Digital Signal Processing Reference
In-Depth Information
may fail either because of the bad propagation conditions over the channel or
because two or more stations transmit at the same time and collide. However,
an ARQ scheme is implemented: in the case of failure, a transmission is re-
peated until a maximum number of transmission attempts is reached.
3. THE H.264 VIDEO CODING STANDARD
We focus on the transmission of video data compressed according to the new
ITU-T H.264 standard. The compression scheme follows the general structure
of the ISO MPEG and ITU-T H.264 video coding standards, with some new
features to achieve a higher compression efficiency. Some of them are briefly
outlined in the following; refer to [8] and [13] for more details.
The base coding unit for transform coding is a 4x4 sample block. Mac-
roblocks are composed of 16 luminance blocks and 4 blocks for each chromi-
nance component. The transform coding is a separable integer transform with
essentially the same properties of the traditional Discrete Cosine Transform
(DCT). Regarding motion compensation, prediction using multiple reference
frames is possible.
Consecutive macroblocks are grouped into a
slice.
The
slice
is important
because it has the property to be independently decodable. This is useful to
subdivide the coded stream into independently decodable packets, so that the
loss of a packet does not affect the decoding of others (not considering the effects
of motion compensation).
One of the most interesting characteristics of the H.264 standard is the at-
tempt to decouple the coding aspects from the bitstream adaptation needed to
transmit it over a particular channel. The part of the standard that deals with the
coding aspects is called Video Coding Layer (VCL), while the other is the Net-
work Adaptation Layer (NAL) [7]. One of the developed NAL is aimed to the
problem of transporting data over an IP network using the Real-Time Transport
Protocol (RTP) [3], which is well suited for real time multimedia transmissions.
Complete separation between the VCL and the NAL is, however, difficult
to achieve. For instance, to improve error resilience, the VCL should create
slices of about the same size of the packets handled by the NAL —which, in
turn, should not split slices (a VCL entity) into different packets. Such cross-
layer approach would benefit the error resilience of the transmission because all
packets could be decoded independently.
In H.264 the subdivision of a frame into slices has not to be the same for each
frame of the sequence; thus the decoder can flexibly decide how to slice each
individual video frame. Slice should not be too short because that would cause a
