Towards Intelligent Zone-Based Content Pre-Caching Approach in VANET for Congestion Control

Area of Science:

• Vehicular Ad Hoc Networks (VANETs)
• Machine Learning Applications
• Network Performance Optimization

Background:

• Vehicular ad hoc networks (VANETs) face challenges with network congestion and high response delays due to content requests.
• Location-based dependencies exacerbate congestion, making efficient content pre-caching crucial for timely data delivery and safety applications.
• Existing periodic data dissemination methods lack awareness of real-time road conditions and vehicle states.

Purpose of the Study:

• To propose a novel machine learning-based, zonal/context-aware content pre-caching strategy for VANETs.
• To enhance content placement and management within the VANET pre-caching framework.
• To improve early content prediction accuracy and reduce network latency.

Main Methods:

• Development of a machine learning model for zonal/context-aware content pre-caching.
• Implementation of three algorithms: zone selection, content dissemination, and supervised machine learning for pre-caching decisions.
• Evaluation of prediction accuracy, cache hit ratio, and average delay against existing techniques (CCMP, MLCP, PCZS+PCNS, RPSS).

Main Results:

• Achieved 99.6% early content prediction accuracy using supervised machine learning.
• Improved cache hit ratio by 13% compared to previous techniques.
• Demonstrated an average 16% improvement in prediction accuracy and significant reduction in average delay with increasing node density.

Conclusions:

• The proposed machine learning-based pre-caching strategy effectively addresses content delivery challenges in VANETs.
• The model significantly enhances content prediction and placement, leading to reduced network congestion and improved performance.
• The strategy offers a robust solution for timely content delivery, crucial for safety and efficiency in urban vehicular environments.