Information Technology Reference
In-Depth Information
underlying network protocol layers in order to avoid redundancy for efficient re-
source utilization [25]. This problem can be resolved by partitioning of video bit
stream and prioritizing its parts by sending them via multiple bearers with differ-
ent characteristics, such as different channel coding, and modulation schemes. In
order to implement such approach, it is necessary to separate the encoded bit
stream optimally into a number of sub-streams, such as in [26].
Joint source-channel coding principles have been adopted here for achieving
optimized transmission. Different segments of the 3D video bit stream are priori-
tized to provide maximum security to the most vulnerable parts. The encoded data
is separated into a number of sub-streams based on the relative importance of dif-
ferent video packets, which is calculated using the estimated perceived distortion
of packets at the encoder. The importance level has a further biasing factor of
depth information related to that particular object. The next section explains the
proposed prioritization algorithm.
3 Cross-Layer Design for Optimized 3D Video Transmission`
Cross layer design techniques for jointly adapting the coding/transmission tech-
niques across all network layers have demonstrated considerable performance
gains for multimedia applications over wireless networks [8]. This approach is fol-
lowed in this work for optimized transmission of 3D video content over WLAN.
3.1 Design of Cross Layer Video Communications System
In a typical home entertainment scenario, a group of users may simultaneously ac-
cess a variety of services, e.g. VoIP, data downloads, and video streaming. The
WLAN router selects the appropriate channel for transmission of voice, video or
data packets. Such a home WLAN enabled environment is shown in Fig. 4. 3D
videos are available at the source server. The objective is to prioritize data across
various layers of transmission chain for optimized 3D video content (color plus
depth) transmission. A segmentation module extracts background and foreground
information from the 3D video color stream. In order to separate the foreground
from the background for streaming application, we need faster segmentation solu-
tion. Details of the segmentation process used in this work are provided in the next
subsection.
Segmented data is then prioritized based upon its estimated distortion at the re-
ceiver. The distortion estimation module calculates optimal QP values for the
three streams (the depth stream, the background object stream and the foreground
object stream) for different network conditions to minimize distortion. The mod-
ule also performs mapping of each video packet into one of the streams at the ap-
plication layer level according to their expected distortion, and the error rate of the
communication channel over which they are to be transmitted. The selection pro-
cedure is discussed in subsection 3.3.
