ASDAE-OPF: an adaptive data reduction and optimization model for efficient data handling and routing in wireless sensor networks

Abstract Wireless sensor networks face severe energy limitations, with data transmission consuming the majority of a node’s power. To address this, the proposed framework combines three energy aware mechanisms. First, an Adaptive Sparse Denoising Autoencoder compresses and reconstructs sensor data, transmitting clean and essential information rather than raw redundant readings. Second, cluster head selection is handled through Hierarchical Agglomerative Clustering integrated with the Firefly Algorithm. It’s enabling energy-balanced and adaptive cluster formation. Third, a Cross-layer Opportunistic Routing Protocol, built on Grey Wolf Optimization and Particle Swarm Optimization, selects efficient forwarding paths that reduce latency while preserving network connectivity. Together, these components reduce transmission overhead, balance energy usage across nodes, and extend overall network lifetime. Simulation results confirm improved energy efficiency and prolonged operational life compared to existing WSN schemes. This hybrid technique combines the three methods, and therefore, it would provide faster convergence, better CH selection and link stability by combining the parameters like distance, residual energy, trust, and link quality. The proposed system shows that the QOS can be improved considerably for different network sizes from 100 nodes to 500 nodes. The nodes framework (100 nodes) delivers 1.0 Mbps of throughput, a 93% packet delivery ratio, an average energy consumption of 28 mJ, an end-to-end delay of 1.1 ms and a network lifetime of 6400 rounds. The framework (500 nodes) can sustain 0.76 Mbps throughput, a 98% packet delivery ratio, a 3% packet loss ratio and an average energy consumption of 90 mJ in a network lifetime of 5440 rounds of network size 500 with an end-to-end delay of 4.8 ms.