Proactive approach for preamble detection in 5G-NR PRACH using supervised machine learning and ensemble model.

Area of Science:

• Telecommunications Engineering
• Machine Learning Applications
• Wireless Communication Systems

Background:

• The physical random access channel (PRACH) is critical for initial user access in cellular networks.
• Achieving low latency in 5G PRACH is challenging with conventional methods.
• Signal collisions and dual peaks degrade PRACH receiver performance when multiple users share preamble signatures.

Purpose of the Study:

• To investigate machine learning classification techniques for proactive PRACH signal analysis.
• To identify the most effective model for mitigating PRACH collisions and dual peaks.
• To enhance the efficiency and reliability of 5G initial access.

Main Methods:

• Implemented supervised learning algorithms: Decision Tree Classification (DTC), Naïve Bayes (NB), and K-nearest neighbor (KNN).
• Utilized signal samples as big data for training and classification into 'peak' and 'false peak' categories.
• Developed a Bagged Tree Ensembler model to reduce variance and improve upon DTC performance.

Main Results:

• Supervised learning models were trained to classify PRACH signal outcomes.
• The Bagged Tree Ensembler demonstrated superior performance compared to individual classifiers.
• The Ensembler method effectively addressed the issue of signal collisions and dual peaks.

Conclusions:

• Machine learning, particularly the Ensembler method, offers a promising proactive solution for PRACH challenges in 5G.
• The Ensembler approach enhances system performance by reducing signal collision impacts.
• This study highlights the potential of advanced ML techniques for optimizing wireless network access.