function, passes through
Local privacy conversion process realizes the local difference Privacy Privacy protection of user terminal, and the sensitive letter of participating user has been effectively ensured
Breath, and attribute Value space is reduced using contiguous function, the dimension-reduction treatment to multidimensional data is realized, to greatly avoid
" dimension disaster ", and improve the utility of publication data.By theory analysis and experimental analysis, experimental result is all confirmed
The present invention is largely effective in secret protection and in terms of guaranteeing data utility, and the present invention is in the compromise side of privacy and effectiveness
Face is better than other multidimensional data difference method for secret protection.
Detailed description of the invention
Fig. 1 is the multidimensional data local method for secret protection process schematic based on contiguous function;
Fig. 2 is the average deviation distance versus figure tested under different secret protection degree for different data collection;
Fig. 3 is the lower average deviation distance versus figure tested for different data collection of different probability distribution estimation;
Fig. 4 is that the average support vector cassification tested under different secret protection degree for different data collection is accurate
Rate comparison diagram.
Specific embodiment
Present invention is further described in detail with reference to the accompanying drawing.
Multidimensional data local secret protection with reference to Fig. 1, in intelligent perception system provided by the invention based on contiguous function
Method, specifically includes the following steps:
The conversion of Step1 perception data local privacy: giving the d dimension perception data D of N number of user, and the data of each user exist
User terminal is directly hashed into Bit String, is then overturn at random to Bit String, the Bit String after obtaining secret protection, concurrently
Server end is given, the expression formula for obtaining difference secret protection is
Wherein, ε indicates difference secret protection degree, and h is hash function number, and f is overturning probability;
Step1 specifically includes the following steps:
Step1-1 Hash mapping: to the d dimension data of each userIn each attribute value into
Row Hash translation, thus by each data valueBeing converted to length is mjBit StringWherein i=1,2 ... N, j=1,
2 ... d, to every dimension attribute AjUse hash functionBy raw valueBeing converted to length is mjBit StringTable
It is up to formula
Step1-2 is overturn at random: based on Randomized response technology to each Bit StringEach bit carry out it is random
Overturning, to Bit StringCarrying out the regular expression formula of random overturning is
Wherein, f ∈ (0,1) is used to control the probability overturn at random, thus specified local secret protection degree;
The connection of Step1-3 Bit String: carrying out after overturning at random, all Bit Strings of each dimension be attached, from
And obtain a dmjThe bit vectors of position, the expression formula of the vector are
And the vector is sent to server end.
Step2 calculates dimensional probability distribution: after the Bit String after getting the secret protection of server end, utilizing
Lasso regression algorithm calculates the dimensional probability distribution of initial data, if to two dimensional attributes (Ak,Av) use hash function difference
ForWithThe expression formula for then calculating Bit String Candidate Set is
Wherein, ΩkAnd ΩvIt is attribute (Ak,Av) codomain;
It is returned based on Lasso and calculates two dimensional attributes (Ak,Av) the expression formula of dimensional probability distribution be
Wherein,It is attribute (Ak,Av) true bit count and group
At vector.
Step3 computation attribute dependency structure: it is based on dimensional probability distribution, calculates the attribute (A two-by-two of multidimensional propertyk,Av)
(k, v=1 ... Pearson's coefficient d)Finally obtain the dependency structure model R of d- dimension attribute.
Step4 is based on contiguous function and is sampled and synthesized: based on inverse cumulative distribution and dependency structure, utilizing multivariate Gaussian
Contiguous function sampling and synthesis Multidimensional Awareness data, are successively sampled and are synthesized to d- dimension attribute, and then obtaining can be external
The generated data collection of publicationFor
Wherein, (X1,X2,...,Xd)∈[0,1]N×dIt is to meet multivariate Gaussian contiguous function d- dimension random vector, implies
Dependence between d- dimension attribute, (F1 -1,F2 -1,...,Fd -1) be d- dimension attribute inverse cumulative distribution function.
With reference to Fig. 2, carried out on five data sets of Retail, Kosarak, USCensus, Adult and TPC-E averagely partially
Gap overturns probability variation range and is set as 0.1~0.9 from experiment, reduces multidimensional number by using contiguous function in the present invention
According to Value space, therefore variation of mean rate distance is universal smaller, thus has preferable data utility;It can be seen with reference to Fig. 2
Out, when overturning probability increase, i.e., when secret protection degree increases, the change trend of average deviation distance of the invention is more slow
Slowly, demonstrate again that the present invention while guaranteeing privacy, can effectively improve the utility of publication data.
With reference to Fig. 3, carried out on five data sets of Retail, Kosarak, USCensus, Adult and TPC-E averagely partially
Gap overturns probability and is set as 0.5 from experiment, is respectively compared the average deviation distance of 1- dimension, 2- peacekeeping 3- dimension probability distribution, can
To find out that the probability distribution of the present invention on different data sets all has lower average deviation distance, i.e., preferable data effectiveness
Property;In addition, can be seen that with reference to Fig. 3 when the increase of the dimension of probability distribution, average deviation distance opposite will increase, but still
So within tolerance interval, demonstrate again that the privacy information of participation can be not only effectively ensured in the present invention, but also can guarantee
Issue the utility of data.
With reference to Fig. 4, average branch is carried out on five data sets of Retail, Kosarak, USCensus, Adult and TPC-E
The experiment of vector machine classification accuracy is held, overturning probability variation range is set as 0.1~0.9, it can be seen that the present invention is in different numbers
According to all having preferable classification accuracy on collection, it can guarantee the validity of publication data;In addition, can be seen that with reference to Fig. 4
When overturning probability increase, i.e., secret protection degree enhance when, the classification accuracy of the present invention on different data sets under
Drop, but still higher classification accuracy, and downward trend is very gentle, it was demonstrated that the present invention can be effectively ensured publication data and exist
Availability in machine learning.
Claims (8)
1. the multidimensional data local method for secret protection in intelligent perception system based on contiguous function, which is characterized in that right first
The data of each user are all based on Randomized response technology and carry out local privacy conversion, the bit after being then based on conversion in system
String calculating dimensional probability distribution and calculating multidimensional property dependency structure, finally synthesize and issue with function is relied on according to probability distribution
Multidimensional Awareness data, specifically includes the following steps:
1) perception data local privacy is converted: giving the d dimension perception data D of N number of user, the data of each user are straight in user terminal
It connects and is hashed into Bit String, then Bit String is overturn at random, the Bit String after obtaining secret protection, and be sent to service
Device end, specifically includes the following steps:
1-1) Hash mapping: to the d dimension data of each userIn each attribute value carry out Hash turn
It changes, thus by each data valueBeing converted to length is mjBit StringWherein i=1,2 ... N, j=1,2 ... d;
1-2) random overturning: based on Randomized response technology to each Bit StringEach bit overturn at random, from
And provide local secret protection;
1-3) Bit String connects: carrying out after overturning at random, all Bit Strings of each dimension is attached, to obtain one
A dmjThe bit vectors of position, and it is sent to server end;
2) it calculates dimensional probability distribution: after the Bit String after getting the secret protection of server end, being returned and calculated using Lasso
The dimensional probability distribution of method calculating initial data;
3) it calculates multidimensional property dependency structure: being based on dimensional probability distribution, calculate the Pearson correlation coefficients of multidimensional property to retouch
The dependence of multidimensional property is stated, to establish the dependency structure model between multidimensional property;
4) it is sampled and is synthesized based on contiguous function: based on inverse cumulative distribution and dependency structure, connecting letter using multivariate Gaussian
Number sampling and synthesis Multidimensional Awareness data, obtain the publication data set with secret protection
2. the multidimensional data local secret protection side in intelligent perception system according to claim 1 based on contiguous function
Method, which is characterized in that in step 1) after user terminal carries out local privacy conversion, each is joined as in intelligent perception system
Local difference secret protection is provided with user, and guaranteeing subsequent step all is carried out on based on the data after secret protection, no
The sensitive data of user can be revealed, and for d- dimension data, the expression formula that step 1) obtains difference secret protection is
Wherein, ε indicates difference secret protection degree, and h is hash function number, and f is overturning probability.
3. the multidimensional data local secret protection side in intelligent perception system according to claim 2 based on contiguous function
Method, which is characterized in that step 1-1) in Hash mapping concrete operations are as follows: to every dimension attribute AjUse hash functionIt will be former
Beginning data valueBeing converted to length is mjBit StringExpression formula be
4. the multidimensional data local secret protection side in intelligent perception system according to claim 3 based on contiguous function
Method, which is characterized in that step 1-2) in Bit StringCarrying out the regular expression formula of random overturning is
Wherein, f ∈ (0,1) is used to control the probability overturn at random, thus specified local secret protection degree.
5. the multidimensional data local secret protection side in intelligent perception system according to claim 4 based on contiguous function
Method, which is characterized in that step 1-3) in carry out Bit String connection concrete operations be: the Bit String of all properties is sequentially connected
Get up, obtains a dmjThe bit vectors of position, the expression formula of the vector are
6. the multidimensional data local secret protection side in intelligent perception system according to claim 5 based on contiguous function
Method, which is characterized in that the concrete operations of dimensional probability distribution are calculated in step 2) are as follows: set in step 1) to two dimensional attributes (Ak,Av)
The hash function used is respectivelyWithThe expression formula for then calculating Bit String Candidate Set is
Wherein, ΩkAnd ΩvIt is attribute (Ak,Av) codomain;
It is returned based on Lasso and calculates two dimensional attributes (Ak,Av) the expression formula of dimensional probability distribution be
Wherein,It is attribute (Ak,Av) true bit count and composition
Vector.
7. the multidimensional data local secret protection side in intelligent perception system according to claim 6 based on contiguous function
Method, which is characterized in that the concrete operations of step 3) are to calculate attribute (A two-by-twok,Av) (k, v=1 ... Pearson's coefficient d)Finally obtain the dependency structure R of d- dimension attribute.
8. the multidimensional data local secret protection side in intelligent perception system according to claim 7 based on contiguous function
Method, which is characterized in that the concrete operations of step 4) are that successively d- dimension attribute is sampled and synthesized, and then obtaining can be right
The generated data collection of outer publicationFor
Wherein, (X1,X2,...,Xd)∈[0,1]N×dIt is to meet multivariate Gaussian contiguous function d- dimension random vector, implies d- dimension
Dependence between attribute, (F1 -1,F2 -1,...,Fd -1) be d- dimension attribute inverse cumulative distribution function.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811301086.6A CN109508559B (en) | 2018-11-02 | 2018-11-02 | Multi-dimensional data local privacy protection method based on connection function in crowd sensing system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811301086.6A CN109508559B (en) | 2018-11-02 | 2018-11-02 | Multi-dimensional data local privacy protection method based on connection function in crowd sensing system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109508559A true CN109508559A (en) | 2019-03-22 |
CN109508559B CN109508559B (en) | 2020-10-27 |
Family
ID=65747446
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811301086.6A Active CN109508559B (en) | 2018-11-02 | 2018-11-02 | Multi-dimensional data local privacy protection method based on connection function in crowd sensing system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109508559B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110309671A (en) * | 2019-06-26 | 2019-10-08 | 复旦大学 | General data based on random challenge technology issues method for secret protection |
CN111144888A (en) * | 2019-12-24 | 2020-05-12 | 安徽大学 | Mobile crowd sensing task allocation method with differential privacy protection function |
CN112016123A (en) * | 2020-09-04 | 2020-12-01 | 支付宝(杭州)信息技术有限公司 | Verification method and device of privacy protection algorithm and electronic equipment |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9032072B2 (en) * | 2012-08-08 | 2015-05-12 | Empire Technology Development Llc | Real-time compressive data collection for cloud monitoring |
CN107493268A (en) * | 2017-07-27 | 2017-12-19 | 华中科技大学 | A kind of difference method for secret protection based on front position vector |
CN108563962A (en) * | 2018-05-03 | 2018-09-21 | 桂林电子科技大学 | A kind of difference method for secret protection based on spatial position service |
-
2018
- 2018-11-02 CN CN201811301086.6A patent/CN109508559B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9032072B2 (en) * | 2012-08-08 | 2015-05-12 | Empire Technology Development Llc | Real-time compressive data collection for cloud monitoring |
CN107493268A (en) * | 2017-07-27 | 2017-12-19 | 华中科技大学 | A kind of difference method for secret protection based on front position vector |
CN108563962A (en) * | 2018-05-03 | 2018-09-21 | 桂林电子科技大学 | A kind of difference method for secret protection based on spatial position service |
Non-Patent Citations (1)
Title |
---|
刘雅辉等: "大数据时代的个人隐私保护", 《计算机研究与发展》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110309671A (en) * | 2019-06-26 | 2019-10-08 | 复旦大学 | General data based on random challenge technology issues method for secret protection |
CN111144888A (en) * | 2019-12-24 | 2020-05-12 | 安徽大学 | Mobile crowd sensing task allocation method with differential privacy protection function |
CN111144888B (en) * | 2019-12-24 | 2022-08-02 | 安徽大学 | Mobile crowd sensing task allocation method with differential privacy protection function |
CN112016123A (en) * | 2020-09-04 | 2020-12-01 | 支付宝(杭州)信息技术有限公司 | Verification method and device of privacy protection algorithm and electronic equipment |
Also Published As
Publication number | Publication date |
---|---|
CN109508559B (en) | 2020-10-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109508559A (en) | Multidimensional data local method for secret protection in intelligent perception system based on contiguous function | |
Bai et al. | Alternating optimization of sensing matrix and sparsifying dictionary for compressed sensing | |
CN111988277A (en) | Attack detection method based on bidirectional generation counternetwork | |
Li et al. | A survey on federated learning | |
CN104835153B (en) | Non-rigid surface's alignment schemes based on rarefaction representation | |
Yang et al. | Avoid-df: Audio-visual joint learning for detecting deepfake | |
CN109614853A (en) | It is a kind of based on body structure divide bilinearity pedestrian identify network establishing method again | |
CN109299436A (en) | A kind of ordering of optimization preference method of data capture meeting local difference privacy | |
Hinz et al. | Metric and spectral triples for Dirichlet and resistance forms | |
CN110378141A (en) | Based on Bayesian network higher-dimension perception data local difference secret protection dissemination method | |
CN104463148A (en) | Human face recognition method based on image reconstruction and Hash algorithm | |
Xu et al. | Domain disentangled generative adversarial network for zero-shot sketch-based 3d shape retrieval | |
He et al. | Multi-attribute data recovery in wireless sensor networks with joint sparsity and low-rank constraints based on tensor completion | |
Ying et al. | Ear recognition based on weighted wavelet transform and DCT | |
Zhang et al. | Attribute-guided collaborative learning for partial person re-identification | |
CN116091260B (en) | Cross-domain entity identity association method and system based on Hub-node | |
Chen et al. | Incremental general non-negative matrix factorization without dimension matching constraints | |
Wang et al. | CCA-Net: A Lightweight Network Using Criss-Cross Attention for CSI Feedback | |
CN109525598A (en) | A kind of fault-tolerant compression method of wireless sense network depth and system based on variation mixing | |
Zhao et al. | Analysis and application of martial arts video image based on fuzzy clustering algorithm | |
CN108846262A (en) | The method that RNA secondary structure distance based on DFT calculates phylogenetic tree construction | |
Ma et al. | A syncretic representation for image classification and face recognition | |
An et al. | Low-resolution face recognition and sports training action analysis based on wireless sensors | |
Yao et al. | Privacy-preserving Judgment of the Intersection for Convex Polygons. | |
Sun et al. | Network Latency Estimation: A Tensor-based Weighted Singular Value Thresholding Method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |