On December 20th, the NEO SPCC developer group attended the MICSECS-2018 Conference at ITMO University in St. Petersburg, where it presented three scientific works pertaining to the development of a NEO distributed decentralized storage platform (DDSP).

NEO SPCC is a Russian NEO development team whose formation was largely driven by City of Zion and Neon Exchange co-founder Fabian Wahle. The team’s core project is the development of a distributed file storage system, a candidate for the NeoFS initiative. Daily operations are run by Sergei Liubich and Dr Anatoly Bogatyrev, former researchers at Samsung and Dell.

The MICSECS Conference (Majorov International Conference on Software Engineering and Computer Systems) welcomed international researchers to present their findings in a variety of fields. The theses presented by NEO SPCC were as follows.

Zero-knowledge data validation method based on the homomorphic hash function in the DDSP

Presented by Anatoly Bogatyrev, the thesis outlined the development of an efficient data validation method that aims to preserve network scalability by minimizing data transfers and computational costs for storage nodes. The approach would serve as a more scalable alternative to the commonly used Merkle tree-based data validation methods.

The DDSP design aims to ensure the integrity of data by using verifiable homomorphic hash signatures, removing the need for real data to be transferred to a validator on the network. The computational complexity of these validations scale linearly with the size of validated data and is designed with parallelization in mind.

The paper credits Stanislav Bogatyrev, Anatoly Bogatyrev, Sergei Liubich and Fabian Wahle. The source code and a demo video may be found here.

NEO SPCC also submitted a provisional patent application on some of the novel engineering methods used in its homomorphic zero-knowledge data validation, which is hoped to protect its use outside of the NEO ecosystem.

The model of network map and data placement in the DDSP

Existing distributed data storage approaches are not well suited for decentralized environments, so NEO SPCC has been researching alternatives with the hope of forming a new model for decentralized storage systems.

Such a storage system would need to distribute data between available storage nodes, as well as intelligently reorganize data when nodes join or leave the network. Flexible constraints on object replica placement are also required to maximize data safety in the event of a fault.

The thesis was presented once again by Anatoly Bogatyrev, who is credited alongside Sergei Liubich, Fabian Wahle, Stanislav Bogatyrev and Alexey Vanin. The source code and a demo video may be found here.

Push-gossip protocol efficiency with network topology propagation

The third thesis was presented by Alexey Vanin and is concerned with the data distribution between nodes. The system must be able to quickly respond to events such as failures, overloads or changes in its structure, which is usually achieved through the use of a reactive notification protocol.

The most common approach used is a gossip protocol, where the nodes that comprise a distributed system may receive statistics about the network by exchanging messages with each other.

The thesis outlines one such use case for a gossip protocol, where subnetworks of nodes may be formed by informing each other about their participation in the work done by the network. The research aims to improve the efficiency of gossip protocols for this use case.

The source code and a demo video may be found here.

Tags: