MKFF: mid-point K-means based clustering in wireless sensor network for forest fire prediction

Forest fires, by disrupting the ecological equilibrium and exacerbating global warming, pose a threat to both wildlife and the overall environmental stability. To safeguard our ecosystems, it is imperative to predict and detect forest fires at an early stage. Wireless Sensor Networks (WSNs) have gained popularity due to their cost-effectiveness, low power consumption, and portability in achieving this goal. This research introduces an innovative method based on mid-point K-means clustering to forecast three forest activity zones: high-active (fire-prone), medium-active, and low-active zones. This system excels in identifying high-active zones with remarkable accuracy (98%). The sensor node at the high-active zone’s center continuously transmits data to the Base Station (BS), promptly notifying the relevant authorities of potential forest fires. In contrast, the medium-active zone’s sensor node periodically shares environmental data, while the low-active zone’s node conserves energy by not transmitting data to the BS, thereby enhancing network longevity and energy efficiency.