A Novel k-Coverage and m-Connectivity based Fire Prediction Model for Consumer IoT Using Fish Swarm Optimization and Logistic Regression Model

Continuous monitoring is crucial for the early detection of mine fires in underground coal mines (UCMs). Internet of Things (IoT), is widely used for continuous monitoring of the environment of UCMs. However, coverage and connectivity issues among the nodes in (UCM) pose challenges for sustained monitoring and communication. To address this, the strategy involves partitioning the UCM using designated target points. The key idea is to confirm that nodes adequately cover these predetermined targets, ensuring comprehensive coverage and connectivity throughout the entire UCM. This study addresses the challenges of node coverage and connectivity in UCMs by proposing a novel k-coverage and m-connectivity model using the Fish Swarm Optimization (FSO) algorithm. The methodology involves partitioning the UCM into designated target points and deploying nodes to ensure comprehensive coverage and connectivity. Sensor nodes gather local environmental data, which is then processed to predict potential fire hazards using a logistic regression model (LRM). The FSO algorithm optimizes node deployment by achieving k-coverage, where each target point is covered by at least k nodes, and m-connectivity, ensuring robust communication routes even when some connections fail. A fitness function is formulated to minimize node count, maximize coverage, and maintain connectivity. The proposed method demonstrates high accuracy and efficiency in simulations, outperforming traditional linear regression and Naive Bayes models with an accuracy of 98% in fire prediction. This approach is scalable and can be adapted for various Consumer IoT applications to enhance safety and operational efficiency.