Energy-efficient wireless sensor network for urban groundwater level monitoring using machine learning and sink mobility

Urban groundwater level monitoring is vital for enabling data-driven decision-making and sustainable urban water resource management. Wireless Sensor Networks (WSNs) offer an effective solution for real-time observation of spatially distributed underground water sources. However, conventional WSN protocols often face significant limitations, such as unbalanced data routing, excessive energy consumption, and the energy hole problem near the sink. To overcome these challenges, this paper proposes an energy-efficient WSN protocol named Sleep Scheduled Data Aggregation with Sink Mobility (SSDA-SM), specifically designed for Urban Groundwater Monitoring (UGM) in heterogeneous sensor networks. The protocol incorporates a machine learning (ML)-based probabilistic clustering mechanism to optimize Cluster Head (CH) selection, considering residual energy, node density, and average network energy. To further conserve energy, a proximity-aware sleep scheduling strategy selectively deactivates redundant nodes, while dynamic sink mobility uniformly balances communication load and mitigates the energy hole problem. Moreover, to reduce transmission overhead, Compressive Sensing (CS) is applied at the CH level for data aggregation, and the original data is accurately reconstructed at the sink using an appropriate decoding algorithm. The SSDA-SM protocol is implemented and simulated in MATLAB. Performance evaluation shows that SSDA-SM significantly outperforms existing protocols such as OCNTMS, MEDF, SEI$$^{2}$$, and MACOA across various metrics including network lifetime, energy consumption per round, data throughput, packet delivery ratio, end-to-end delay, cluster stability, compression ratio, and reconstruction accuracy. These results demonstrate that SSDA-SM is a robust, scalable, and energy-efficient solution for long-term urban groundwater level monitoring using heterogeneous WSNs.