The design of the Distributed Decentralized Storage Platform (DDSP) requires to develop an e cient data validation method taking into account the network scalability issue, possibility to check data without knowledge of content and privacy protection of users data. DDSP is aimed at ensuring the availability and integrity of data. The aim of this research is to develop a zero-knowledge data validation method for the decentralized storage system for minimizing data transferring to maintain the network scalability and minimizing computational cost on the side of a storage node to maintain a large number of parallel interactions, and on the side of a validating node. The integrity guarantee is achieved due to the proposed data validation method based on the homomorphic hash function which allows to verify the integrity of data on a storage node without transferring real data to a validating party over the network. The computational complexity of validations depends linearly on the size of validated data and is well suited for parallelization.
Nowadays, building an e cient, reliable and scalable decentralized data storage architecture is an actual problem in both corporate and academic communities since decentralization requires new approaches that adapt existing models.
In the decentralized storage system, where users no longer physically own
the storage of their data, traditional cryptographic primitives for the purpose of
data security protection cannot be directly adopted [
A group of works [1{4] have been done focusing on attempt to solve a remote data validation task in cloud storage. These methods can be classi ed as Proof of Data Possession(PDP) and Proof of Retrievability(PoR).
The PDP scheme initially has been presented by Ateniese [
An e cient PDP method with privacy protection of users data from external
auditors on a client and server sides is needed for decentralized storage. From the
perspective of data privacy protection, this drawback greatly a ects the security
of the protocols [
The design of the Distributed Decentralized Storage Platform (DDSP)
requires to develop an e cient data validation method [
This paper proposes a zero-knowledge data validation method for the
decentralized storage system for minimizing data transferring to maintain the network
scalability and minimizing computational cost on the side of a storage node to
maintain a large number of parallel interactions, and on the side of a validating
node [
Requirements for the data validation [
In order to ful ll the requirements, the data validation method based on
homomorphic hashing [
A hash function has to have the following properties:
{ it should be easily computable;
{ it should be computationally di cult to nd collisions [
Homomorphic hash is a hash function that can compute a hash of a composite block from hashes of individual blocks.
The computational complexity of validations depends linearly on the size of validated data and is well suited for parallelization.
In the proposed method, the challenge-response model is applied. The challenge can be formed as a collective game for all storage nodes that contain validated data. In the simplest case, a request for an arbitrary set of hashes from the data being veri ed is considered. The obtained hashes have to produce the expected hash according to the rule of homomorphism. Detailed description of the challenge and the analysis of resistance to attacks are the subjects of future research.
Validation of each individual object is a computationally expensive task for a decentralized system with an unlimited number of users and the amount of stored data. A storage group is introduced to reduce a computational cost and the amount of stored meta-information for validation. 2.1
Data Validation Complexity: Storage Group The concept of a storage group is introduced to reduce a validation complexity dependence on the number of stored objects in the system. The storage group encapsulates a group of objects' identi ers and a set of homomorphic hashes required for data validation. The safety and accessibility of multiple objects in the network are achieved by the storage group validation without storing metainformation and conducting validation of each object. The concept of a container of objects is introduced. One container can have any number of storage groups. The storage group is an immutable structure, however, new storage groups can still be formed based on the existing ones (the merge operation) or the storage group can be removed (the objects will be deleted). A group of homomorphic hashes is generated for data validation. One hash is created from all the ordered data of all storage group objects (zero-level hash) and a xed number of hashes is created from zones ( rst-level hashes). The zone`s size is a multiple of the size of the object`s block. A xed and equal number of homomorphic hashes of each level is stored for all the groups in DDSP. As an example { retrieving data for the storage group, where the zero-level hash consists of six objects. The case, where it is required to store four rst-level homomorphic hashes, is shown on Fig. 1 below. The actions to form the storage group are following: { getting of a sorted list of homomorphic objects` hashes: h(O1) ::: h(O6); { getting of a single structure from ordered objects` data; { dividing of the data structure into four zones D1:1; D1:2; D1:3; D1:4 in order to form homomorphic hashes; { getting of rst-level hashes H1:1 = h(D1:1) ... H1:4 = h(D1:4).
In the simplest case, a validating node selects a random rst-level homomorphic hash during data validation and requests some number of second-level homomorphic hashes from the placement group nodes. This procedure is a collective game. Merging of the second-level homomorphic hashes, a validating node con rms the validation if the obtained hash corresponds to the rst-level hash being stored. 3
The proposed method allows to minimize the load on the network and validating nodes without transferring a real data over the network. The storage group
Container
Storage Group formation (SG1)
Object 1 Object 2 Object 3 Object 4 Object 5
Object 6 Storage Group 1 Storage Group 2 Storage Group k ... allows to keep a xed amount of meta-information needed for validation process regardless of the size and the number of objects from the storage group. The comparison of the proposed method with the existing ones, a detailed challenge description and the analysis of resistance to malicious attacks are the subjects of further research.