Abstract
With raising improvements in various sensor applications, wireless sensor networks (WSNs) are considered as a significant field for research. It is composed on numerous sensor nodes (SNs) to observe sense and monitor environment based on application need. Moreover, SNs are battery constrained (some may be dead). Henceforth, number of energy-efficient protocols are considered and implemented in state-of-the-art works. However, choosing an optimal path of SN and base station is measured as a crucial crisis. To get rid of this issue, learning-based routing is considered to enhance routing mechanism stronger. In all iteration, node caching and node stacking is done using learning approach to evaluate shortest path between SNs and cluster head (CH). Here, network is being trained by considering diverse WSNs features and to identify which SNs must be selected for subsequent hops to provide shorted path between SNs and CH. The proposed model is termed as learning-based stacking and caching (LSC) model. Here, simulation is done in NS-2 environment and shows that the proposed technique may outperform prevailing protocols like LEACH and A-LEACH. Performance metrics like network lifetime, packet delivery ratio, end-to-end (E2E) delay, energy consumption, residual energy, error computation, cache hit ratio and hop reduction ratio are evaluated. The residual energy of a cluster size with 20 nodes is 0.82 J. The network lifetime of LSC is 180 s with higher PDR. Similarly, the delay is 5 ms for 400 rounds. In LSC, CHR is 10, 25, 32, 45, 55, 63 and 65 while HRR is 20, 25, 26, 28, 27, 28 and 30 respectively.