Adaptation of Blockchain using Ethereum and IPFS for Fog based E-Healthcare Activity Recognition System

Abstract

The cloud storage is far away from us and it is not capable of handling huge bandwidth data due to network latency. The goal of the Fog computing is to decrease the data that needs to be transferred to the cloud for data processing and to increase the efficiency. Fog computing improves the QoS and also reduces network bandwidth. All machine learning algorithm performances are dependent on the quality of the training data. If the training data is inadequate or it is modified by attackers, then the machine learning algorithm will miss predict and may give invalid results. In order to avoid modification of training data, it is preferable to store it in Blockchain. Blockchain is a decentralized model and its data structure is practically difficult to forge, hence it has attracted both industry and research now-a-days. Proposed system uses Ethereum platform to implement Blockchain. Ethereum allows users to create and run decentralized applications (DApps) to make agreements and to conduct transactions directly with each other without any third party by making use of smart contracts. Blockchain is convenient to store only a small amount of data, hence alternative solution to store large amount of data, for example in healthcare, in Blockchain is possible with the help of IPFS (Interplanetary File System). In e-healthcare applications, Activity Recognition System (ARS) is the most significant undertaking in remote checking of patients experiencing physical medical issues for taking quick action. Hence this paper, especially concentrate on the overall framework of the e-healthcare ARS and implementation of Blockchain to store e-healthcare training data to avoid forging, ultimately which improves ARS results. Our implementation results also show that constant and less transaction fee required to store into Blockchain irrespective of size of training data with help of IPFS and also proved transaction throughput increases and network delay decreases with help of IPFS.
 HIGHLIGHTS
 
 All machine learning algorithm performances are dependent on the quality of the training data. If the training data is inadequate or it is modified by attackers, then the machine learning algorithm will miss predict and may give invalid results
 To avoid modification of training data, it is preferable to store it in Blockchain. Blockchain is a decentralized model, and its data structure is practically difficult to forge
 Proposed system uses Ethereum platform and IPFS to implement Blockchain using smart contracts for health care applications
 Our implementation results also show that constant and less transaction fee required to store into Blockchain irrespective of size of training data with help of IPFS, and proved transaction throughput increases and network delay decreases with help of IPFS
 
 GRAPHICAL ABSTRACT