Trade-Off Analysis of Hardware Architectures for Channel-Quality Classification Models

Area of Science:

• Computer Science
• Electrical Engineering
• Telecommunications

Background:

• Vehicular networks like Software-defined vehicular networks (SDVN) and Vehicular ad-hoc networks (VANETs) require channel quality assessment for adaptive behavior.
• Transmission channel quality impacts network performance, affecting bandwidth utilization and necessitating fault-tolerant schemes that can reduce speed and increase energy consumption.
• An efficient, low-power mechanism is needed to sense channel quality and automatically select optimal transmission channels.

Purpose of the Study:

• To analyze the trade-offs of hardware implementations for identifying high or low quality communication channels.
• To evaluate the effectiveness of four machine learning algorithms for real-time channel quality sensing and selection.
• To identify the most efficient and accurate algorithm for hardware deployment in vehicular networks.

Main Methods:

• Implemented and evaluated four machine learning algorithms: Decision Trees, Multi-Layer Perceptron, Logistic Regression, and Support Vector Machines.
• Performed a hardware implementation trade-off analysis to measure performance metrics.
• Focused on accuracy and efficiency (Mbps/LUT) for channel quality classification.

Main Results:

• The Decision Tree algorithm achieved the highest accuracy at 95.01%.
• The Decision Tree implementation demonstrated superior efficiency with 9.83 Mbps/LUT.
• A Decision Tree with a depth of five provided the optimal balance between accuracy and hardware resource utilization.

Conclusions:

• Hardware-based Decision Trees offer an efficient and accurate solution for real-time vehicular communication channel quality assessment.
• This approach enables adaptive channel selection, optimizing bandwidth use and potentially reducing the overhead of fault-tolerant schemes.
• The findings contribute to the development of more robust and energy-efficient vehicular communication systems.