arXiv:2307.12417v2 Announce Type: replace-cross 
Abstract: While 5G New Radio (NR) networks offer significant uplink throughput improvements, these gains are primarily realized when User Equipment (UE) connects to high-frequency millimeter wave (mmWave) bands. The growing demand for uplink-intensive applications, such as real-time UHD 4K/8K video streaming and Virtual Reality (VR)/Augmented Reality (AR) content, highlights the need for accurate uplink throughput prediction to optimize user Quality of Experience (QoE). In this paper, we introduce UplinkNet, a compact neural network designed to predict future uplink throughput using past throughput and RF parameters available through the Android API. With a model size limited to approximately 4,000 parameters, UplinkNet is suitable for IoT and low-power devices. The network was trained on real-world drive test data from commercial 5G Standalone (SA) networks in Tokyo, Japan, and Bangkok, Thailand, across various mobility conditions. To ensure practical implementation, the model uses only Android API data and was evaluated on unseen data against other models. Results show that UplinkNet achieves an average prediction accuracy of 98.9% and an RMSE of 5.22 Mbps, outperforming all other models while maintaining a compact size and low computational cost.

تم تقديم UplinkNet كشبكة عصبية مدمجة مصممة للتنبؤ بمعدل نقل البيانات في الشبكات المستقلة 5G (SA)، باستخدام بيانات النقل السابقة ومعلمات RF من واجهة برمجة التطبيقات الخاصة بنظام Android. تم تدريب النموذج، الذي يحتوي على حوالي 4000 معلمة، على بيانات من العالم الحقيقي من طوكيو وبانكوك، مما يلبي الطلب المتزايد على التطبيقات التي تتطلب نقل بيانات مرتفع مثل بث الفيديو بدقة UHD ومحتوى الواقع الافتراضي/المعزز.

Se ha presentado UplinkNet como una red neuronal compacta diseñada para predecir el rendimiento de subida en redes 5G autónomas (SA), utilizando datos de rendimiento pasados y parámetros de RF de la API de Android. El modelo, que contiene aproximadamente 4,000 parámetros, fue entrenado con datos del mundo real de Tokio y Bangkok, abordando la creciente demanda de aplicaciones intensivas en subida como la transmisión de video UHD y contenido de VR/AR.

UplinkNet a été introduit comme un réseau de neurones compact conçu pour prédire le débit de liaison montante dans les réseaux 5G autonomes (SA), en utilisant des données de débit passées et des paramètres RF de l'API Android. Le modèle, qui contient environ 4 000 paramètres, a été formé sur des données du monde réel provenant de Tokyo et de Bangkok, répondant à la demande croissante d'applications intensives en liaison montante telles que le streaming vidéo UHD et le contenu VR/AR.

UplinkNet has been introduced as a compact neural network designed to predict uplink throughput in 5G Standalone (SA) networks, utilizing past throughput data and RF parameters from the Android API. The model, which contains approximately 4,000 parameters, was trained on real-world data from Tokyo and Bangkok, addressing the growing demand for uplink-intensive applications like UHD video streaming and VR/AR content.

UplinkNet: Practical Commercial 5G Standalone (SA) Uplink Throughput Prediction

Was this article worth reading? Share it

Whyline

NowKnow

Bifrost