Smart Partitioned Blockchain

Area of Science:

• Computer Science
• Information Technology
• Machine Learning

Background:

• Blockchain technology offers potential advancements across various fields.
• Integrating blockchain with smart spaces and the Internet of Things (IoT) presents challenges due to diverse transaction sensitivities (correctness, time, specialization).
• Existing blockchain models may not adequately address the nuanced requirements of sensitive data processing in interconnected systems.

Purpose of the Study:

• To introduce a customized blockchain model, the smart partitioned blockchain, for smart spaces and IoT applications.
• To leverage machine learning and deep learning for classifying transactions based on sensitivity levels.
• To demonstrate the model's adaptability and effectiveness in meeting specific application requirements.

Main Methods:

• Development of the smart partitioned blockchain model.
• Utilizing machine learning (Random Forest) and deep learning (Sequential Deep Learning) for transaction classification.
• Mapping classified transaction pools to specific, potentially permissioned or permissionless, parts of the blockchain architecture.
• Experimental validation using bank and medical datasets with predefined sensitivity thresholds.

Main Results:

• The Random Forest model achieved 100% accuracy in classifying critical bank transactions.
• Sequential Deep Learning achieved 91% accuracy for classifying speculative medical transactions.
• The smart partitioned blockchain model successfully mapped transactions to appropriate blockchain segments based on sensitivity.
• Acceptability of classification was evaluated against predefined sensitivity thresholds for each dataset.

Conclusions:

• The smart partitioned blockchain model offers a flexible and customizable solution for integrating blockchain with sensitive IoT and smart space data.
• Machine learning and deep learning are effective tools for managing transaction sensitivity in blockchain applications.
• The model's performance demonstrates its potential for secure and efficient data handling in critical sectors like finance and healthcare.