IoT-Based Harmful Toxic Gases Monitoring and Fault Detection on the Sensor Dataset Using Deep Learning Techniques

Table 1

Comparison of algorithms with used dataset.

Author	Algorithm/method	Dataset	Metrics	Purpose
Sharma et al. [4]	Rule-based (heuristic) methods, histogram-based method, and hidden Markov model (HMM)	Berkeley dataset, Great Duck Island (GDI) dataset, NAMOS dataset	Accuracy, false positive, false negative	For diagnosing faults, heuristic rules are developed to detect abnormal sensor behavior based on sensor correlations. Anomaly sensor measurements are flagged as faults by leveraging this information
Chen [19]	Centralized anomaly data detection	Real time dataset	Fault detection rate	To enhance the fault detection technique of correlated irregular information analysis and abnormal event detection
Tuan Anh Nguyen [23]	Fault-injection algorithms	Sensor Scope dataset	Fault detection rate	To detect and classify faults in sensor data
Mahmudul Hasan	Artificial neural network	DS2OS	Accuracy, precision, recall	Identify the attack and detection anomalies in the internet of objects (IoT) infrastructure
	Support vector machine
	Decision tree
Tsai et al. [15]	Automatic threshold selection, impermanent faults, fault recovery	Airbox open data,	Accuracy, fault detection rate	Design the architecture of detection system and prevent anomaly amongst sensors based on machine learning
		Intel lab data