Mobile Television

1
Mobile
Television

• 
  Valluri Kalyan
• 
  April 26

2
Topics Covered

• Introduction
• DVB-H Standard
• Issues
• Proposed Solutions
• Conclusion
• References

3

• TheDVB-H(DigitalVideoBroadcasting for handheld)
  is an extension of DVB-T. Delivery of rich
  multimedia content was missing in DVB-T and DVB-H
  technology enhanced DVB-T by adding the feature
  of multimedia content.
• The physical layer has four extensions to the
  existing DVB-T physical layer. First, the bits in
  transmitter parameter signaling (TPS) have been
  upgraded to include two additional bits t

4

• Second, a new 4K mode orthogonal frequency
  division multiplexing (OFDM) mode is adopted for
  trading off mobility and single-frequency network
  (SFN) cell size,
• Third, a new way of using the symbol interleaver
  of DVB-T has been defined
• Finally, the addition to DVB-T physical layer is
  the 5-MHz channel bandwidth

5
A Conceptual DVB-H Receiver
6
Issues

• Resolution Requirements
• Voltage Scaling
• Prioritizing Data In MPEG-4

7
Resolution Requirements

• The Major problem for Mobile Television Devices
  such as handheld PDAs ,Phone etc is the extent
  unto which the image can be reduced, without
  affecting the perceived quality of the image.
• By doing so bandwidth also can be saved to a
  greater extent
• There is a tradeoff between screen size and
  portability of the device

8

• Reducing the image size can have two opposing
  effects
• Firstly a smaller image resolution will
  facilitate bitrate savings as there is less
  information to be coded.
• Hence for a fixed encoding bitrate,it is possible
  that the percieved quality is increased as
  bandwidth budget per pixel is increased when the
  image resolution is reduced

9

• A negative impact of reducing image size is that
  we have only a fewer pixels to represent the
  information to the user.
• This creates problem in content types such as
  sports.
• We will have only fewer pixels available to
  display important details, for instance details
  about the location of the football.

10
This phenomenon is illustrated in the graph below
11

• Another important factor for mobile Tv is how the
  bitrates allocated to audio and video streams
  interact to affect the perceived Qos.
• Multimodal quality for generic clips is predicted
  by a regression equation of the form
  AQVQ(AQVQ) where AQ and VQ stand for Audio and
  Video quality respectively.
• For high motion clips a reduced equation with
  just the interaction term (AQVQ) gave the best
  predictions of multimodal quality

12

• Now we examine the effects of varying image
  resolution and encoding bitrate on service
  acceptability.
• When the bitrate available for video content is
  scarce, reducing the image resolution could free
  up valuable encoding bitrate to improve perceived
  quality.
• Similarly when bitrate is abundant there will be
  less loss of detail as the image resolution is
  reduced

13
Image resolution effects depend on the content

• Careful analysis in the form of a survey
  indicates different acceptable ratings for varied
  content types
• News
• In this case the largest image failed to obtain
  highest acceptability ratings. In fact
  acceptability increased when the resolution was
  reduced to 208156.

14
Acceptability of News Content
15

• Number of reasons such as text details, the
  headline text, clock, the logo or the captions
  for the people being interviewed and factors such
  as facial details, shot types etc were the main
  reasons for unacceptability by the viewers of the
  news channel.

16
Sports Content

• Live sporting events like the NBC playoff still
  need to takeoff with high acceptability ratios.
• The main problem in such sporting events is loss
  of object detail. The major problem being
  locating the ball and identifying the players.
• The second most common complaint is about certain
  short types specifically long shots of the
  entire pitch to be seen on the small screen

17

• Another prominent drawback is inability to read
  text detail about the teams and the scores.
• To a large extent these problems can be dealt by
  increasing both Image resolution and Video
  bitrate.

18
Acceptability of Sports Content
19
Animation Content

• A reduction in image size had very little impact
  on the acceptability factor.
• Problems such as identifying animal species when
  the image resolution was very small and images
  being very dark persisted but are not very
  prominent.

20
Why Was Quality Unacceptable
21

• Across content types, the effect of reducing
  image resolution is more pronounced when
  bandwidth is abundant.
• The best feasible solutions in such scenarios
  would be to encode at the largest image
  resolution possible for any content type
• This also has two exceptions News and very low
  bandwidth Music videos.

22

• More effective way is to stream text information
  separately to the device.
• Protocols such as the SMIL should be integrated
  into mobile TV production process to synchronize
  text and video streams to mobile devices.

23

• Still evaluation is required to understand how
  audio and video qualities interact to bias users
  perception of video quality acceptance.

24
Voltage Scaling

• Saving on energy is the most critical factor for
  mobile multimedia applications.
• A challenging problem is how to minimize energy
  consumption while provisioning QoS.
• Previously implemented DVS (Dynamic Voltage
  Scaling) algorithm reduces the CPU energy by
  reducing the CPU speed based on the application
  CPU demand.

25

• But the effectiveness of such an algorithm is
  dependent on the prediction of application
  demand.
• Over prediction may waste energy and under
  prediction may degrade the application
  performance.
• The solution put forth here is PDVS (Practical
  voltage scaling) algorithm.

26

• The PDVS extends traditional real-time scheduling
  by deciding what execution speed in addition to
  when to execute what applications.
• PDVS makes these decisions based on
• 1.Discrete speed levels of the CPU
• 2.The total power of the device at different
  speeds.
• 3.The probability distribution of CPU demand of
  multimedia applications.

27

• The goal of PDVS is to minimize the total energy
  of the mobile device while providing soft
  performance guarantees to each multimedia task.
• The purpose is to save energy since CPU may run
  slower without affecting application performance.
• Initially we can set a uniform sped for all
  concurrent tasks until the task set changes.

28

• Assume there are n tasks and each task is
  allocated Ci cycles per period Pi. The total
  number of cycles demanded per second (Hz) by all
  tasks
• The uniform speed can be the lowest speed that is
  no less than the total demand i.e
• If each task uses its allocated cycles exactly,
  this will minimize CPU energy.

29

• Multimedia tasks often change their cycle demand
  dynamically. A task may complete a job earlier
  even before utilizing the allocated cycles.
• This causes the CPU to be idle resulting in
  wastage of power during the idle time.
• The PDVS dynamically adapts the CPU speed of each
  job execution in a manner that minimizes the
  total energy consumed during job execution while
  bounding the jobs execution time.

30

• To attain this a time budget is allocated for
  each job. Consider there are n concurrent tasks
  and each task is allocated Ci cycles per period
  Pi then the ith task is allocated
• time units per period Pi.
• But another inherent problem is how to
  dynamically adapt the execution speed for each
  individual task.

31

• To achieve this PDVS sets a speed for each of the
  cycles allocated to the job.
• If a cycle x,1ltxltc is executed at speed s(x),the
  execution time is 1/s(x).
• Energy consumed by device in this time interval
  is
• Here ps(x) represents the total power consumed by
  the device.

32

• Expected energy for this task is given by the
  equation
• Where F(x) is the tail distribution function.
• The speed adaptation problem can be formulated as

33

34

• Although we have arrived at a solution,it is not
  quite feasible as number of allocated cycles C
  may be very large.
• Secondly overhead for optimization may be
  unacceptably large.
• Hence a piece wise approach is followed.
• We use set of points b1,..bk1 to divide the
  allocated cycles into k groups each with a size
  gi.

35

• Thus we get a new constrained optimization
  equation as

36

• PDVS stores speed schedule of a task into its
  process control block and updates the speed
  schedule when the tasks time budget changes.

37
Architecture of PDVS Implementation
38
Prioritization of Data in MPEG-4

• All the applications on mobile tv require a real
  time transmission of video data over fixed and
  mobile networks with varying bandwidth
  requirements and error rate characteristics.
• These signals must be compressed before
  transmission in order to optimize the required
  bandwidth to provide a multimedia service.

39

• As MPEG-4 coded video data is highly sensitive to
  information loss and channel bit errors, the
  decoded video quality is bound to suffer at high
  channel bit error rates.
• Coded bit streams are transmitted in form of
  packets and the enclosed video payloads are
  exposed to channel errors and excessive delays
  and thereby information loss.

40

• The worst effect of bit error occurs when
  synchronization is lost and the decoder is no
  longer able to identify the received information
  belongs to which part of the frame.
• Real time video transmissions are sensitive to
  time delays, therefore there is no point of
  retransmitting the erroneous video data.

41

• To overcome these issues one of the approach is
  multi-layered video coding.
• The compressed bit stream of each video object
  plane (VOP) in the video sequence consists of a
  number of layers.
• There are two layers namely the base layer and
  enhancement layers.
• The base layer contains essential information
  regarding texture reconstruction.

42

• A similar method employed to improve the quality
  of video transport over networks is the
  prioritization of different parts of the video
  bit stream by sending data as different separate
  streams.
• This enables video encoder to demand the network
  to send the data using channels with different
  priorities.
• Thereby allocating more important and error
  sensitive data to more reliable and secure
  channels.

43

• MPEG-4 detects entities in the video frame that
  the user can access and manipulate thereby
  providing user with content based functionalities
  for processing and compression of any video scene
• The entities are called video object plane.
• The diagram shows Vide object plane consists of
  Vop headers and video packets.

44
a) Video object plane b) Video object
45

• The Vop and Vp contain synchronization code and
  compression parameters.
• Each video frame starts with a start code
• Start code are unique combination of bits that
  never occur in the video data.
• The MPEG-4 in addition to coding the texture and
  motion information as in traditional block based
  video coders,it also codes the shape of each VOP.

46

• Errors have detrimental effect on decoded video
  quality due to interdependencies of video data.
• Synchronization word error is caused by bit rate
  variability characteristic, the decoder in such
  cases skips all the correct bits waiting to
  recover the state of synchronization.
• This can be eliminated by placing the coded video
  data into regular sized packets, with
  resynchronization words between each packet

47

• MPEG-4 data partitioning succeeds in ensuring
  that much of the packet data is not very
  sensitive to error.
• An example of a walking person sequence is used
  to analyze the error sensitivity of data in first
  and second partitions of an MPEG-4 video packets.

48
a)Error free sequence b)shape data c)motion data
d)Texture Data
49

• Here we can see that texture errors can be
  concealed, the concealment of motion and shape
  data results in reconstructed frames contain a
  high degree of distortion.
• Corruption of texture has negligible effects
• Whereas shape data proves to be highly sensitive

50

• We now examine as to how prioritization is
  effective.
• In this case we consider a GPRS data is
  transmitted over packet data traffic channel
  after being error protected using one of the
  channel protection schemes CS-1,CS-2,CS-3 or
  CS-4.

51

• To eliminate these problems video frame is MPEG-4
  encoded and separated into two segments with high
  priority and low priority.
• Then the video data are produced by the encoder
  and the RTP/UDP/IP protocols are encapsulated to
  them.

52
Two separated segments
53

• The diagram below shows the objective quality for
  single and partitioned stream transmissions.

54
Conclusion

• The concept of Mobile Tv is still in its naïve
  stages and many issues related to broadcast,
  handoff, uninterrupted transmission still require
  further study for successful takeoff in the near
  future.
• Mobile Television faces the challenge of being
  compared with traditional television.

55

• Though it is very difficult to match the
  standards, maximum acceptability can still be
  achieved by enhancing performance to the peak
  levels and banking on the concept of being mobile
  and connected at any place.

56
References

• 1. DVB-H digital broadcast services to
  handheld devices Faria, G. Henriksson, J.A.
  Stare, E. Talmola, P. Proceedings of the IEEE
  Volume 94, Issue 1, Jan. 2006 Page(s)194 - 209
• 2. Systems 2 mobility and video Can small be
  beautiful? assessing image resolution
  requirements for mobile TV Hendrik Knoche, John
  D. McCarthy, M. Angela Sasse November 2005
  proceedings of the 13th annual ACM international
  conference on Multimedia MULTIMEDIA '05
  Publisher ACM Press
• 3. Prioritisation of Data Partitioned MPEG-4
  for Streaming Video in GPRS Mobile Networks
  Jafari,M.Kasaei,S. Internet, 2005.The First
  IEEE and IFIP International Conference in Central
  Asia on 26-28 Sept. 2005

57

• Performance Analysis and multimedia over
  WirelessPractical voltage scaling for mobile
  multimedia devicesWanghong Yuan, Klara Nahrstedt
  October 2004  Proceedings of the 12th annual ACM
  international conference on Multimedia
  Publisher ACM Press