Is burst loss worse than random loss for video transmission
Jimin Xiao, Tammam Tillo, Chunyu Lin
Abstract
Jimin Xiao, Tammam Tillo, Chunyu Lin
Abstract
Many papers have shown that burst loss causes more severe damage to the reconstructed video quality than random loss for the same packet loss rate. With traditional interleaving methods, the consecutive lost packet positions will be interleaved to random positions, and after interleaving, the PSNR of burst loss can be close to the PSNR of random loss for the same packet loss rate. In this paper, a novel slice reorder and packetization (SRP) approach is proposed which fully utilizes the temporal correlation between slices in successive frames. Theoretical analysis is presented that explains why with the proposed approach, burst loss can be better than random loss for the same packet loss rate. The effect of the proposed approach is validated with H.264/AVC coded video. The simulation results show that with the proposed approach, the PSNR of the burst loss can be up to more than 2dB higher than random loss without adding any redundant information.
A significance statement is not available in the OpenAlex record.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Many papers have shown that burst loss causes more severe damage to the reconstructed video quality than random loss for the same packet loss rate. With traditional interleaving methods, the consecutive lost packet positions will be interleaved to random positions, and after interleaving, the PSNR of burst loss can be close to the PSNR of random loss for the same packet loss rate. In this paper, a novel slice reorder and packetization (SRP) approach is proposed which fully utilizes the temporal correlation between slices in successive frames. Theoretical analysis is presented that explains why with the proposed approach, burst loss can be better than random loss for the same packet loss rate. The effect of the proposed approach is validated with H.264/AVC coded video. The simulation results show that with the proposed approach, the PSNR of the burst loss can be up to more than 2dB higher than random loss without adding any redundant information.
Key concepts: Interleaving, Packet loss, Computer science, Data loss, Transmission loss, Real-time computing, Algorithm, Network packet