O.S. Mezentseva, A.I. Alekseev

The paper deals with the methods of performance improvement in the distributed data storage systems based on the threshold schemes (on example of Shamir-s secret sharing scheme). The factors affecting the performance of distributed data processing systems are analyzed, the ways of enhancing performance are presented, namely, through the use of hardware-emulated neural networks, parallel matrix algebra algorithms (recursive block scheme of matrix multiplication in the quaternary trees format, a parallel Strassen-s method modification) and modular arithmetic algorithms. The paper proposes the use of the mathematical apparatus of residual classes, what leads to neural network-based calculations and independent processing. The NM6403 neuroprocessors-based hardware platform required for the implementation of the proposed methods is discussed.