Swarm-Based Decentralized Memory Sharing for Resilient Wireless Sensor Deployments

Wireless sensor networks are increasingly used in monitoring and safety?critical environments, where data must be preserved even if nodes fail. Nodes can die from battery depletion, hardware faults, or harsh conditions, breaking routes and permanently losing the measurements stored on them. Traditional designs often rely on fixed routes and central sinks, so failure of key nodes disrupts connectivity and destroys locally held data. The proposed work introduces a swarm-based decentralized memory sharing scheme that adds a lightweight redundancy layer. Each node stores its own readings and also keeps small, encoded fragments of data from nearby nodes. Periodically, sensed data are split into multiple erasure-coded fragments, and subsets are shared with neighbours, which buffer them in limited memory. If a node stops sending heartbeats and is considered failed, surviving neighbours forward their stored fragments to a recovery point, where the original readings are reconstructed as long as enough fragments arrive. Simulations show that this approach significantly improves post-failure data recovery compared to a conventional architecture, while keeping communication and energy overhead much lower than full data replication, making it well-suited for long-lived industrial, environmental and safety-critical monitoring deployments.