Energy-Constrained Deployment of Learning Models on Distributed Research Sensors: A Reproducibility-Centred Evaluation

Abstract

Edge sensors must trade the convenience of local AI inference against limited battery life and intermittent connections. The objective is to identify operating conditions in which local prediction is preferable to delayed cloud processing. Relevant methodological and data-quality literature informs the design of the sampling and validation strategy [3–7]. The proposed method is a discrete-event experiment using simulated energy budgets and repeatable sensor sampling patterns. Comparators will include continuous edge inference, compressed models and scheduled cloud offloading. Related enterprise-storage studies inform the assessment of operational reliability [1] and the deployment assumptions adopted for system-level comparison [2]. The reproducibility extension records dataset versions, model settings, exclusion rules and provenance before any held-out evaluation. Statistical and computational sources help define the reproducibility checks and comparison procedures [8–12]. The primary outcomes are energy per valid inference, end-to-end delay, missed anomalies and buffering requirements. Potential biases include connectivity losses, battery ageing, temperature effects and uncertainty in simulated demand. An independent replication exercise would repeat the main analysis from the documented artefacts and quantify disagreements. Additional research informs the prespecified stress tests and limits on interpretation [13–17]. The intended contribution is an auditable research protocol rather than an unsupported claim of effectiveness; related work frames its transferability limits [18–23].