Mathematical models for distributed storage and processing of high-dimensional data with a high degree of fault-tolerance

The paper presents mathematical models for distributed storage and processing of high-dimensional data. The models presented are based on the threshold schemes and provide a high degree of fault-tolerance. Presented parallel algorithms use modular arithmetic to reach noticeable speed-up as compared to equivalent serial schemes, what makes the proposed algorithms suitable for large-scale real-time systems. Implementation possibilities are also discussed with the focus on NM6403 neuroprocessor