[Paper Review] Mobile Edge Computing, Blockchain and Reputation-based Crowdsourcing IoT Federated Learning: A Secure, Decentralized and Privacy-preserving System.

Yang Zhao, Jun Zhao+3 more|arXiv (Cornell University)|Jun 26, 2019

Privacy-Preserving Technologies in Data9 references81 citations

TL;DR

This paper proposes a secure, decentralized, and privacy-preserving federated learning system for IoT device manufacturers that leverages mobile edge computing, blockchain, and reputation-based crowdsourcing. It enables users to collaboratively train a personalized model using locally collected IoT data with differential privacy, while blockchain ensures trustless aggregation and an incentive mechanism encourages participation through coin rewards.

ABSTRACT

Internet-of-Things (IoT) companies strive to get feedback from users to improve their products and services. However, traditional surveys cannot reflect the actual conditions of customers' due to the limited questions. Besides, survey results are affected by various subjective factors. In contrast, the recorded usages of IoT devices reflect customers' behaviours more comprehensively and accurately. We design an intelligent system to help IoT device manufacturers to take advantage of customers' data and build a machine learning model to predict customers' requirements and possible consumption behaviours with federated learning (FL) technology. The FL consists of two stages: in the first stage, customers train the initial model using the phone and the edge computing server collaboratively. The mobile edge computing server's high computation power can assist customers' training locally. Customers first collect data from various IoT devices using phones, and then download and train the initial model with their data. During the training, customers first extract features using their mobiles, and then add the Laplacian noise to the extracted features based on differential privacy, a formal and popular notion to quantify privacy. After achieving the local model, customers sign on their models respectively and send them to the blockchain. We use the blockchain to replace the centralized aggregator which belongs to the third party in FL. In the second stage, miners calculate the averaged model using the collected models sent from customers. By the end of the crowdsourcing job, one of the miners, who is selected as the temporary leader, uploads the model to the blockchain. Besides, to attract more customers to participate in the crowdsourcing FL, we design an incentive mechanism, which awards participants with coins that can be used to purchase other services provided by the company.

Motivation & Objective

• To address the limitations of traditional surveys in capturing real user behavior by leveraging actual IoT device usage data.
• To enable IoT manufacturers to build accurate predictive models of user behavior without compromising user privacy.
• To eliminate reliance on a centralized aggregator in federated learning by replacing it with a blockchain-based decentralized model aggregation process.
• To incentivize user participation in crowdsourced federated learning through a coin-based reward mechanism.
• To ensure model integrity and participant accountability using a reputation system integrated with blockchain.

Proposed method

• Customers collect IoT usage data via their smartphones and perform local model training with feature extraction on mobile devices.
• Differential privacy is enforced by adding Laplacian noise to the extracted features before local model training.
• Each customer signs their local model using a digital signature to ensure authenticity and integrity.
• Local models are submitted to a blockchain, replacing the traditional centralized aggregator in federated learning.
• Miners on the blockchain compute the global model by averaging the signed local models received from participants.
• A temporary leader miner uploads the final aggregated model to the blockchain, ensuring finality and transparency.

Experimental results

Research questions

• RQ1 How can a decentralized federated learning framework be designed to preserve user privacy while enabling accurate model training from IoT device data?
• RQ2 What role does blockchain play in replacing the centralized aggregator and ensuring trustless model aggregation in federated learning?
• RQ3 How can differential privacy be effectively integrated into the local training phase to protect user data?
• RQ4 What incentive mechanism can effectively encourage widespread user participation in crowdsourced federated learning?
• RQ5 How can reputation and accountability be maintained in a decentralized, trustless environment for IoT federated learning?

Key findings

• The integration of mobile edge computing significantly reduces the computational burden on end-user devices, enabling efficient local model training.
• The addition of Laplacian noise to features ensures differential privacy, quantitatively protecting user data during local training.
• Blockchain-based model aggregation eliminates the need for a trusted central authority, enhancing system decentralization and trust.
• The incentive mechanism using coins successfully encourages user participation, with rewards distributed based on contribution and reputation.
• The system achieves a balance between model accuracy, privacy preservation, and decentralization through a multi-layered trust and security architecture.