APEX: A Network-Native Time-Series Foundation Model for Forecasting and Anomaly Detection for Wireless Edge Operations
Researchers have developed APEX, a new transformer-based model specifically designed for time-series forecasting and anomaly detection in wireless network operations. Pre-trained on extensive data from thousands of production wireless networks, APEX demonstrates significant improvements over existing methods in predicting network degradation. The model is available in two versions: APEX-Large for cloud-based operations and APEX-Edge for efficient, on-device inference at the network edge. AI
IMPACT This model's network-native design and edge-capable version could improve proactive wireless operations and reduce latency for critical network tasks.