Upload
britton-may
View
217
Download
0
Tags:
Embed Size (px)
Citation preview
Scalable Video Transport over Wireless IP Networks
Dapeng WuElectrical & Computer Engineering
University of Florida
2
Bandwidth FluctuationsBandwidth Fluctuations
Domain A
Domain B
Domain C
InternetSource
Receiver
AccessAccessSWSW
AccessAccessSWSW
GatewayGateway
64 kb/s
1 Mb/s
Cellular NetworksWireless LAN
Mobile PCMobile PC
3
ChallengesChallenges• UnreliabilityUnreliability
– Fading– Noise
• Bandwidth fluctuationsBandwidth fluctuations– Moving between different networks (LAN to
WAN)– Hand-off– …
• Heterogeneity for multicastHeterogeneity for multicast
4
Unicast vs. MulticastUnicast vs. Multicast
Unicast Multicast
5
Three Independent TechniquesThree Independent Techniques
• Scalable video codingScalable video coding
• Network-aware adaptation of end systemsNetwork-aware adaptation of end systems– Network awareness– Adaptation
• Adaptive QoS support from networks: Adaptive QoS support from networks: adaptive servicesadaptive services
6
Scalable Video RepresentationsScalable Video Representations
Layered video encoding/decoding. D denotes the decoder.Layered video encoding/decoding. D denotes the decoder.
7
An Application: IP MulticastAn Application: IP Multicast
8
Our ApproachOur Approach
• Unify the three techniques: Unify the three techniques:
an adaptive frameworkan adaptive framework
9
OutlineOutline
• Challenges for video over wireless IP Challenges for video over wireless IP networksnetworks
• An adaptive framework for video over An adaptive framework for video over wireless IP networkswireless IP networks– Scalable video representations– Network-aware end systems– Adaptive services
• SummarySummary
10
Network-aware End SystemsNetwork-aware End Systems
• Why using network-aware end systems?Why using network-aware end systems?– All layers may get corrupted with equal
probability without awareness of channel status
• How?How?– Discard enhancement layers at the sender based
on network status
• Network-aware adaptation:Network-aware adaptation:– Network monitoring: collect information– Adaptation: adapt video representations based on
network status
11
Taxonomy of Network Monitoring Taxonomy of Network Monitoring
CriteriaCriteria Type of monitoringType of monitoring
Method of monitoringMethod of monitoring ActiveActive PassivePassive
Monitoring frequencyMonitoring frequency On demandOn demand ContinuousContinuous
Replication of informationReplication of information CentralizedCentralized DistributedDistributed
12
Adaptation/ScalingAdaptation/Scaling
An architecture for transporting scalable video from a mobile terminal An architecture for transporting scalable video from a mobile terminal to a wired terminal.to a wired terminal.
13
ScalingScaling
• The operations of a scalerThe operations of a scaler– Drop the enhancement layers– Do not scale the video
• Scaling based on network statusScaling based on network status– Available bandwidth– Channel quality (BER)
14
OutlineOutline
• Challenges for video over wireless IP Challenges for video over wireless IP networksnetworks
• An adaptive framework for video over An adaptive framework for video over wireless IP networkswireless IP networks– Scalable video representations– Network-aware end systems– Adaptive services
• SummarySummary
15
Adaptive ServicesAdaptive Services
• Objective: Objective: – achieve smooth change of perceptual quality in
presence of bandwidth fluctuations.
• Functions:Functions:– Reserve a minimum bandwidth for the base
layer– Adapt enhancement layers based on available
bandwidth and the fairness policy
16
Adaptive Services (cont’d)Adaptive Services (cont’d)
• ProvisioningProvisioning– End-to-end deployment (our focus)– Local deployment
• Components:Components:– Service contract– Call admission control and resource reservation– Substream scaling– Substream scheduling– Link-layer error control
17
Service ContractService Contract• A service contract consists of multiple A service contract consists of multiple
subcontractssubcontracts– Bandwidth reservation for the base layer – No QoS guarantee for enhancement layers
• EnforcementEnforcement– Shaping– Priority
18
Call Admission Control (CAC)Call Admission Control (CAC)
• Objective:Objective:– Provide a QoS guarantee while efficiently
utilizing network resources
• The operation of CAC: checkThe operation of CAC: check– whether QoS for existing connections is
violated– whether the incoming connection’s QoS can be
met
19
Resource ReservationResource Reservation
• Different from the counterpart in wired Different from the counterpart in wired networks:networks:– The reserved bandwidth may not be rigidly
guaranteed in wireless networks
• Two parts of resource reservationTwo parts of resource reservation– Reserve resources along the current path– Reserve resource on the paths from the current
base station to neighboring base stations
20
Mobile multicast mechanismMobile multicast mechanism
• Objective:Objective:– Provide seamless QoS during a handoff
• Multicast mechanism: Multicast mechanism: – Multicast the base layer to the neighboring base
stations
21
Substream ScalingSubstream Scaling
• Objective:Objective:– Adapt video streams during bandwidth
fluctuations and/or under poor channel conditions
• Scaling decision based on utility fairness or Scaling decision based on utility fairness or max-min fairnessmax-min fairness– Utility fairness is based on utility functions– Max-min fairness is based on revenue
22
Substream Scaling (cont’d)Substream Scaling (cont’d)
An architecture for transporting scalable video from a wired terminalAn architecture for transporting scalable video from a wired terminal
to a mobile terminal.to a mobile terminal.
23
Utility FunctionsUtility Functions
24
Substream SchedulingSubstream Scheduling
25
Link-layer Error ControlLink-layer Error Control• Forward error correction (FEC)Forward error correction (FEC)
• Automatic repeat request (ARQ)Automatic repeat request (ARQ)
• Truncated type-II hybrid ARQTruncated type-II hybrid ARQ
• Delay constrained hybrid ARQDelay constrained hybrid ARQ– A receiver sends request based on delay bound
of the packet
26
Delay-constrained Hybrid ARQDelay-constrained Hybrid ARQ
RCPC: Rate compatible punctured convolution
27
SummarySummary• Objective: end-to-end solution to providing QoS for Objective: end-to-end solution to providing QoS for
video transport over wireless IP networksvideo transport over wireless IP networks• Our approach: Our approach: an adaptive frameworkan adaptive framework
– Scalable video representations
– Network-aware end systems
– Adaptive services
• Advantages of the adaptive frameworkAdvantages of the adaptive framework– Perceptual quality is changed gracefully
– Resources are shared in a fair manner