Multimedia Communications: Models of Channels and Networks

1
Multimedia CommunicationsModels of Channels and
Networks

• mmc7.01

2
Benefits of Models
Simplification of Complex Problems Scientific
Understanding of Performance and Fundamental
Limits Simulations of Systems Comparisons and
Calibrations of Algorithms
mmc7.02
3
Examples of Models in Communications
Analog Digital Channels Shannons
Formula Bit Errors Signal
Power Networks Multimedia Traffic Packet
Losses OSI Model Sources
Entropy Activity Error
Sensitivity Users Mobility Interactivity
mmc7.03
4
Shannons Formula
C W log (1 SNR) where C channel
capacity W channel bandwidth
SNR signal-to-noise ratio log base 2
mmc7.04
5
Shannons Limit on Error Rate
mmc7.05
6
Signal Power Model0.375 inch CoAx Cable
mmc7.06 Ahamed and Lawrence 97
7
Signal Power ModelWireless Channel
mmc7.07 Goodman 97
8
Multipath Fading
mmc 7.08 Steele 92
9
Multipath Fading Models
K Fraction of Power in Dominant Path
mmc 7.09 Steele 92
10
Interference Models

• Adjacent Channel Interference Minimized in
  Cellular Radio Significant in AM and FM Radio
• Co-Channel Interference Minimized in
  Cellular Radio Non-negligible in AM and FM
  Radio
• Often modeled as noise in CDMA, Cable Systems
• Advanced techniques use interference cancellation

mmc7.10
11
Cellular Radio Plan
mmc7.11 Goodman 77
12
Gilbert Model for Bursty Errors
Prob (losing block) PGB / (PBG PGB)
mmc7.12
13
Network Traffic Models
Systematic Patterns Long Term and Short Term
Variations Poisson Model Autoregressive
Model Self-similar (Fractal) Model Dependence on
Media and Media Compression
mmc7.13
14
Packet Voice Modeling
Two-state model
?
Talk spurt (active)
Silence (inactive)
?
15
Packet Voice Modeling
Composite model for N multiplexed voice sources
...
...
Probability that i sources are active at the same
time
16
Markov Model
17
Markov Model
? M 0
18
Poisson Process

• Oldest traffic model
• Inter-arrival times exponentially distributed

• An Inter-arrival times
• Counting process satisfying

• N(t) Number of arrivals in an interval of length
  t
• The number of arrivals in disjoin intervals is
  statistically independent

19
Video Traffic Modeling
First-order autoregressive model
Where ?(n) bit rate (bits/pixel) of frame
n w(n) stationary i.i.d. WGN with Ew ? a, b
constants
20
Video Traffic Modeling

• Advantages of first-order autoregressive model
• Simple representation of coded video sources
  statistics
• Resembles the statistics of videoconference/videop
  hone signals
• Disadvantages
• Cannot be readily applied to the determination of
  network parameters (buffer sizes, delays)
• Does not capture the effect of scene change in
  video sequences
• Solution Markov-Modulated Poisson Process (MMPP)

21
Source Models
Unequal error-sensitivity Unequal
delay-sensitivity Unequal priority Constant
Quality-Variable Bit Rate Constant Bit Rate-
Variable Quality Adaptive Bit Rate
mmc7.14
22
QoS Models
Best Effort Guaranteed QoS
mmc7.15
23
User Models
Up-link and Down-link Usage Interactivity Patience
and Attention Span Quality on Demand
mmc7.16
24
Open System Interconnect Model

• Application Layer
• Presentation Layer
• Session Layer
• Transport Layer
• Network Layer
• Data Link Layer
• Physical Layer

mmc7.17