Multicore distributed processing architecture with Miss Table in radar systems for real-time severe weather analysis

Abstract

Although the first use of radar was for military purposes, radar technology has been introduced into many other fields including weather analysis. Today, radar is the primary tool for weather forecasting. To cover local, national, and global weather conditions timely and accurately, more powerful radar system is in need. Recent studies indicate that newly introduced multicore distributed architecture can provide high processing speed required by modern radar systems. In spite of the fact that multicore architecture improves performance, it poses some serious challenges including execution time unpredictability which is not acceptable for real-time weather analysis. In this work, we model and simulate a popular Intel-like multicore architecture with Miss Table (MT) and investigate the effectiveness of the multicore architectures for real-time severe weather analysis. Miss Table holds information about block addresses and helps improve execution time predictability and performance/power ratio. Experimental results show that execution time predictability can be further improved by applying techniques like cache locking with Miss Table. Therefore, we find multicore architecture with Miss Table very promising for future radar systems for real-time severe weather analysis. In addition, this kind of powerful radar systems can have significant role in assessing global climate change and developing future defense applications.