Securing Blockchain based Supply Chain in Agriculture using Isolated Forests with Local Outlier Factor for Anomalies Detection

In recent days, the agriculture supply chain has transformed from a localized, autonomous system to a complex, globally interconnected network of diverse stakeholders, influencing every stage of food production and delivery to consumers. However, the existing Random Forest (RF) utilized for the anomaly detection, but it has high computational complexity. Hence, this research proposes an Isolated Forest with Local Outlier Factor (IF-LOF) for agriculture supply chain anomalies detection. This method uses real-time data collected from Device Network Logs (DNL) and the collected data is secured by utilizing block chain mechanisms. Then, the smart contracts block the suspicious IP address occurrence from the data, which is given by oracle database. Next, min-max normalization technique is utilized to preprocess the secured data and IF-LOF is introduced to detect anomalies by selecting random samples along with threshold value. When the LOF value is less than the value of threshold, then the anomalies are present in data otherwise not. As per the results, the proposed IF-LOF approach offered outperforming results in terms of accuracy (99.73%), precision (97.95%), recall (98.47%) and F1-score (97.32%) respectively.