Design and Optimization for Storage Mechanism of the Public Blockchain Based on Redundant Residual Number System

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

This paper presents a novel approach to solving one of blockchain’s most persistent scaling issues: the "storage explosion." By applying a mathematical framework known as the Redundant Residual Number System (RRNS), the authors demonstrate a way to slash node storage requirements by over 90% without sacrificing the decentralization or security that makes blockchain valuable.

TL;DR

The researchers propose replacing full-ledger storage with a distributed "remainder" system based on modular arithmetic. This allows individual nodes to store only a tiny fraction of account data while enabling the network to reconstruct the full state and detect malicious tampering with high mathematical certainty.

What Problem Does This Paper Address?

In a public blockchain like Bitcoin or Ethereum, "full nodes" are the backbone of the network. To ensure trust, every full node traditionally stores a complete copy of the entire ledger. As of late 2018 (the paper's context), Ethereum exceeded 110GB and Bitcoin 190GB. This "storage bottleneck" creates a barrier to entry: only users with high-end hardware can run full nodes, leading to centralization.

Existing solutions like "light nodes" rely on full nodes for verification, which weakens decentralization. "Sharding" reduces storage but introduces massive architectural complexity. The authors sought a way to keep all nodes "equal" and independent while dramatically reducing the amount of data each one must hold.

Methods

The core innovation is the application of the Redundant Residual Number System (RRNS) to blockchain account states.

  1. Residual Number System (RNS): Instead of storing a large integer $X$ (like an account balance), RNS represents $X$ as a set of remainders $\{x_1, x_2, ... x_n\}$ relative to a set of coprime moduli $\{m_1, m_2, ... m_n\}$.
  2. Distributed Storage: In this proposed model, each storage node is assigned one small module ($m_i$). It only stores the remainder ($x_i$) of the account balance. Because the module is small (e.g., 16 bits) compared to the account data (256 bits), storage needs plummet.
  3. Redundancy for Fault Tolerance: By adding "redundant" modules beyond what is mathematically necessary to reconstruct the number, the system gains the ability to detect and correct errors. This is vital for a Byzantine environment where some nodes ("devil nodes") might provide false data.
  4. Reconstruction: Consensus nodes use the Chinese Remainder Theorem (CRT) to aggregate remainders from various storage nodes and reconstruct the original 256-bit account value for transaction verification.

Datasets

The authors did not use a public transaction dataset. Instead, they built a simulation platform using Python 2.7 to evaluate the theoretical framework. They simulated a system with:

  • 20,000 accounts.
  • 256-bit account information (standard for Bitcoin/Ethereum).
  • A module set of 32 residual bases (16-bit primes).
  • Randomly injected "devil nodes" to test fault tolerance.

Key Insights and Ideas

  • Mathematical Compression: The storage reduction is a direct result of the bit-width difference. Storing a 16-bit remainder instead of a 256-bit integer results in a 93.75% reduction in storage for that specific data.
  • Parallel Updates: A major "aha" moment is that linear operations (addition/subtraction of balances) can be performed directly on the remainders. If an account balance changes by $\Delta X$, a node only needs to update its local remainder: $x_{new} = (x_{old} + \Delta X) \pmod{m_i}$. The node never needs to know the full balance to update its share.
  • Intrinsic Security: The "Redundant" part of RRNS acts as a built-in checksum. If a malicious node provides a tampered remainder, the CRT reconstruction will produce a value that falls outside the "valid" range, immediately flagging the error.

Key Results

  • Storage Efficiency: In the demo system (16-bit modules vs 256-bit data), the storage volume was reduced to 1/16th of the original size (a 93.75% compression ratio).
  • Fault Tolerance: The system proved highly resilient. In a 32-node simulation, the success probability for recovering correct data remained near 100% even when up to 10 nodes (31.25% of the network) were malicious.
  • Computational Overhead: Error correction and data restoration for 10 malicious nodes took approximately 50ms, suggesting the method is performant enough for real-world block times.

Limitations and Future Work

  • Network Traffic: While storage is reduced, the system requires consensus nodes to fetch remainders from multiple storage nodes to verify a transaction. The paper does not deeply analyze the potential increase in network latency or bandwidth consumption.
  • Scale of Simulation: The test was limited to 32 nodes. While the math scales, real-world public blockchains involve thousands of nodes, which may introduce new challenges in module distribution and discovery.
  • Scope: The optimization focuses on "account information" (the current state). While this is a huge part of the storage burden, the paper notes that transaction history (the chain of blocks) might still require external solutions like IPFS.

Summary updated

Abstract

Blockchain has a broad development potential and application prospects. At present, huge storage volume on each node has become one of the primary bottlenecks that restrict the expansibility of blockchain. Optimization for storage mechanism has become a necessary way to accelerate the improvement of blockchain. There are a variety of solutions to relieve the storage burden by modifying the architecture of blockchain, but most of them weaken the decentralization property of the system, which is the core advantages of the blockchain. In this paper, an optimization scheme based on the redundant residual number system is proposed to dramatically reduce the storage volume of nodes in the blockchain system, and fault tolerance mechanism is designed based on the new storage scheme. The simulation experiments are performed to prove the effectiveness of the proposed mechanism.
BibTeX
@inproceedings{guo2019design,
  author = {Zhaohui Guo and Zhen Gao and Haojuan Mei and Ming Zhao and Jinsheng Yang},
  title = {{Design and Optimization for Storage Mechanism of the Public Blockchain Based on Redundant Residual Number System}},
  booktitle = {IEEE Access},
  year = {2019},
  doi = {10.1109/access.2019.2930125},
}
Figure