Broadband Multimedia Networks Assignment

Broadband Multimedia Networks Assignment

Question 1
In MPEG-DASH server movie is stored as multiple files (Kapetanakis, Zampoglou,
Malamos, Panagiotakis & Maravelakis, 2014). Each file is segmented as different quality levels.
Because MPEG-DASH player facilitates the adaptive bit rate feature. This feature helps to select
the highest possible bitrate based on the current network strength etc. So that only the movie is
stored at different bit rates.
MPD stands for Media Presentation Description. It describes the segment information
like timing, resolution, etc. Here one or more representation of the video files are there in the
server. Here the bit rate is selected based on the network conditions and user preferences
(Kesselman, Kogan & Segal, 2010).
MPEG-DASH Server
One of the well sought and applicable bitrate streaming processes in the market which enables
the streaming of the media in high quality by the internet is MPEG-DASH; service delivered by
HTTP Servers. This is one of the first streaming processes that are of international standard and
works on adaptive bitrate HTTP. This server is sometimes confused with a transport protocol,
whereas the protocol used by this server is TCP (Kapetanakis et al., 2014) (Park and Kim, 2015)
(Thomas et al., 2017).
MPD (Media Presentation)
Segment information such as Media properties, URL, and Timings are described by a Media
Presentation Description Commonly termed as MPD. This MPD can be classified in a number of
1

ways according to the specification of the segment such as base, template, list, timeline and use
case. Segment are the parts which can bear any kind of data related to media, although the
certain set of instructions are to be used for two types of containers, A Mp4 file format which
lies on the ISO file as a base and MPEG, a type of system which uses 2 different transport
streams.
Question 2
200 kbps needs 1 sec. 400 kbps needs 2 seconds. This is for bit rate 200kbps
500 kbps needs 1 sec. 1000 kbps needs 2 seconds. This is for bit rate 500kbps
1000 kbps needs 1 sec. 2000 kbps needs 2 seconds. This is for bit rate 1000kbps
200 kb will be produced in 1 sec. 1600 kb will be produced in 8 seconds. This is for 200-bit rate.
500 kb will be produced in 1 sec. 1600 kb will be produced in 8 seconds. This is for500-bit rate.
1000 kb will be produced in 1 sec. 8000 kb will be produced in 8 seconds. This is for 1000 bit rate
seconds (Mustafa & Talab, 2016).
Time Requirement
Chunk size 400Kb 400/1600 0.25 sec
Chunk size 1000Kb 1000/1600 0.625 sec
Chunk size 2000Kb 2000/1600 1.25 sec
Chunk size 1600Kb 1600/1600 1 sec
Chunk size 4000Kb 4000/1600 2.5 sec
2

Chunk size 8000Kb 8000/1600 5 sec
The semantic gap is one of the toughest factors while creating segmentation for any given movie
towards a semantically correlated units. This does not feed you with information necessary to
correlate between the different shot and sometimes it is even harder to detect the scenes which
bear the capability of dynamic content. In this particular article, the method proposed is the
representation of key frames of the video shots by the local descriptors that do not vary. The
method proposed in this particular article is proved by the numerical analysis and states this
method results in higher detection rates. The analysis also opines that the given method will
highly likely to make a good balance between passion and recall.
This method proposed is temporal soothing that is carried out upon words histogram that is
extracted from the video shots. For every under consideration shot, any particular number of key
frames may be extracted and then these key frames are assessed upon the scale of local
descriptors that are invariant. Every possible descriptor may be joined and form a cluster and
then made a visual word of it. For every given word a histogram of visual words is computed
accordingly. This histogram later than managed according to their preceding or exceeding visual
words using a Gaussian kernel which helps in the preservation of the information effectively. By
adjusting these by the Gaussian kernel the detection of both scene and boundaries of any under
study video.
Question 4
3