An efficient approach for rice prediction from authenticated Block chain node using machine learning technique

Abstract With the advent of Internet of Things (IoT), even devices communicate without human intervention. Even though many cryptographic schemes provide secure communication between devices, errors happen through wireless link. Wireless communications are prone to errors such as session hijacking, masquerading, impersonation attack, Distributed Denial of Service (DDoS) attack, non-repudiation attack, tampering and many more. Block chain provides immutable transaction with authorized communication. To address the problem of non-repudiation attack in IoT and log false information, block chain based authentication using Rivest–Shamir–Adleman (RSA) digital signature is proposed. This method prevents the existence and active participation of fake nodes. The performance of a public key operation is also measured with the crypto metrics of latency (204.2ms, 1642.7ms) and throughput metric (37.2, 12) connections per sec for RSA key sizes of 1024 and 2048 . The environmental factors such as temperature, humidity, pH (potential for hydrogen) concentration and rainfall are collected from sensors deployed at agricultural lands through node Micro Controller Unit (MCU) and transferred to Block chain enabled edge device. Furthermore, edge intelligence is performed to predict rice production with Machine Learning (ML) algorithm and the best appropriate model is selected from the regression metric which includes, Mean Absolute Error (MAE), Mean Square Error (MSE), Root Mean Square Error (RMSE) and Coefficient of Determination (R2).