AI Based Short Term Prediction of Q-Band Rain Attenuation

Kamolideen Abolarinwa, Ali Roula, Abdulkareem Sarki Karasuwa, Ifiok Otung

Research output: Contribution to conferencePaperpeer-review

Abstract

The increasing demand for high-throughput broadband services continues to drive the use of Extremely High Frequency (EHF) bands for satellite communication [1]. However, despite the advantages offered by these bands, rain attenuation remains a challenge to the full exploitation of EHF bands for communication. To address this challenge, various Fade Mitigation Techniques (FMTs) have been developed to enhance the robustness of signals against rain attenuation. Effective operation of these FMTs relies on real-time information about link conditions, enabling configuration adjustments and ensuring high Quality of Service (QoS) [2].

This study focuses on the application of Artificial Intelligence (AI), specifically machine learning, for short-term rain attenuation prediction in the Q-band. Our approach incorporates both rain attenuation measurements and meteorological data, including temperature, atmospheric pressure, wind speed, wind direction, relative humidity, and rain rate. The data was collected over a two-year period at Chilbolton Observatory in Southern England.

Although AI has been acknowledged for its potential in accurate instantaneous rain attenuation prediction [3], significant challenges persist. The delay associated with combined forward and return links impacts the timeliness of link attenuation reporting to the gateway. Additionally, the unavailability or outdated nature of historical rain attenuation data during satellite communication operations poses a challenge for rain attenuation prediction [4].

Therefore, our investigation focuses on identifying the relevant factors among the various meteorological data that can enhance prediction accuracy. We aim to simplify and improve predictions by discarding less significant factors. Furthermore, we explore the influence of past data on prediction accuracy, determining how far back in time historical data can impact predictions. Our ultimate goal is to define an AI model that selects the appropriate combination of meteorological and attenuation data for accurate short-term rain attenuation prediction on a satellite link. This prediction model will serve as a foundation for designing a fade mitigation strategy that optimally allocates satellite system resources based on predicted attenuation.
Original languageEnglish
Publication statusAccepted/In press - 24 Jul 2023
Event40th International Communications Satellite Systems Conference (ICSSC) and its Colloquium: Direct Satellite to Cellular Mobile Systems - Midland Hotel, Bradford, United Kingdom
Duration: 23 Oct 202326 Oct 2023
Conference number: 40
https://www.kaconf.com/

Conference

Conference40th International Communications Satellite Systems Conference (ICSSC) and its Colloquium
Abbreviated titleICSSC
Country/TerritoryUnited Kingdom
CityBradford
Period23/10/2326/10/23
Internet address

Fingerprint

Dive into the research topics of 'AI Based Short Term Prediction of Q-Band Rain Attenuation'. Together they form a unique fingerprint.

Cite this