Transcript Media Streaming over Power-Controlled Wireless Links

Control Mechanisms for Video
Streaming Wireless Links
Athina Markopoulou
Electrical Engineering Dept.
Stanford University
1
Real-Time Multimedia over Packet Networks
 Characteristics
•
Continuous Stream Playout, Real-Time
t
src
ideally
t
rcv
t
src
network
t
rcv
loss
jitter
delay
 Requirements
•
low loss, delay, delay jitter
2
Multimedia - Networking
Application
Network


Problems at the interface between multimedia
applications and underlying network
Control mechanisms
•
•

in the network and/or
at the end-systems
Challenges depend on underlying network(s)
3
The Bigger Picture
2
backbone
1
wireless
4
Media Streaming over Wireless Last-Hop
Wireline
Server

Mobile
Terminal
Streaming to
•

Access Point,
Server/Proxy
Wireless
laptops, cellphones, PDAs, wireless TV displays
Challenges
•
•
limited resources, time variation
strict application requirements
5
Example
Ideally
Tx
Tx
Rx
Rx
Play
Play
Tx
Control at the Tx
Tx
Rx
Rx
Play
Play
Over wireless
Control at the Rx
6
Problem Statement
N
p
i
Tx

Scenario

Objective
•
•
•
•
•
r
Rx
pre-stored media content at Tx.
interference i, according to a Markov chain with
deliver and play entire content
maximize the playout quality
minimize the power cost
7
System State and Controls
n
p
Tx
i
b
r
s(p,i) Rx
(p,r) = system controls in current time slot
(n,i,b, r’) = system state in current time slot
n = remaining packets at Tx
i = channel interference
b = available packets at Rx
r’ = playout rate in previous slot
8
System Controls at Tx
n
p
i
s(p,i) Rx
Tx
Control p: transmission power in current slot
•
s(p,i) : probability of successful reception
1
0.9
0.8
0.7
0.6
s(p,i)

r
b
0.5
0.4
•
Power Cost: Φ = p
–
–
battery lifetime
interference stress
0.3
0.2
0.1
0
0
2
4
6
8
10
SIR p/i
9
Dynamic Programming Formulation

System Evolution

Define
to be the
minimum expected cost-to-go
from n=N….
Power ·W + Quality
(n,i,b;r’)
·W
(p,r)
s(p,i)qij
(n-1,j,b+1-r;r)
(1-s(p,i))qij
(n,j,b-r;r)
… until n=0
10
Computing the Optimal Control
 A stationary optimal solution (p*,r*) exists
and can be obtained by value iteration
 Optimal policy:
•
•
table p*(n,i,b;r’) and r*(n,i,b;r’)
obtain offline & store in lookup table
11
Special Cases: Individual Controls


Scenario
Control at Tx
Control at Rx
No control
fix p
fix r=R
Power Control
adapt p
fix r=R
Power Control &
Re-buffering
adapt p
r in {0, R}
Playout Control
fix p
r in {0, r1, r2…R}
Re-buffering
fix p
r in {0, R}
Joint Control
adapt p
r in {0, r1, r2…R}
Similar formulations – obtain optimal policies
Compare: no control, special cases, joint control
12
Power-Quality Tradeoff (1)
No control
Better Performance
13
Power-Quality Tradeoff (2)
No control
Playout Only
14
Power-Quality Tradeoff (3)
No control
Playout Only
Power Only
Power+Rebuffering
15
Power-Quality Tradeoff (4)
No control
Playout Only
Joint control
Power Only
Power+Rebuffering
16
Heuristics


Why heuristics?
Justified vs. ad-hoc heuristics
•

mimic properties of optimal control
Steps
•
•
•
Power-only heuristic
Playout-only heuristic
Joint power-playout heuristic
17
Power Heuristic
n
p
i
r
Fix playout r=R
s(p,i) Rx
Tx

Optimal power:

Backlog pressure X(n,i,b)
•
b
n=N
has structural properties:
X

^
Heuristic: approximate X
•
mimicking those properties
n=1
b
18
Playout - today
n
p


i
fix p
Tx
B2
s(p,i)
s(i)
Rx
B1
r=R/2
r=R
b
r
r=0
Purpose: choose r(b)
Fixed Threshold Heuristic
•
•
L thresholds for buffer occupancy b
r
r=rl
r=0
Bl
r=R
Bl+1
BL
b
19
Playout Heuristic
B2
i
s(i)
Rx
r=R
B1
r=R/2
b
r
r=0

Observation: channel not taken into account yet

Adaptive Threshold Heuristic
•
•
adapt rate
and adapt thresholds
20
Joint Power-Playout Heuristic
n
p
Tx

•
•

•
•
•
i
s(p,i)
b
r
Rx
Tx side:
^
compute X(n,b)
compute power p:
Rx side:
estimate i, compute p and s(p,i)
adjust thresholds, compute playout r
feedback to Tx
21
Joint heuristic performs well
Optimal Playout
Optimal Power+Rebuffering
Joint Heuristic
Joint Optimal
22
Demo: no-control vs. joint heuristic



For the same
interference scenario
For the same power
consumption
Compare the playout
quality
original
no control
Joint heuristic
23
Comparison Details

Controls off

Joint Heuristic
24
Wireless Video - Summary

Contributions
•
•
•

Joint power-playout control
Modeled in a dynamic programming framework
Developed simple, efficient heuristics
Extensions
•
•
•
•
Additional Channels and Responsive Interference
Additional Controls
Content-Aware Control
Apply to protocols (802.11h)
“Joint Power-Playout Control Schemes for Media Streaming over Wireless
Links”, in IEEE Packet Video 2004, Markopoulou joint work with Y.Li,
N.Bambos, J.Apostolopoulos
25
Extension: adding more controls
control
scheduling
n
R(t)
Tx

b
Rx
Additional Controls:
•
Tx: control scheduling
–
•
how many units to transmit and which to drop
Rx: motion-aware playout
–

content-aware
playout (r)
slowdown video scenes with low or no motion
Results:
•
•
trade-off: playout speed variation vs. distortion
effect of playout variation is less perceived
“Joint Packet Scheduling and Content-Aware Playout Control for Media
Streaming over Wireless”, invited paper in IEEE MMSP 2005, A.
Markopoulou joint work with Y.Li, N.Bambos, J.Apostolopoulos
26
Example of Motion-Aware Playout
Motion-aware playout
Motion-unaware playout
27
Future Directions

Multimedia over IP
•
•

Network Dependability
•
•

Cross-layer optimization
Content distribution
From traditional QoS to Reliability & Security
Measurements for diagnosis and control
Network shared by independent selfish entities
•
•
•
Network-adaptive applications
How bad is selfish routing?
Selfishness in other contexts?
Interaction …
28
Appendix
29
Responsive Interference - Setup
 Primary Media link, background PCMA links
 Pairs of Tx-Rx randomly chosen from area
(500x500 wrapped in a torus)
 Background: geometric durations, Bernoulli
arrivals
 Free space path loss G~1/d^4, noise 1^(12)
 Estimate I using previous timeslot
 N=100, initial 5slots, buffer B=10
 Heuristic gains: 60% in power, 66% in QoS
30
Responsive Interference- Power
31
Responsive Interference- Quality
32
Adding mode control: (p, m, r)
 Add a control m:
• Packets transmitted in a time slot
 Add a cost Psi(m)
 Modify
 Bellman equations
33
Power-only heuristic
 Fix playout r=R and find optimal power p*
•
similarly to [B&R (1997), B&K(2000), B&Li(2004)]
•
where
p*
X
i
aggres
sive
soft
backoff
hard
backoff
 Heuristic: approximate X, plug it in p.
34
[Performance evaluation cont’d]
 Simulated other channels
 Simulated responsive interference
 Found low sensitivity to r-parameters
35