CN104052750A - Method for protecting privacy in convex quadratic programming outsourcing on basis of affine transformation - Google Patents

Abstract

The invention provides a method for protecting privacy in convex quadratic programming outsourcing on the basis of affine transformation and belongs to the field of calculation outsourcing. The method includes that steps that (1) a user side inputs calculation parameters; (2) encrypted calculation parameters are formed by randomly generating a matrix and hiding parameter information through affine transformation; (3) the encrypted calculation parameters are transmitted to an outsourcing side; (4) the outsourcing side receives the encrypted calculation parameters transmitted by the user side, a solver is selected for calculation, and a calculation result is returned to the user side; (5) the user side receives the calculation result returned by the outsourcing side; (6) an optimal solution is obtained by decrypting the calculation result, and calculation is finished.

Description

A kind of method based on privacy in the outsourcing of affine transformation protection convex quadratic programming

Technical field

The invention belongs to and calculate outsourcing field, be specifically related to a kind of method based on privacy in the outsourcing of affine transformation protection convex quadratic programming, ensure user privacy information safety in outsourcing calculation task.

Background technology

Along with the high speed development of public cloud platform technology, public cloud can provide for user computing capability and the memory space of magnanimity, its advanced Intel Virtualization Technology and elasticity computing capability can effectively reduce user's construction cost and O&M cost simultaneously, and a set of perfect operation mode of paying is as required provided.Increasing user selects the program of local runtime and storage information, transfers to public cloud platform from this locality, effectively reduces commercial cost, has promoted greatly user job efficiency.

But affecting the extensive main factor of promoting of public cloud platform is safety problem, and user after cloud platform, has lost the ability to supervise to own resource by the resource outsourcing trustship of oneself to a certain extent.Due to the Intel Virtualization Technology of cloud computing, user cannot learn that its resource operates in the concrete physical machine of which platform, in this physical machine, also moving which other user's resource, the keeper of the public cloud platform sensitive information directly checked user etc. of whether having the ability.For these misgivings, a lot of users hold wait-and-see attitude for the use of public cloud platform, especially privacy are required to higher user, may abandon the use to public cloud platform just, and this can affect the universal of public cloud platform to a certain extent.

Do not carry out conservation treatment outsourcing account form, as shown in Figure 1.Traditional outsourcing is calculated by the administration means such as route of transmission and confidentiality agreement of strict control information; the safety of protection cleartext information; in transfer of data, do not carry out corresponding secrecy processing, this secret protection mode is not also suitable for the emerging outsourcing computing platforms such as public cloud platform.

Summary of the invention

The object of the invention is to solve the difficult problem existing in above-mentioned prior art, a kind of method based on privacy in the outsourcing of affine transformation protection convex quadratic programming is provided, utilize the random invertible matrix that generates as supplementary mode, a kind of mechanism of hiding convex quadratic programming privacy information is provided, user utilizes this mechanism to want Solve problems hidingization of sensitive information and to keep the canonical form of problem constant, the effective solution that ensures convex quadratic programming outsourcing computationally secure is provided, simply realize user outsourcing calculate in sensitive information hide, thereby meet user's low cost and utilize the demand for security of the outsourcing platforms such as cloud platform.

The present invention is achieved by the following technical solutions:

Based on a method for privacy in the outsourcing of affine transformation protection convex quadratic programming, comprising:

(1) user side input calculating parameter;

(2) hide parameter information by random generator matrix and affine transformation, form the calculating parameter after encrypting;

(3) calculating parameter after described encryption is sent to outsourcing end;

(4) calculating parameter after the encryption that outsourcing termination receipts user side transmits, chooses solution musical instruments used in a Buddhist or Taoist mass and calculates and result of calculation is returned to user side;

(5) user side receives the result of calculation that outsourcing end returns;

(6) decipher described result of calculation and obtain optimal solution, complete calculating.

Described step (2) is to utilize formula below to realize:

$\left\{\begin{array}{c}{Q}^{\′}=H^T\mathrm{QH}\\ c^{\′}={(c^{T}h-r^T\mathrm{QH})}^T\\ E^{\′}=\mathrm{NEH}\\ d^{\′}=N(d+\mathrm{Er})\\ A^{\′}=(A-\mathrm{ME})H\\ b^{\′}=b-\mathrm{Md}+(A-\mathrm{ME})r\end{array}\right.$

Wherein, N, M are the random matrix generating for hiding parameter information, are the parts of key; Ф is the calculating parameter of user side input in step (1), Ф=(Q, E, A, c, d, b), wherein A is m × n matrix, and E is k × n matrix, and Q matrix is n dimension symmetry square matrix, c is n × 1 vector, and b is m × 1 vector, and d is k × 1 vector, is the parameter of convex quadratic programming problem; H is in affine transformation, generates at random n dimension nonsingular square matrix, and r is in affine transformation, the random n dimensional vector generating, according to operational formula above, obtain calculating parameter (new argument after hiding) the Ф k after encrypting, Ф k=(Q ', E ', A ', c ', d ', b '), utilize braces formula just to obtain the parameter encryption from user's input parameter.

In described step (4), choose solution musical instruments used in a Buddhist or Taoist mass and refer to the solution musical instruments used in a Buddhist or Taoist mass of choosing for convex quadratic programming, separate musical instruments used in a Buddhist or Taoist mass and calculate according to the calculating parameter Ф k after encrypting, obtain result of calculation y.

Described step (6) is achieved in that

By result of calculation y, by formula x=Hy-r, obtain the result x after deciphering, complete calculating.

Compared with prior art, the invention has the beneficial effects as follows:

(1) solution that this method provides the convex quadratic programming outsourcing of perfect protection user privacy information to calculate for user; the enforcement of this scheme only need to be carried out simple encryption and decryption preliminary treatment at user side; do not need to consume too much local computing resource; ensure that the calculating parameter that is transferred to subcontractor is without practical significance; and the result of calculating does not have actual value yet, need to after user side deciphering, could obtain final result.Than original mode comparison of carrying out privacy protecting by treaty and faith mechanism, be more conducive to the popularization of the emerging outsourcing platforms such as public cloud platform.

(2) this method has only increased the privacy processing of encryption and decryption in user's side, do not change the canonical form that convex quadratic programming solves, namely do not change the algorithm solution musical instruments used in a Buddhist or Taoist mass of outsourcing computing platform, enforcement of the present invention does not need to transform on a large scale, there is good inheritance, outsourcing platform only need to be paid close attention to and separate the performance optimization of musical instruments used in a Buddhist or Taoist mass and without paying close attention to user's cipher mode, this programme has good isolation.

Brief description of the drawings

Fig. 1 is the outsourcing scheme that does not carry out conservation treatment.

Fig. 2 is the solution that the present invention hides privacy information.

Fig. 3 is the step block diagram of the inventive method.

Embodiment

Below in conjunction with accompanying drawing, the present invention is described in further detail:

User is in the time solving the science computational problems such as convex quadratic programming at present, because the problem in reality is comparatively complicated, scale is conventionally larger, exceed the disposal ability of PC, need user to consume a large amount of funds and build and calculate cluster and need special messenger to be responsible for the normal operation of service equipment, universal along with public cloud platform, the proposition of the theory of paying as required, the outsourcing of science calculation task has larger market potential, and user's personal secrets are one of numerous problems that need solution.The guard method that the present invention is based on the convex quadratic programming outsourcing calculating of affine transformation just provides this safety protecting mechanism; by affine transformation, the coordinate system of former problem is transformed in another territory; and utilize the random invertible matrix generating that coefficient information is hidden, realized coefficient during outsourcing is calculated and the crypticity of result.

The convex quadratic programming outsourcing Solve problems that the present invention is based on affine transformation is a kind of processing mode of the privacy information of user's sensitivity being hidden by the mode of affine transformation.The mode of outsourcing computing problem is often according to the canonical form of Solve problems, the solver interface of corresponding problem is provided, user only need be uploaded to service provider by parameter required processing problem form according to the rules, and calculating finishes rear service provider result is returned to user.When affine transformation technical finesse convex quadratic programming problem, the domain of definition of former problem and Value space can be mapped to new space, and the form of problem remains unchanged substantially, do not need the solver of service provider to make any change, reach and can use the demand of calculating outsourcing, can ensure that again the parameter that outsourcing service provider obtains has been without any practical significance, realize in parameter transmitting procedure and the secret and safe of operation result.

Convex quadratic programming outsourcing account form based on affine transformation Information hiding, as shown in Figure 2, user processes by calculative problem is carried out to hidingization of sensitive information on local backstage, being sent to outsourcing calculation services business calculates again, the parameter that outsourcing service provider obtains has like this been the information without any practical significance, but can directly apply its general solution musical instruments used in a Buddhist or Taoist mass program, the result of calculation obtaining is not final result of calculation yet, resolve this locality that need to be transferred to user side, obtain final result of calculation, the privacy information that ensures user is all hidden in user side processing in addition.

Convex quadratic programming outsourcing Solve problems based on affine transformation is a kind of processing mode of the privacy information of user's sensitivity being hidden by the mode of affine transformation.The mode of outsourcing computing problem is often according to the canonical form of Solve problems, the solver interface of corresponding problem is provided, user only need be uploaded to service provider by parameter required processing problem form according to the rules, and calculating finishes rear service provider result is returned to user.When affine transformation technical finesse convex quadratic programming problem, the domain of definition of former problem and Value space can be mapped to new space, and the form of problem remains unchanged substantially, do not need the solver of service provider to make any change, reach to use and calculated the demand of outsourcing, can ensure again in parameter transmitting procedure and the secret and safe of operation result.

In sum, when convex quadratic programming problem quantity is a lot of or solve when huge, often need a large amount of high performance equipment, by problem solving outsourcing can effectively be reduced to the requirement of user to local device performance, and by corresponding processing means such as affine transformations, user privacy information is hidden, make user can relieved use public cloud platform etc. outsourcing service platform.

One, system enciphering and deciphering algorithm design

General, quadratic programming can be described as the optimization problem of following canonical form:

$f\left(x\right)=\left(\frac{1}{2}\right)x^T\mathrm{Qx}+c^{T}x$

min

s.t.Ax≤b

Ex＝d

Wherein, A is m × n matrix, and E is k × n matrix, and under normal circumstances, A, E matrix are row full rank; X, c are n × 1 vector, and b is m × 1 vector, and d is k × 1 vector.The Q matrix of target function f (x) is n dimension symmetry square matrix, and in the time that Q is positive semidefinite matrix, matrix exgenvalue is all non-negative, and f (x) is convex function, and optimization problem is convex quadratic programming; In the time that Q is positive definite matrix, matrix exgenvalue is all positive number, and f (x) is strict convex function, and optimization problem is strict convex quadratic programming.

Affine transformation is the conversion of shining upon one by one, and convexity is constant under affine transformation, and original function is convex function, and the function after conversion is still convex function.In affine transformation, generate at random n dimension nonsingular square matrix H, n dimensional vector r, makes:

$y=H^{-1}(x+r)\⇔x=\mathrm{Hy}-r$

Like this, the standard type of quadratic programming problem can be converted into

$g\left(y\right)=f(\mathrm{Hy}-r)=\left(\frac{1}{2}\right){(\mathrm{Hy}-r)}^{T}Q(\mathrm{Hy}-r)+c^T(\mathrm{Hy}-r)$

min

s.t.A(Hy-r)≤b

E(Hy-r)＝d

Further arrange and obtain:

$g=\left(y\right)=\left(\frac{1}{2}\right)y^T\left(H^T\mathrm{QH}\right)y+(c^{T}H-r^T\mathrm{QH})y+\frac{1}{2}{r}^T\mathrm{Qr}-c^{T}r$

min

By random generation invertible matrix, the coefficient information in constraints to be hidden, final arrangement encrypted form is out as follows:

$g\left(y\right)=\left(\frac{1}{2}\right)y^T\left(H^T\mathrm{QH}\right)y+(c^{T}H-r^T\mathrm{QH})y$

min

s.t.NEHy＝N(d+Er)

(A-ME)Hy≤b-Md+(A-ME)r

If $\left\{\begin{array}{c}{Q}^{\′}=H^T\mathrm{QH}\\ c^{\′}={(c^{T}h-r^T\mathrm{QH})}^T\\ E^{\′}=\mathrm{NEH}\\ d^{\′}=N(d+\mathrm{Er})\\ A^{\′}=(A-\mathrm{ME})H\\ b^{\′}=b-\mathrm{Md}+(A-\mathrm{ME})r\end{array}\right.$ Former problem is converted into

$g\left(y\right)=\left(\frac{1}{2}\right)y^T{Q}^{\′}y+c^{\′T}y$

min

s.t.E′y＝d′

A′y≤b′

In sum, key K is made up of H, M, N, r, i.e. K=(H, M, N, r), and Ф=(Q, E, A, c, d, b), Ф k=(Q ', E ', A ', c ', d ', b ').Because key produces by random, security intensity is limited, so each calculating should regenerating key, guarantees computationally secure.

When deciphering, by the result that obtains calculating, by the inverse transformation of affine transformation, can obtain final optimal solution:

$y=H^{-1}(x+r)\⇔x=\mathrm{Hy}-r$

Two, system flow

The schematic flow sheet of the inventive method as shown in Figure 3, comprising:

User side input calculating parameter;

Hide parameter information by random generator matrix and affine transformation: by above formula, can learn, the parameter of user's input is Ф, and the random matrix generating is H, according to the operational formula of braces, obtained the new argument Ф k after hiding.；

Calculating parameter after encrypting is sent to outsourcing end;

Outsourcing termination is received the calculation task that user side transmits, and chooses solution musical instruments used in a Buddhist or Taoist mass and (chooses solution musical instruments used in a Buddhist or Taoist mass and refer to, choose the solution musical instruments used in a Buddhist or Taoist mass for convex quadratic programming, due to its solutions comparison maturation, which kind of being chosen and determined by cloud platform service business, is the larger part of computing cost, therefore outer contracting out.Cloud service business may have the solution musical instruments used in a Buddhist or Taoist mass for different problems in a large number, not only for convex quadratic programming problem.And the application only pays close attention to the secret protection of convex quadratic programming problem.Whether the former problem of result substitution, see and satisfy condition, and can take a decision as to whether the result of raw parameter.) calculate and result is returned to client (this separates musical instruments used in a Buddhist or Taoist mass is general solution musical instruments used in a Buddhist or Taoist mass, is the service that subcontractor provides, and need not be concerned about and how to realize, and only needs to ensure that data are normal mutual.Server acquires after parameter Ф k, calculates result y, and returns to client);

User side receives the result of calculation that outsourcing end returns;

Deciphering result of calculation obtains optimal solution, completes calculating: by the result of calculation y returning, by formula x=Hy-r, obtain the result x after deciphering, complete calculating.

The present invention proposes to realize by affine transformation and the random mode that generates invertible matrix the secret protection solution that convex quadratic programming problem solves in outsourcing platform for the first time; entirety encryption and decryption scheme is simple; the in the situation that of one-time pad, bring enough safety guarantee.

The present invention has directly utilized the solution musical instruments used in a Buddhist or Taoist mass of public outsourcing platform, by the appropriate design to encryption and decryption in client, the form that has ensured canonical solution musical instruments used in a Buddhist or Taoist mass is constant, ensures public outsourcing platform (as the versatility of public cloud platform) versatility, is conducive to Rapid Popularization of the present invention

Technique scheme is one embodiment of the present invention, for those skilled in the art, the invention discloses on the basis of application process and principle, be easy to make various types of improvement or distortion, and be not limited only to the described method of the above-mentioned embodiment of the present invention, therefore previously described mode is just preferred, and does not have restrictive meaning.

Claims (4)

1. the method based on privacy in the outsourcing of affine transformation protection convex quadratic programming, is characterized in that: described method comprises:

(1) user side input calculating parameter;

(2) hide parameter information by random generator matrix and affine transformation, form the calculating parameter after encrypting;

(3) calculating parameter after described encryption is sent to outsourcing end;

(4) calculating parameter after the encryption that outsourcing termination receipts user side transmits, chooses solution musical instruments used in a Buddhist or Taoist mass and calculates and result of calculation is returned to client;

(5) user side receives the result of calculation that outsourcing end returns;

(6) decipher described result of calculation and obtain optimal solution, complete calculating.

2. the method based on privacy in the outsourcing of affine transformation protection convex quadratic programming according to claim 1, is characterized in that: described step (2) is to utilize formula below to realize:

$\left\{\begin{array}{c}{Q}^{\′}=H^T\mathrm{QH}\\ c^{\′}={(c^{T}h-r^T\mathrm{QH})}^T\\ E^{\′}=\mathrm{NEH}\\ d^{\′}=N(d+\mathrm{Er})\\ A^{\′}=(A-\mathrm{ME})H\\ b^{\′}=b-\mathrm{Md}+(A-\mathrm{ME})r\end{array}\right.$

Wherein, N, M are the random matrix generating for hiding parameter information, are the parts of key; Ф is the calculating parameter of user side input in step (1), Ф=(Q, E, A, c, d, b); H is in affine transformation, generates at random n dimension nonsingular square matrix, and r is in affine transformation, and the random n dimensional vector generating, according to operational formula above, has obtained the new argument Ф k after hiding, Ф k=(Q ', E ', A ', c ', d ', b ').

3. the method based on privacy in the outsourcing of affine transformation protection convex quadratic programming according to claim 2; it is characterized in that: in described step (4), choose solution musical instruments used in a Buddhist or Taoist mass and refer to the solution musical instruments used in a Buddhist or Taoist mass of choosing for convex quadratic programming; separate musical instruments used in a Buddhist or Taoist mass and calculate according to the calculating parameter Ф k after encrypting, obtain result of calculation y.

4. the method based on privacy in the outsourcing of affine transformation protection convex quadratic programming according to claim 3, is characterized in that: described step (6) is achieved in that

By result of calculation y, by formula x=Hy-r, obtain the result x after deciphering, complete calculating.