AI-Based Estimation of Bandwidth Availability for Data Offloading in Edge-Cloud Computing

Edge-cloud computing requires dynamic and accurate bandwidth estimation for intelligent data offloading decisions. Within this context, this article proposes an AI-based mechanism to estimate bandwidth availability using lightweight network measurements for data offloading, called Ensemble Learning for Bandwidth Estimation (ELBE). The proposal employs a stacking ensemble of regression models, as well as defines a Bandwidth Availability Index (BAI) based on network measurements. The experiments performed, using data from Brazil’s national research network, indicate a suitable performance of ELBE, achieving low error (around 3%) and high accuracy (up to 99%), while keeping computational efficiency (execution around 1ms).