• Tiada Hasil Ditemukan

Proposed Static Remote Data Auditing Method

Chapter 7: Conclusion

4.2 Proposed Static Remote Data Auditing Method

We propose a novel dynamic remote data auditing method to verify the correctness of outsourced data in the cloud computing. The proposed method incorporates the fea-tures of algebraic properties, which incurs low computation cost on clients and server and makes it suitable for various tasks in large-scale storage systems.

In traditional data auditing methods, the homomorphic cryptosystem is usually used to generate a tag and verify the integrity of the outsourced data remotely that considerable

Figure 4.1: Cloud Data Storage Audit Architecture

computation cost on the auditor. To address this issue, Static Remote Data Auditing (SRDA) method focuses on utilizing the algebraic signature for generating the tags and checking the correctness of the data in the cloud storage system.

The architecture of the SRDA method in cloud and mobile cloud computing domain comprises of four main components (Figure 4.1), as follows:

DO who outsources the data in the cloud through a mobile device, laptop or personal

computers. DO is able to check the integrity of the data and update the data in the cloud by using insert, append, delete operations.

CSP: is an entity with a large amount of space to store owner’s data. The CSP is

respon-sible for generating the response message to prove the possession of the outsourced data.

TPA: Because performing an auditing task introduces performance overheads at the DO

side, the audit task is preferably delegated to an entity called the TPA. The TPA is a trusted entity with expertise and capability to perform the auditing task on behalf of the DO. By employing a TPA, users (especially with limited computing power, storage and connectivity) are alleviated from the burden of expensive auditing tasks.

Although the TPA is considered as a trustworthy and reliable entity for data audit-ing, it may be curious at the same time. Therefore, one of the important issues in the data audit service in the presence of a TPA is to prevent data leakage during auditing and preserve the privacy of data.

User (A/B/C): is an entity (individual or enterprise) who is registered by the DO and

granted access to the outsourced data for reading the data.

Remote data auditing services follow response-challenge procedure in which (1) The DO, firstly, pre-processes her file, generates some metadata, and hands over metadata to TPA. At this point, DO is not required to be involved in the audit process anymore, (2) To check correctness of data in the cloud, TPA generates a random challenge message and sends it to the CSP (the DO is also able to generate the challenge message when the TPA is not supported by RDA service), (3) When Cloud storage receives the challenge, computes the corresponding response and sends it to the TPA, and (4) After receiving a response from CSP, the verification is carried out to find out whether or not the CSP has correctly stored the outsourced file. It is important to mention that in order to reduce the overhead of the audit process only a small fraction of whole file is queried.

In the rest of this section, a secure algorithm is presented to remotely audit out-sourced data in cloud computing based on the algebraic signature technique. the pro-posed remote data auditing method consists of the four phases. such as (1) Setup phase, (2) Challenge phase, (3) Response phase, and (4) Verification phase. The following sec-tions explain the aforementioned steps.

4.2.1 Setup Phase

The Data Owner (DO) firstly divides the input file F in to m blocks F[1], F[2], . . . , F[m].

Each data block F[i]is also divided into n sectors indicated as F[i] [1], F[i] [2], . . . , F[i][n]

by using the data fragment technique. If the number of sectors in a data block is less than n, the DO simply increases the number of sectors to n by appending the number of zero

as new sectors to the considered block. Afterward, the DO generates: (1) a secret key by executing the KeyGen algorithm, (2) a unique file id(f id)for each file by using FidGen algorithm, and (3) a unique tag(Ti)for each block of data file on the basis of the alge-braic signature technique by calling the TagGen algorithm. Finally, the DO sends the data blocks along with the corresponding tags and file ID(f id,F[i],Ti,Ci)mi=1 to the CSP and delete the local copy of the file and tags from the local storage.

KeyGen(1k)→(pk,sk).This probabilistic key generation algorithm takes a security parameter k as an input and produces a pair of public and private key. The private key includes a secret hash key skh to compute hash H(.)and a file id secret key skFid to sign a message. The public key is also used to verify the signature.

FidGen(F,skFid)→(f id).This algorithm creates a unique file id(f id)for each file (F). To achieve this goal, the DO (data owner) firstly generates a unique fingerprint (f print=Filename||m||n)for each file consisting of the file name(Fname), number of blocks (m), and number of sectors (n). After that, the fid is generated for each file by concatenating the fprint and its signature that is computed under skFid,as follows:

f id= f print||SignskFid(f print) (4.1)

TagGen(F,skh,skFid)→(T,C). This algorithm is used to compute a single tag for each block based on the algebraic signature of the considering block. The tag of each block (i)consists of two parts, as a block tag(Ti)and tag controller(Ci), that are

com-puted by using the formula:

Ti=Sγ(f[i]||Hskh(τ)) (4.2) Ci=nγ

n j=1

Hskh(τ).γj1 (4.3)

where, τ indicates the concatenation of fid and index of block(τ= f id||i), andγ is a non-zero elements in the Galois field for generating algebraic signature.

4.2.2 Challenge Phase

To verify the integrity of outsourced data, the auditor requires sending a challenge message to the server. The challenge message is composed of a subset of c random ele-ment of set[1,m]data blocks. The auditor also generates a random coefficient for each block to make sure that server possesses each one of the challenged blocks. The indices and coefficients ({csii}ci=1) are computed by using pseudo-random permutation (Im-pagliazzo, Levin, & Luby, 1989) keyed with a fresh randomly-chosen key for each chal-lenge message. It is because of preventing the server from anticipating the block indices.

Moreover, the pseudo-random permutation prevents server from keeping a combination of the data blocks instead of the original blocks. Finally, the auditor sends the challenge message(chal={csii}ci=1)to the server.

4.2.3 Response Phase

Upon receiving the challenge message(chal={csii}ci=1), the server computes the response message on the basis of the challenge message.The response message consists of two parts(µ,σ), as follows.

(i)µ part. The first part of response message (µ) is computed by linear combining of the selected c blocks of the challenge phase. In other words, µ contains n sectors

µ12,···µn that are computed by making XOR considered sectors of selected c blocks (Equation 4.4).

µj=

csc

i=cs

1

νi.f[i][j] (4.4)

For example, assume that the auditor selected blocks 2, 5, and 7 as a challenge message and sent them as a challenge message to the server. Therefore,µ involves n sectorsµ12,···µnof such three blocks that are computed by making XOR the sectors of 2th,5th, and 7th data blocks of the outsourced file in the cloud storage, as follows (for simplifying this example, the value of the coefficients(ν) are not considered).

µ =

























µ1= f[2][1]⊕f[5][1]⊕f[7][1]

µ2= f[2][2]⊕f[5][2]⊕f[7][2]

...

µn= f[2][n]⊕f[5][n]⊕f[7][n]

(4.5)

Figure 4.2 shows how the sectors of µ (µ12, ...,µn) are generated by using n sectors of three blocks F[2],F[5]and F[7].

(ii)σ. part The second part of the response message(σ)is calculated by aggregating the authenticator tags of the selected blocks in the received challenge by the following Equation.

σ=

csc

i=cs

1

νi.(TiCi) (4.6)

Finally, the server sends the response message(f id,µ,σ)to the verifier.

1 2 3 4 5 7 9 m

n

m Blocks

n Sectors

Figure 4.2: Linear combination of requested data blocks as a part of response message (µ).

4.2.4 Verification Phase

When the response message is received, the auditor firstly verifies the signature of the f id under DO’s public key (pk) and reject the message if the signature is invalid.

Otherwise, the auditor recovers the file name(f name), number of blocks(m), and number of sectors(n). Then, the auditor checks the integrity of the outsourced blocks by:

σ =?

n j=1

µjγj1 (4.7)

Figure 4.3 shows the process of our data auditing scheme for cloud computing.