A Coverage-Preserving Dual-Graph Neural Sleep Scheduler for Wireless Sensor Networks

This project develops and tests a Dual‑Graph Neural Sleep Scheduler (DGNSS) that aims to lengthen the life of dense wireless sensor networks (WSNs) while keeping sensing coverage and communication links intact. Conventional sleep‑scheduling methods often fail to balance the need for each node to save energy with the overall network’s topology requirements. The approach represents the sensor layout with two complementary graphs: * A **communication graph** that shows which nodes can talk to each other via radio. * A **coverage‑overlap graph** that captures how sensing areas overlap and provide redundancy. Each graph is fed into its own Graph Sample and Aggregate (GraphSAGE) encoder. The encoders’ outputs go into a gated

fusion module that produces a “keep” score for every node. These scores consider the node’s position in the graphs, its remaining energy, and its time‑dependent behavior. Instead of using a simple probabilistic threshold, the system uses a deterministic feasibility‑by‑construction decoder to fine‑tune the scores. The decoder performs: * Multi‑start pruning, * Exact grid‑coverage testing, * Verification that all active nodes stay connected to the sink, * Temporal rotation of energy usage. A simulation framework built around this design lets researchers evaluate how long the network lasts, how well coverage is preserved, and how many packets get through as node density changes. The work offers a clear path toward more energy‑efficient

management of wireless sensor networks.

Message: Hi FE, I need help with “A Coverage-Preserving Dual-Graph Neural Sleep Scheduler for Wireless Sensor Networks” in “Electronics & Communication Engineering”