Mixed streaming of video over wireless networks
In recent years, transmission of video over the Internet has become an important application. As wireless networks are becoming increasingly popular, it is expected that video will be an important application over wireless networks as well. Unlike wired networks, wireless networks have high data loss rates. Streaming video in the presence of high data loss can be a challenge because it results in errors in the video.Video applications produce large amounts of data that need to be compressed for efficient storage and transmission. Video encoders compress data into dependent frames and independent frames. During transmission, the compressed video may lose some data. Depending on where the packet loss occurs in the video, the error can propagate for a long time. If the error occurs on a reference frame at the beginning of the video, all the frames that depend on the reference frame will not be decoded successfully. This thesis presents the concept of mixed streaming, which reduces the impact of video propagation errors in error prone networks. Mixed streaming delivers a video file using two levels of reliability; reliable and unreliable. This allows sensitive parts of the video to be delivered reliably while less sensitive areas of the video are transmitted unreliably. Experiments are conducted that study the behavior of mixed streaming over error prone wireless networks. Results show that mixed streaming makes it possible to reduce the impact of errors by making sure that errors on reference frames are corrected. Correcting errors on reference frames limits the time for which errors can propagate, thereby improving the video quality. Results also show that the delay cost associated with the mixed streaming approach is reasonable for fairly high packet loss rates.
video, streaming, wireless, mixed
Master of Science (M.Sc.)