Time-Efficient Blockchain-Based Federated Learning

Federated Learning (FL) is a distributed machine learning method that ensures the privacy and security of participants’ data by avoiding direct data upload to a central node for training. However, the traditional FL typically applies a star structure with cloud servers as the central aggregator for the model parameters from different terminals, leading to problems such as central failure, malicious tampering and malicious participants, resulting in training errors or system crashes. To address these issues, a permissioned blockchain is used to build a secure and reliable data-sharing platform among participating terminals, replacing the central aggregator in the traditional FL called blockchain-based federated learning. However, the block generation method of the blockchain system may introduce significant latency in the federated learning where distributed model parameters upload randomly, resulting in low efficiency of the federated learning. To overcome this, we propose a block generation strategy that groups terminals and generates a block for each group, which minimizes the latency of a single round of federated learning, and an optimal block generation algorithm that considers data distribution, terminal resources, and network resources is provided. The analysis shows that the proposed algorithm can effectively obtain the optimal solution of block generation to minimize the authentication time, and we conduct extensive experiments that demonstrate the time efficiency of the proposed algorithm.

I. Introduction

The exponential growth of network applications and the increasing number of end devices connected to the Internet have led to an unprecedented rise in the volume of data generated in networks [1]. At the same time, privacy protection and trust management have become a crucial concern [2], [3]. Federated learning (FL) is a distributed machine learning that trains a model without uploading the original data of participants [4], [5], [6], [7]. This method prevents the central server from collecting private raw data of participants; instead, the model parameters learned locally from participants are collected and updated iteratively by a center aggregator. In Fig. 1, the illustration showcases a federated learning system, where a central aggregator is responsible for collecting model parameters ,.. from the terminal set during the tth round of learning. Each terminal i trains its local model parameters using its dataset . Following the local training, terminals independently upload their trained model parameters to the aggregator. To update the global model parameters, the aggregator performs a weighted aggregation of the uploaded model parameters, expressed as . This aggregation incorporates the size of each terminal dataset , ensuring a fair combination of the model parameters from different terminals. The resulting global model parameters from the tth round of learning are then downloaded to all terminals. Subsequently, terminals utilize the new global model parameters to commence the next round of learning, until the global model meets the desired accuracy. However, the central aggregator still risks central failure, and the uploaded model parameters still carry the risk of being decrypted and tampered with during the uploading process, which may expose the local information. To enable safe and secure distributed learning, this paper applies a blockchain-based decentralized federated learning system. This system uses the blockchain network to replace the parameter aggregation function of the central server. This approach ensures that any malicious attack or tampering with the upload parameters can be traced back through the blockchain [8], [9], [10]. Blockchain technology is a decentralized and distributed ledger system that offers a secure and transparent platform for recording and validating transactions or data. It operates by organizing data into blocks that are appended in a virtual chain, forming an immutable record. Through a consensus mechanism, participants collectively validate and agree on the state of the ledger, ensuring trust and integrity. In the context of model parameter aggregation of federated learning, blockchain can be leveraged to securely record and aggregate model parameters from individual participants in a transparent manner. This approach guarantees data integrity and prevents unauthorized tampering, fostering a reliable and auditable framework for federated learning. Fig. 1.

Federated learning.