Blockchain based Internet of Things (Debashis De, Siddhartha Bhattacharyya etc.)

An Efﬁcient Blockchain-Based IoT

System Using Improved KNN Machine

Learning Classiﬁer

Roseline Oluwaseun Ogundokun

, Micheal Olaolu Arowolo

Sanjay Misra

, and Robertas Damasevicius

Abstract Theintroductionofblockchaintechnology(BT)hasevolvedintoadistinc-

tive, disturbing, and trendy technology in recent years. Data security and privacy

are prioritized in BT’s decentralized database. New security concerns raised by BT

include common attacks and double-spending. To address the above-mentioned chal-

lenges, data analytics on a blockchain-based IoT network is necessary to protect data.

The value of emerging technology Machine Learning (ML) is highlighted through

analytics on this data. When ML and BT are combined, very precise results may

be obtained. This study, therefore, aimed to give a comprehensive study on the

use of machine learning to make Blockchain-based IoT network smart applications

that are more robust to handle network attacks. To investigate these attacks on a

blockchain-based IoT network, an improved K-Nearest Neighbor (KNN) classiﬁer

was postulated. Improved KNN (I-KNN) surpassed traditional KNN (T-KNN) with

an accuracy of 96.7% and 81.6% for the I-KNN classiﬁer and T-KNN, respectively.

Keywords Blockchain · IoT · K-Nearest Neighbor · Machine learning ·

Classiﬁcation

R. O. Ogundokun · M. O. Arowolo

Department of Computer Science, Landmark University Omu Aran, Omu-Aran, Nigeria

M. O. Arowolo

S. Misra (B)

Department of Electrical and Information Engineering, Covenant University to Department of

Computer science and Communication, Ostfold University College, Halden, Norway

R. Damasevicius

Department of Software Engineering, Kaunas University of Technology, Kaunas, Lithuania

D. De and S. Bhattacharyya (eds.), Blockchain based Internet of Things,

Lecture Notes on Data Engineering and Communications Technologies 112,

https://doi.org/10.1007/978-981-16-9260-4_7

171