Adaptive CSI and feedback estimation in LTE and beyond: a Gaussian process regression approach

• DOI: 10.1186/s13638-015-0388-0
• Publication Date: 2015-06-11T16:01:11Z
• Authors: Alessandro Chiumento; Mehdi Bennis; Claude Desset; Liesbet Van der P...; Sofie Pollin
• Keywords: Channel state information

The constant increase in wireless handheld devices and the prospect of billions connected machines has compelled research community to investigate different technologies which are able deliver high data rates, lower latency better reliability quality experience mobile users. One problems, usually overlooked by community, is that more require proportionally signalling overhead. Particularly, acquiring users’ channel state information necessary order for base station assign frequency resources. Estimating this with full resolution time generally impossible, thus, methods have be implemented reduce In paper, we propose a estimation method based on concept Gaussian process regression predict states varying user mobility profiles. Furthermore, present dual-control technique determine most appropriate prediction each keep packet loss rate below pre-defined threshold. proposed makes use active learning exploration-exploitation paradigm, allow controller choose autonomously next sampling point so exploration control space limited while still reaching an optimal performance. Extensive simulation results, carried out LTE-A simulator, show provide consistent gain, terms rate, users low average mobility, its efficacy reduced high-velocity then applied, impact analysed multicell network proportional fair maximum throughput scheduling mechanisms. Remarkably, it shown presented approach allows reduction overall over 90 % keeping 5 schedulers, as well 60 scheduling.