A CSI Prediction Scheme for Satellite-Terrestrial Networks

Guey Yun Chang, Chia Kai Hung, Chi Hao Chen

Research output: Contribution to journalArticlepeer-review

3 Scopus citations

Abstract

Recently, low Earth orbit (LEO) satellites have been suggested as a promising solution for global coverage of 5G networks. However, long propagation delay makes obtained channel state information (CSI) (which is employed for efficient spectrum utilization) outdated before usage. Note that CSI is closely related to elevation angle and relative location. Although elevation angle and relative location could be estimated by terrestrial device position and LEO ephemeris, maintaining these two informations requires a high cost for most IoT devices. Due to regular and high-speed orbital movement of LEO, for an arbitrary terrestrial device, LEOs repeat the same behavioral pattern (e.g., satellite rise and satellite set) again and again. Satellite rise and set result in different temporal correlations in elevation angle and relative location. By the aid of the temporal correlations mentioned above, we proposed a CSI prediction scheme for satellite-terrestrial networks without terrestrial device position and LEO ephemeris. Simulation results show that our scheme has a rather low prediction error under various LEO altitude/obit, terrestrial device location/mobility, ground environments, weather, and elevation angle when handover.

Original languageEnglish
Pages (from-to)7774-7785
Number of pages12
JournalIEEE Internet of Things Journal
Volume10
Issue number9
DOIs
StatePublished - 1 May 2023

Keywords

  • 5G network
  • channel state information (CSI)
  • recurrent neural network (RNN)
  • satellite-terrestrial

Fingerprint

Dive into the research topics of 'A CSI Prediction Scheme for Satellite-Terrestrial Networks'. Together they form a unique fingerprint.

Cite this