Due to the continuous increase in communication bandwidth and the use of highly efficient yet nonlinear power amplifiers, Digital Predistortion (DPD) algorithms are becoming increasingly complex. In particular, neural network (NN) based DPD approaches using Phase-Normalized NN architectures often incur substantially higher computational costs than widely deployed polynomial-based methods, such as the Memory Polynomial (MP) and Generalized Memory Polynomial (GMP) models. To bridge this gap between research performance and practical implementation, we propose a low-complexity Feature Selection NN DPD architecture. The proposed method employs an offline feature-engineering pipeline based on the Least Absolute Shrinkage and Selection Operator (LASSO) and the Minimum Redundancy Maximum Relevance (MRMR) algorithm to construct a compact and informative input representation. Using measured wideband FR3 power amplifier datasets that are publicly released with this work, we demonstrate up to 30% reduction in computational complexity while maintaining comparable linearization performance.
Cel Thys, Rodney Martinez Alonso, A. Alsarraf et al.· 0 citations
Research on low-power Internet of Things (IoT) systems has gained significant momentum within the broader context of green and sustainable IoT. In this setting, batteryless (BL) IoT has emerged as a promising solution to reduce maintenance costs and environmental impact by eliminating the need for battery replacements. As large-scale IoT deployments for monitoring and sensing applications continue to expand, important challenges arise in the design and operation of sustainable BL IoT networks, including long-term reliability, performance evaluation, and the analysis of energy-harvesting behavior under real-world conditions. To help address these challenges, this article presents a comprehensive dataset capturing the behavior of BL IoT devices deployed in an indoor environmental sensing network. The dataset comprises 100 days of measurements collected from sensors installed throughout an office building and includes data from two types of IoT devices: 1) plugged-in (PI) sensors with continuous power supply; and 2) BL sensors powered exclusively by energy harvested from indoor light. This dual-device deployment enables direct comparison of sensing performance, reliability, and energy dynamics between powered and energy-harvesting systems. The final dataset contains over three million samples collected from 28 sensors (14 PI and 14 BL) deployed across approximately 300 m<inline-formula><tex-math notation="LaTeX">${}^{2}$</tex-math></inline-formula> of office space spanning nine rooms. The dataset provides a valuable resource for the systematic investigation of BL IoT systems, enabling rigorous analysis of deployment strategies, spatial–temporal sensing dynamics, energy-harvesting behaviors, system performance, and long-term sustainability considerations in next-generation self-powered IoT sensing networks.</p> <p><bold>IEEE SOCIETY/COUNCIL</bold> Communications Society (COMSOC)</p> <p><bold>DATA TYPE/LOCATION</bold> Comma Separated Values (CSV); KU Leuven, Belgium</p> <p><bold>DATA DOI/PID</bold> <ext-link ext-link-type="uri" xlink:href="https://doi.org/10.21227/TF0N-SB66">10.21227/TF0N-SB66</ext-link>
Jimmy Fernandez Landivar, Ihsane Gryech, A. Colpaert et al.· IEEE Data Descriptions· 0 citations