Low-Power Hardware Design of Binary Arithmetic Encoder in H.264

Ben Hamida Asma ,Nedra Jarray,Zitouni Abdelkrim

doi:10.14569/ijacsa.2017.080756

Abstract

Context-Based Adaptive Binary Arithmetic Coding (CABAC) is a well-known bottleneck in H.264/AVC, owing to the highly serialized calculation and high data dependency of the binary arithmetic encoder. This work presents a hardware architecture for the sub-module binary arithmetic encoder of the CABAC. Moreover, a clock gating technique is inserted into our design for power saving. An FPGA design of the proposed architecture can work at a frequency up to 268 MHz on Virtex 5. The suggested design can achieve 17% of power consumption saving, which allows it to be applied for low power video coding applications.

Highlights

In the H.264/AVC standard, two entropy encoders are defined: Context-based Adaptive Variable Length Coding (CAVLC) and Context-based Adaptive Binary Arithmetic Coding (CABAC)
It is to clear that the bypass Binary Arithmetic Encoder (BAE) operates at a higher clock frequency compared to the regular mode
Our design uses a higher frequency compared to the work [5], which was implemented in the same FPGA technology

Summary

Introduction

In the H.264/AVC standard, two entropy encoders are defined: Context-based Adaptive Variable Length Coding (CAVLC) and Context-based Adaptive Binary Arithmetic Coding (CABAC). The CAVLC is a low-complexity entropy coding technique based on the use of switched contextadaptively sets of variable-length codes. Compared to CABAC, The compression efficiency improvement is obtained at the cost of an inevitable complexity overhead. Software-based complexity analysis results show that switching from CAVLC to CABAC usually leads to complexity increasing by 25–30% for encoding and 12% for decoding. A hardware implementation of CABAC encoder is always required. The bit-serial nature of the CABAC algorithm and the strong data dependency between contiguous bits make it hard to improve the throughput and to parallelize the encoding process

Objectives

Methods

Results

Conclusion