An Enhanced Anomaly Forecasting in Distributed Wireless Sensor Network Using Fuzzy Model

Wireless sensor networks (WSNs) grieve from a wide range of limitations and aberrations which hinder their smooth functioning. Detecting anomaly in WSNs is a decisive research area, which emphasizes making sensor nodes to be reliable in handling data. Owing to energy restrictions and less computation capability of sensor networks, anomaly detection ought to concentrate on the fundamental limitations of sensor networks. Anomaly detection and dipping noisy data transmission are essential to recover the network life span of sensor networks by promising data integrity. Henceforth, academicians and researchers are frequently getting motivated in finding methods to improve the accuracy of data held by the sensor nodes. In such surroundings, the investigators focus on semi-supervised anomaly detection which uses real data to distinguish incidences that are conflicting with the widely-held data. The proposed idea is acquainted with a fuzzy-based anomaly detection model for semi-supervised anomaly detection in an effort to recover accuracy. The proposed model is compared with other leading existing procedures and methodology over robustness and other substantial metrics. Comparatively, our proposed prototype achieves a high-performance score with 99.70% accuracy, 99.14% precision, 99.27% detection rate, 98.56% specificity, 98.78% F 1 score, and 0.8 correlation coefficient. It is detected that our proposed model diminishes false alarms up to 1.20% through detection which is a key concern in WSNs.