Machine learning for resource-efficient data transfer in mobile crowdsensing

Chapter in Machine Learning for Future Wireless Communications (F. L. Luo, ed.), John Wiley & Sons Inc, February 2020

Abstract:

This chapter provides an overview of applications and requirements for vehicular crowdsensing as well as anticipatory data transmission. As a case study, it presents an opportunistic, context-predictive transmission scheme that relies on machine learning-based data-rate prediction for channel quality assessment, which is executed online on embedded devices. The chapter provides an overview of requirements and current work in mobile crowdsensing and anticipatory data transfer. It summarizes the work on machine learning channel-aware transmission (ML-CAT) and ML-predictive CAT (ML-pCAT), which extend the established transmission schemes CAT and pCAT with ML-based data-rate prediction. The proposed mechanisms implement client-side anticipatory networking and bring together different research directions of anticipatory optimization such as ML-based channel quality assessment, mobility prediction, and connectivity maps. The chapter presents the methodology setup for the real-world performance evaluation, and discusses the results of the empirical performance evaluation about the proposed solutions.