Indonesian J our nal of Electrical Engineering and Computer Science V ol. 16, No. 2, No v ember 2019, pp. 1065 1069 ISSN: 2502-4752, DOI: 10.11591/ijeecs.v16i2.pp1065-1069 r 1065 Pri v acy pr eser ving outsour cing algorithm f or tw o-point linear boundary v alue pr oblems Nedal M. Mohammed 1 , Laman R. Sultan 2 , Santosh S. Lomte 3 1,3 Department of Computer Science , Dr . Babasaheb Ambedkar Marathw ada Uni v ersity , Aurang abad, India 2 Department of Po wer Mechanics, Basra T echnical Institute, Southern T echnical Uni v ersity , Al-Basrah, Iraq Article Inf o Article history: Recei v ed Jan 25, 2019 Re vised Apr 15, 2019 Accepted May 3, 2019 K eyw ords: Cloud computing Secure computation outsourcing V erifiable computing Secure and pri v ac y BVP . ABSTRA CT One of a po werful application in the age of cloud computing is the outsourcing of sci- entific computations to cloud computing which mak es cloud computing a v ery po w- erful computing paradigm, where the customers with limited computing resource and storage de vices can outsource the sophisticated computation w orkloads into po wer - ful service pro viders. One of scientific computat ions problem is T w o-Point Boundary V alue Problems (BVP) is a basic engineering and scientific problem, which has appli- cation in v arious domains. In this paper , we propose a pri v ac y-preserving, v erifiable and ef ficient algorithm for tw o-point BVP in outsourcing paradigm. W e implement the proposed schema on the customer side laptop and using A WS compute domain elastic compute cloud (EC2) for the cloud side. Copyright c 2019 Institute of Advanced Engineering and Science . All rights r eserved. Corresponding A uthor: Nedal M. Mohammed, Department of Computer Science, T aiz Uni v ersity , T aiz, Y emen. Email: dr .nedal.mohammed@gmail.com 1. INTR ODUCTION The po werful adv antages of cloud computing is called outsourcing, where the customers with li mited computing resource and storage de vices can outsource the sophisticated comput ation w orkloads i nto po werful service pro viders. Despite the tremendous benefits, there are man y challenges and security concerns because the cloud serv er and customer are not in the same trusted domain, to a v oid these problems [1, 2, 3, 10]. First, to combat the security c o nc ern is applying encryption techniques to customer’ s sensiti v e information before outsourcing to the cloud b ut still, there is a challenge ho w mak es the task of computation o v er encrypted data [2, 4, 8]. Second, no guarantee from the cloud on the quality of the computed data and results. F ocusing on scientific and engineering applications problems we notice that the dif ferential equati ons problems (Boundary v alue problems and initial v alue problems). The BVP & IVP frequently appear in v arious fields and has a number of applications, b ut solving these lar ge-scale BVP & IVP problems i s usually computa tionally so e xpensi v e [5, 7, 9, 11]. The computers with limited computing resource and storage de vices are f acing the challenge of solving lar ge-scale BVP & IVP . T o address this challenge (solving lar ge-scale BVP & IVP), an alternati v e option is outsourcing to cloud computing. This w ork, focusing on secure outsourcing BVP . W e are proposing secure, ef ficient and v erifiable scheme to of fload lar ge-scale BVP computation to cloud side. This paper is or g anized as follo ws Section 2. describes our proposed system model t o secure outsourcing the BVP problem. In Section 3. outsourcing algorithm of tw o-point BVP . In Section 4. security analysis of proposed algorithm. In Section 5. we pro vide e xperimental result analysis for proposed schema. At last the w ork conclusion is presented in section 6. J ournal homepage: http://iaescor e .com/journals/inde x.php/ijeecs Evaluation Warning : The document was created with Spire.PDF for Python.
1066 r ISSN: 2502-4752 2. SYSTEM MODEL W e consider an asymmetric architecture in v olv ed tw o parties. On the first side, there is a resource- constrained customer who has the BVP problem to solv e Ho we v er , due to limited computing resources the customer cannot do this computation to solv e the problem locally . On second side, ( S ) cloud with computa- tionally po werful de vices and huge storage f acilities, b ut cannot be trusted with the sensiti v e information. Then to a v oid security problems the procedure goes as sho wn in Figure 1. Figure 1. System model of secure outsourcing of BVP problem. W e c on s ider the procedure is go wing as follo w the customer C outsource an BVP computation task : D ! M to a cloud serv er S. Ho we v er , S is not fully trusted by C either semi-honest or malicious. So to protect the input pri v ac y , the C transfers the original BVP to linear system equation (LSE) then encrypts into an encrypted k with a secret K e y K then. Then k is outscrced from the C to the S. The S runs optimization algorithm to solv e k when recei ving the encrypted BVP problem k . After getting the result, the C v erifies whether the solution returned is correct or not in encryption domain. If the result can not pass through v erification, C requires the cloud to compute it ag ain. Otherwise, the customer decrypting the correct result to get the solution to the BVP . 3. OUTSOURCING ALGORITHM OF TW O-POINT BOUND AR Y V ALUE PR OBLEMS In this section, we propose algorithm for securely outsourcing lar ge-scale system of finite dif ference method to find the solution of linear boundary v alue problems. T o achie v e goals of our w ork, we design the follo wing sub-algorithms. 3.1. Using Finite Differ ence Method T w o-Point linear boundary v alue problems formulated as[6, 5, 9]: u 00 = p ( t ) u 0 + q ( t ) u + r ( t ) ; t 2 [ a; b ] ; u ( a ) = ; u ( b ) = : (1) The problem for securely outsourcing lar ge-scale Linear BVP with Finite Dif ferences can be for - mulated as follo ws: The client C seeks for the solution to a lar ge-scale Linear Boundary V alue Problems A h y h = b h , where A h 2 R N N is gi v en matrix in the form [8, 9] A h = 0 B B B B @ 2 1 0 : : : 0 0 0 0 1 2 1 : : : 0 0 0 0 : : : : : : : : : : : : : : : : : : : : : : : : 0 0 0 : : : 0 1 2 1 0 0 0 : : : 0 2 1 0 1 C C C C A and b h 2 R N is a v ector of the form Indonesian J Elec Eng & Comp Sci, V ol. 16, No. 2, No v ember 2019 : 1065 1069 Evaluation Warning : The document was created with Spire.PDF for Python.
Indonesian J Elec Eng & Comp Sci ISSN: 2502-4752 r 1067 b h = 0 B B @ r ( t 1 ) + h 2 r ( t i ) ; i = 2 ; 3 ; : : : ; N 1 r ( t N ) + h 2 1 C C A 3.2. K ey Generation Algorithm. In k e y generation algorit hm, client firstly picks a random disguising coef ficient v ector r 2 R n and M ; N 2 R n n ; tw o random sparse matrices as belo w . M ( i ; f ) = m i m ( i ) f 1 i ; f m ; (2) N ( i ; f ) = n i n ( i ) f 1 i ; f n ; r ( i ) = r i 1 i n: Here m i ; n i ; r i are all randomly generated from the space and m ; n are tw o dif ferent bijection functions. These tw o k e y matrices and one v ector constitut e the k e y which can be written as K ( M ; N ; r ) and must be k ept secret by client. 3.3. Outsour cing Algorithm Due to the lack of computing resources, C could be infeasible to carry out such e xpensi v e computa tion as O ( n ) for 2 < 3 locally . Therefore, client will outsource the computation w orkloads to cloud serv er in a pay-per -use manner [11]. Client chooses M ; N 2 R nn tw o random sparse matrices which generate from 3.2. to hide A h then computes K = M A h N . So that the input of problem generation (ProbGen) algorithm is a c oef ficient matrix A h u h A 2 R n n and a coef ficient v ector b h 2 R n . From abo v e transformation the original matrix of finite dif ference method of BVP can be re wri tten as A ( x + r ) h = c . Then, C computes K = M AN and d = M c . W ithout loss of generality , we denote u = N 1 ( x + r ) , where N 1 is the in v erse of matrix N . Note that in our algorithm, no party needs to compute N 1 . It appears h e re only for representing the form of u . In f act, if N 1 had to be computed, the algorithm w ould no longer be ef ficient as the time and computational comple xities incurred by computing N 1 w ould be v ery undesirable. Note that K u h = M A h N pN 1 ( u h + r ) = M A ( u h + r ) = M c = d , then outsource this K ; d to the cloud and get th solution as a coef ficient v ector u h such t hat K h u h = d h . Using the random disguising technique, we can pro v e the pri v ac y of proposed outsourcing schema for A h ; b h and the solution x . Besides, The proposed schema only requires one round interaction between cloud and serv er and does not require an y special encryption techniques, so the comple xity to compute K is O ( n 2 ) . 4. SECURITY AN AL YSIS Theorem 1: In the fully malicious model, the algorithms (C, S) ar e privacy for A h ; b h , and u h . Pr oof . W e first pro v e the pri v ac y for input b h and output u h of B VP . Note that the adv ersary A h can only kno w K and d throughout the whole algorithm. Besides, we ha v e b h = M 1 d Ar ; and u h = N x r . Since r is a random blinding coef ficient v ector in R n , both b and u h are blinded by r in the sense of indistinguishability . W e then pro v e the pri v ac y for input A h . Let M = ( m ij ) ; N = ( n ij ) ; M 0 = ( m ij ) ; and N 0 = ( n 0 ij ) be four random nonsingular sparse matrices generated by C. Gi v en tw o nonsingular dense matrices A = ( aij ) and A 0 = ( a 0 ij ) which are chosen by the adv ersary A h , C computes K = M A h N = ( k ij ) and K 0 = M 0 A h 0 N 0 = ( k 0 ij ) , where k ij = n X i =1 n X j =1 m if a k f n if Privacy pr eserving outsour cing algorithm for two-point linear ... (Nedal M. Mohammed) Evaluation Warning : The document was created with Spire.PDF for Python.
1068 r ISSN: 2502-4752 and t 0 ij = n X i =1 n X j =1 m 0 if a 0 f i n 0 ij Note that the numerical v alue and position of all non-zero elements of four matrices M ; N ; M 0 and N 0 are randomly chosen by C, thus k ij and k 0 ij are computationally indistinguishable. As a result, the adv antage of A h to distinguish between K and K 0 is ne gligible. Theorem 2: In the fully malicious model , the algorithms (C, S) ar e an O ( 1 n ) ef ficient implementation of sc hema. Pr oof . In the propos ed algorithm, C needs to perform four matrix-v ector multiplication (we omit the v ector -addition operations), which tak es O ( n 2 ) computations. Besides, C also ne eds to compute K = M AN , which also tak es O ( n 2 ) computations. Thus, the ef ficient implementation of our algorithm (C, S) are an O ( 1 n ) . 5. EXPERIMENT AL RESUL T AN AL YSIS The e xperimental results are the a v erage of multiple trials. W e design numerical e xperiments to e v al- uate the ef ficienc y of the mechanism. W e implemented it using Matlab (2013a), with the system configuration is C P U I ntel R C or e TM i 3( C P U s ) ˜ 1 : 8 GH Z 4 GB R am on a laptop and Amazon Elastic Compute Cloud (EC2) cluster . The test benchmark for randomly generated sparse matrices only focuses on the lar ge- scale problems where n ranges from 5 ; 000 to 20 ; 000 ; as T able 1. T able 1. Performance of the proposed scheme Pr oblem K eyGen Pr obGen V erify Solv e Client Size Algorithm Algorithm Algorithm Algorithm Cost T ime 1 n = 5000 10.15 116.88 6.13 10.54 133.03 2 n = 10000 19.20 231.23 10.89 14.02 261.32 3 n = 15000 23.13 347.72 20.54 17.43 391.39 4 n = 20000 33.97 580.73 27.85 18.56 642.55 T o measure the ef ficienc y of our proposed mechanism, we simulated all four phases (i.e., K e yGen, ProbGen, V erify and solv e). The corresponding time costs for dif ferent size of problems are sho wn in Figure 2. 0.5 1 1.5 2 x 10 4 0 100 200 300 400 500 600 700 Problem Size Tim in (sec)     Cost of  ProbGen Cost of Cloud(slove) Cost of KeyGen Cost of Client Cost of Verify Figure 2. T ime cost for each phase of secure outsourcing BVP algorithm. Indonesian J Elec Eng & Comp Sci, V ol. 16, No. 2, No v ember 2019 : 1065 1069 Evaluation Warning : The document was created with Spire.PDF for Python.
Indonesian J Elec Eng & Comp Sci ISSN: 2502-4752 r 1069 6. CONCLUSION In this paper , we proposed a ne w secure, ef ficient algorithm for securely outsourcing of lar ge-s cale finite dif ference method to find the solution of linear BVP computation using cloud computing. W e implement schema on the customer side laptop and using A WS compute domain elastic compute cloud (EC2) for the cloud side. W e find that the proposed schema is suitable to approxi mate solutions to dif ferential equations (BVP) problem and only requires O ( n 2 ) computational o v erhead e v en in the fully malicious model. REFERENCES [1] C. Hu, A. Alhothaily , A. Alra w ais, X. Cheng, C. Sturti v ant and H. Liu, ”A secure and v erifiable out- sourcing scheme for matrix in v erse computation, IEEE INFOCOM’17 (Atlanta, GA, USA) , V ol. 4, pp. 2304–2312, 2017. [2] N. M. Mohammed and S. S. Lomte, ”Recent adv ances on secure computations outsourcing in cloud computing, Asian Journal of Mathematics and Computer Research , V ol. 24, pp.192–205, 2017. [3] K. Zhou and J. Ren, ”CASO: Cost-A w are Secure Outsourcing of General Computational Problems, IEEE T ransactions on Services Computing , 2018. [4] W . Shen, Y . Bo, C. Xianghui, C. Y u and S. Xuemin, ”A distrib uted secure outsourcing scheme for solving linear algebraic equations in ad hoc clouds, IEEE T ransactions on Cloud Computing , V ol. 4, 2017. [5] H. C. Sax ena, ”Finite-Dif ferences and Numerical Analysis, Thirteen Re vised Edition, Published by S. Chand& Compan y Ltd. Ne w Delhi , 1997. [6] U. Ascher , R. Mattheij and R. Russell, ”Numerical Solution of Boundary V alue Problems for Ordinary Dif ferential Equations, Prentice-Hall , 1988. [7] N. M. Mohamm ed and S. S. Lomte, ”Secure Computations Outsourcing of Mathematical Optimization and Linear Algebra T asks: Surv e y , International Journal for Research in Engineering Application & Management , pp. 6–11, 2019. [8] A. Geor ge and J. Liu, ”Computer Solution of Lar ge Sparse Positi v e-definite Systems, Prentice-Hall , 1981. [9] D. M. Y oung, ”Iterati v e Solution of Lar ge Linear Systems, Academic Press , 1971. [10] N. M. Mohammed and S. S. Lomte, ”Secure and Ef ficient Outsourcing of Lar ge Scale Linear Fractional Programming, International Conference on Computing in Engineering and T echnology(I CCET) , 2019, T o Appear . [11] S. Salinas, C. Luo, X. Chen, W . Liao and P . Li, ”Ef ficient Secure Outsourcing of Lar ge-scale Sparse Linear Systems of Equations, IEEE T ransactions on Big Data , 2017. DOI 10.1109/TBD A T A.2017.2679760, Privacy pr eserving outsour cing algorithm for two-point linear ... (Nedal M. Mohammed) Evaluation Warning : The document was created with Spire.PDF for Python.