Low-Complexity and Hardware-Friendly H.265/HEVC Encoder for Vehicular Ad-Hoc Networks

Area of Science:

• Computer Science
• Electrical Engineering
• Telecommunications

Background:

• Real-time video streaming in vehicular ad-hoc networks (VANETs) presents significant computational challenges for road safety.
• High Efficiency Video Coding (HEVC) is crucial for efficient video transmission but demands high computational resources.

Purpose of the Study:

• To reduce the computational complexity of HEVC encoders specifically for VANET environments.
• To enable efficient real-time video streaming for critical road safety applications.

Main Methods:

• Developed a coding tree unit depth decision algorithm using a novel spatiotemporal neighborhood set to control the depth search range.
• Implemented a Bayesian classifier for prediction unit decision in inter-prediction, utilizing a Gibbs Random Field model for prior probability calculation.

Main Results:

• Achieved significant reductions in encoding time, up to 63.96% compared to HEVC reference software HM16.0.
• Maintained a low increase in Bjontegaard delta bit-rate, averaging only 0.76-0.80%.

Conclusions:

• The proposed HEVC encoder algorithm offers a low-complexity and hardware-friendly solution for video codecs in mobile vehicles.
• This optimization is vital for deploying real-time video streaming in VANETs for enhanced road safety.