A Bandwidth-efficient AFEC-based Voice Quality Control Mechanism for Embedded Systems

• Main Authors: Li, Kai-Zhen, 李開振
• Other Authors: Huang, Yu-Lun
• Format: Others
• Language: en_US
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/98808561820894345370

碩士 === 國立交通大學 === 資訊科學與工程研究所 === 97 === As the massive deployment of packetized voice over the Internet, the voice quality has become an important topic in recent years. Besides the codec type, the voice quality can also be affected by network conditions, especially the network loss. To remedy this issue, many mechanisms were proposed to model or recover the loss in packet networks. Despite these mechanisms perform well in many cases, however, they are not suitable for real-time and interactive applications such as VoIP, where the voice quality is sensitive not only to network loss, but also to jitter and end-to-end delay. In this thesis, we propose a model to address the quality of packetized voice under network loss and redundancy for loss recovery. Based on this model, we present a control mechanism to recover the voice quality of VoIP in a lossy network with less bandwidth overhead. Our mechanism is built on Adaptive Forward Error Correction (AFEC) with Reed- Solomon codes and performed at the packet level to better reflect the real world loss pattern. We try to reduce the computing complexity of our mechanism so that it can be easily realized on embedded systems. As a proof of our work, we implement the proposed mechanism on a home gateway with VoIP capability. We also perform some experiments to verify the effectiveness of our model. The experiments show that by using the proposed mechanism, under the loss ratio up to 30%, the voice quality can still be maintained above the designated threshold with the best bandwidth efficiency compared to the traditional AFEC mechanisms.