Storage Mechanism Optimization in Blockchain System Based on Residual Number System

Haojuan Mei, Zhen Gao, Zhaohui Guo, Ming Zhao, Jinsheng YangIEEE Access 2019 · DOI

Abstract

Huge storage volume is one of the main bottlenecks for the development of blockchain, so how to release the burden by optimizing the storage mechanism has become an important problem. Most of the current solutions would modify the architecture of blockchain, which weakens the characteristics of the decentralization, such as cloud storage. In this paper, a storage optimization mechanism based on residual number system is proposed to reduce the storage volume on each node. In addition, the recovery procedure of CRT-II (The new Chinese Remainder Theorem) is used to detect garbled data from devil nodes, which enable the proposed storage mechanism with strong fault tolerance capability. Both theoretical analysis and simulation results prove the effectiveness and reliability of the proposed scheme.

Introduction

The rapid growth of blockchain technology has brought the "storage explosion" problem to the forefront of scalability challenges. As decentralized ledgers like Bitcoin and Ethereum continue to grow—surpassing 190 GB and 110 GB respectively by late 2018—the requirement for every full node to maintain a complete copy of the history becomes increasingly prohibitive. This trend threatens decentralization, as high hardware requirements push average users toward centralized third-party service providers.

The proposed optimization targets the foundational Data Layer of the blockchain architecture.

Existing attempts to solve this often involve off-chain storage or heavy modifications to the blockchain's core architecture, which can compromise security or traceability. This paper introduces a novel storage optimization mechanism that leverages the mathematical properties of Residual Number Systems (RNS) to distribute data without sacrificing the system's decentralized nature.

System Architecture

The proposed system redefines the roles within a blockchain network to support distributed remainder-based storage. Instead of every node acting as a full ledger holder, the network is composed of different node types:

  • User Nodes: Initiate transactions and manage personal account data.
  • Storage Nodes: Store only "partial blocks" or remainders of the global state.
  • Verification Nodes: Responsible for reconstructing data and validating transactions using the Chinese Remainder Theorem.
Network architecture showing the distribution of partial blocks across storage nodes and the verification process.

The RNS-Based Storage Approach

The core innovation lies in using Residual Number Systems (RNS) and Redundant RNS (RRNS). In a standard system, an account value $X$ (typically 256 bits) is stored in its entirety. In the proposed RNS scheme:

  1. Data Shredding: A value $X$ is converted into a set of small remainders $(x_1, x_2, ..., x_n)$ based on a set of coprime moduli $(m_1, m_2, ..., m_n)$.
  2. Distributed Storage: Each storage node only stores a single 16-bit remainder $x_i$ instead of the full 256-bit value.
  3. Homomorphic Updates: RNS allows for parallel arithmetic. If a transaction changes an account balance by $\Delta X$, each node updates its local remainder as $x_i' = (x_i + \Delta X) \pmod{m_i}$. This allows for independent updates without needing to reconstruct the full value $X$ at every step.

Fault Tolerance and Data Recovery

Storing data in fragments introduces risks, particularly from "devil nodes"—malicious actors that provide incorrect remainders during reconstruction. To counter this, the authors employ a hybrid fault-tolerance strategy:

  • Anchor Nodes via Raft: A small subset of nodes uses the Raft consensus algorithm to maintain two "anchor" remainders ($x_1, x_2$). These are guaranteed to be correct and serve as a mathematical ground truth.
  • CRT-II Reconstruction: When a verification node needs to recover the full value $X$, it collects remainders from various storage nodes. It then uses the New Chinese Remainder Theorem (CRT-II).
  • Overflow Detection: By comparing the collected remainders against the trusted anchors, the system can detect if the reconstructed value falls into an "invalid range" (the overflow zone). If an inconsistency is found, the system identifies the "garbled" data and replaces it until a valid reconstruction is achieved.

Performance Evaluation

The researchers tested the system using a simulation mirroring standard blockchain parameters (256-bit account values and 16-bit remainders).

  • Storage Efficiency: The scheme achieved a compression rate of approximately 6.25% (1/16). This drastically reduces the hardware barrier for participating in the network.
  • Reliability: The system demonstrated remarkable resilience. Even with 20% malicious nodes, the probability of successful data recovery remains high. With up to 50% malicious nodes, the iterative replacement algorithm can still recover the correct data with 99.99% probability within just 10 attempts.
  • Overhead: While fetching remainders adds some network communication, the small size of the 16-bit remainders ensures that this overhead is manageable compared to the massive savings in local storage.

Conclusion

By integrating Residual Number Systems into the blockchain data layer, this research provides a viable path toward solving the storage explosion problem. The proposed mechanism offers a significant reduction in storage requirements while maintaining high fault tolerance against malicious actors. Future work will focus on optimizing network latency and scaling the consensus mechanisms used for anchor nodes.

Summary updated

Abstract

Huge storage volume is one of the main bottlenecks for the development of blockchain, so how to release the burden by optimizing the storage mechanism has become an important problem. Most of the current solutions would modify the architecture of blockchain, which weakens the characteristics of the decentralization, such as cloud storage. In this paper, a storage optimization mechanism based on residual number system is proposed to reduce the storage volume on each node. In addition, the recovery procedure of CRT-II (The new Chinese Remainder Theorem) is used to detect garbled data from devil nodes, which enable the proposed storage mechanism with strong fault tolerance capability. Both theoretical analysis and simulation results prove the effectiveness and reliability of the proposed scheme.
BibTeX
@inproceedings{mei2019storage,
  author = {Haojuan Mei and Zhen Gao and Zhaohui Guo and Ming Zhao and Jinsheng Yang},
  title = {{Storage Mechanism Optimization in Blockchain System Based on Residual Number System}},
  booktitle = {IEEE Access},
  year = {2019},
  doi = {10.1109/access.2019.2934092},
}
Figure