Time is near when we will be able
to walk in our homes, or anywhere around the world, and seamlessly
access our preferred movie or music independently of where we
are, and where the content is physically stored or created. In
some limited examples, this has already started to happen.
Yet, it takes more that just bundling a multimedia codec and
a networking device to make an optimal multimedia streaming system.
Complex interactions need to be carefully taken into account,
and all aspects of this new kind of devices must be evaluated.
The particular nature of still images, video and audio signals,
the best-effort nature of the Internet Protocol, the lossy nature
of most wireless links, the need to protect content from unauthorized
fruition pose new challenges to the engineering and business community.
New scalable, interoperable, adaptive multimedia codecs, secure
transport and access rights management, SW application stacks
and APIs, content metadata descriptions, networks, wired and wireless
links, implementation architectures: all these components are
needed to make the system work.
In this ST Journal issue we try to give an insight on some of
these topics with the publication of where the state-of-the-art
research is today.
We start with a global adaptation framework, by M. van der Schaar
et al., that represents what could be the end of the process,
with automatic adaptation of terminals to the network and processing
power conditions. Towards this end, one of the key building blocks
we need is a scalable video coder. Beside current standardization
efforts (MPEG SVC), original works in wavelet scalable video coding
show the full potential of scalability, as discussed in the contribution
by N. Adami et al.
Yet, we have to face the fact that it will take a few years before
these schemes, although very promising, will enter our houses.
Today our video terminals (at best!) understand MPEG-4 AVC, and
if we want to improve video performance today we need to work
around that. The paper by D. Lefor et al. describes how one can
adapt an MPEG-2 AVC video transmission to different networks/terminals
by means of very efficient transcoding algorithms. P. Baccichet
et al. describe what one can do to mitigate data losses that unavoidably
occur in wireless transmission by means of the MPEG-4 AVC’s
FMO tool and improved error concealment.
We have seen that we can try to actively adapt our transmission
to achieve the best possible performance. An alternative approach
commonly found in the literature is to insert explicit redundancy
into the stream in order to increase its passive resistance to
losses. The implied trade-off is to compress less, taking advantage
of the fact that each bit, if lost, can be recovered by other
correlated bits, assuming the total amount of losses is below
a given threshold. There are two approaches one can take: introduce
redundancy at source level (as proposed by R. Bernardini et al.),
or introduce it at channel coding level (e.g., by using, LDPC
coders, as described by M. Rovini et al.).
Facing these new challenges does not mean that one can neglect
the classical algorithmic and architectural cross-optimization
problems. Finding the best cost/performance implementation is
always instrumental toward the success of any technology. D. Alfonso
et al. discuss the implementation optimizations for motion estimation
in the latest MPEG-4 AVC coder.
In order to have true multimedia applications, we need to address
audio as well. The same points we discussed for video still hold:
adaptation of the signal to the terminal/network conditions, passive
resilience by the receiver, and optimal implementation challenges.
The first topic is presented by A. Vitali et al., who discuss
compressed domain MP3 bitstream reshaping. Effects of network
jitter/losses are mitigated by one single algorithm, adaptive
playout, as described by A. Servetti et al., who also describe
implementation into a current STMicroelectronics’ SoC. Finally,
multi-processors implementation of an AC3 decoder is presented
by Y. K. Soni et al.
Multimedia streaming is such a broad and fascinating topic that
we couldn’t possibly give a complete presentation of the
challenges and state-of-the-art solutions in a single issue of
this journal, but we hope we have at least given you a glimpse
of some of these topics and instilled in you the technical curiosity
to delve deeper into the literature on the subject, and perhaps
inspired you to became yourself a researcher, or at least an educated
user of this technology! |
