Scholarlink

Introduction <p>Block withholding attacks are severe threats in blockchain networks that substantially decrease mining efficiency and jeopardize a fair reward distribution. Current approaches to mitigation depend on either a central detection system or reach a consensus on what mitigation is, both of which are not scalable, adaptable or privacy preserving. Moreover, traditional incentive mechanisms are unable to deter malicious miners, and result in performance reduction.</p> Methods <p>To solve the above problems, the Modified Tactical Flight-Federated Deep Reinforcement Learning (MTF-FDRL) framework is proposed in this study. The framework consists of the federated deep reinforcement learning method for decentralized and privacy preserving miner classification, anomaly filtering by Density-Based Spatial Clustering of Applications with Noise (DBSCAN), and a modified version of the tactical flight optimizer for detection of malicious miners. Moreover, a dynamic income allocation mechanism is embedded to guarantee fair distribution of income and stop adversarial exploitation.</p> Results <p>A simulated proof-of-work blockchain network is used to conduct the experiments to verify the accuracy of the proposed system for detecting malicious miners and detecting the malicious miners, which the proposed system achieves a detection accuracy of 99.41% and a malicious miner identification rate of 99.9%, respectively. This system significantly reduces the amount of energy used, increases the overall performance of the mining pool, and optimizes the reward distribution.</p> Conclusions <p>The framework proposed by MTF-FDRL is efficient, scalable and secure solution to mitigate block withholding attacks. The framework combines intelligent learning, anomaly detection, and optimization methods, resulting in a more reliable, fair, and energy-efficient blockchain system that can be adopted for future decentralized systems.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Secure and Energy Efficient Framework for Block Withholding Attack Mitigation Using Federated Learning Approach

  • 期刊论文

  • 发表日期：

    2026年7月4日

查看全文
  • Karthik. S,
  • Saravanan. S,
  • Bhaggiaraj. S,
  • Muthaiah. U

摘要

Introduction

Block withholding attacks are severe threats in blockchain networks that substantially decrease mining efficiency and jeopardize a fair reward distribution. Current approaches to mitigation depend on either a central detection system or reach a consensus on what mitigation is, both of which are not scalable, adaptable or privacy preserving. Moreover, traditional incentive mechanisms are unable to deter malicious miners, and result in performance reduction.

Methods

To solve the above problems, the Modified Tactical Flight-Federated Deep Reinforcement Learning (MTF-FDRL) framework is proposed in this study. The framework consists of the federated deep reinforcement learning method for decentralized and privacy preserving miner classification, anomaly filtering by Density-Based Spatial Clustering of Applications with Noise (DBSCAN), and a modified version of the tactical flight optimizer for detection of malicious miners. Moreover, a dynamic income allocation mechanism is embedded to guarantee fair distribution of income and stop adversarial exploitation.

Results

A simulated proof-of-work blockchain network is used to conduct the experiments to verify the accuracy of the proposed system for detecting malicious miners and detecting the malicious miners, which the proposed system achieves a detection accuracy of 99.41% and a malicious miner identification rate of 99.9%, respectively. This system significantly reduces the amount of energy used, increases the overall performance of the mining pool, and optimizes the reward distribution.

Conclusions

The framework proposed by MTF-FDRL is efficient, scalable and secure solution to mitigate block withholding attacks. The framework combines intelligent learning, anomaly detection, and optimization methods, resulting in a more reliable, fair, and energy-efficient blockchain system that can be adopted for future decentralized systems.