Transcript 3-4-1_QoS Intro adap..

Outline
• Wireless introduction
• Wireless cellular (GSM, CDMA, UMTS)
• Wireless LANs, MAC layer
• Wireless Ad hoc networks
– routing: proactive routing, on-demand
routing, scalable routing, geo-routing
– wireless Ad hoc multicast
– TCP in ad hoc networks
– QoS, adaptive voice/video apps
• Sensor networks
QoS Support
• Objective:
– Guarantee Quality of Service to real time users
• QoS users:
– Multimedia
– Voice over IP
– Time critical applications
• QoS metrics:
– Bandwidth / data rate
– Delay
– Delay variance / jitter
– Packet loss
– Bit Error rate
– etc
QoS Wireless Scenarios
• Scenarios
– Multihop ad hoc network
– Wireless Internet access (from W-LAN or ad hoc net)
– Wireless mesh networks
• The above scenarios are all challenging:
– Channel parameter fluctuations (motion, obstacles,
interference)
– Change in S/N ratio at receiver
– Mobility (path breaks, must be repatched, etc)
– Multiaccess channel (distributed queue scheduling)
– Interference from nodes beyond one hop
• QoS support truly a cross-layer design:
– From physical layer to applications
Physical Layer QoS
• As distance between tx and rcv increases, and
noise/interference increases, S/N ratio decreases.
• This requires a reduction in data rate to maintain
acceptable BER
• In 802.11 a, the rate can be adjusted in the following
steps: 6,9,12,18, 24,48,54 Mbps
• In 802.11b: 1,2,5.5, 11 Mbps
• ARF (Auto Rate Feedback):
– Rate reduced by transmitter if ACK missed twice
– Rate restored if 10 consecutive ACKs are
correctly received
• RBAR (Receiver Based Auto Rate):
– Receiver measures S/N ratio from RTS; returns
allowed rate in CTS
MAC Layer QoS
• IEEE 802.11e is addressing QoS:
– DCF -> Enhanced DCF (EDCF); 8 priority levels
defined by different AIFS values
– PCF -> Hybrid CF (HCF): can “poll” nodes also in
contention free period
• 802.11e is mainly directed to infrastucture LANs
• In ad hoc, multihop network one can use some of the
EDCF features
• Another problem in ad hoc nets is the simultaneous
scheduling of multiple transmissions (spatial reuse)
• Several solutions:
– TDMA (eg, cluster TDMA)
– Non interfering flow graph (SCOPE)
Network layer QoS
• QoS routing (either proactive or on demand)
• QoS extension of LANMAR, Fisheye
• Real time bandwidth measurement in mobile
ad hoc networks
• Bandwidth allocation (or re-allocation)
• Bandwidth reservation
• Call Acceptance Control
• Mobility adaptation
• Scheduling and policing
• Coexistence with best effort traffic
Application layer QoS
• Rate adaptation of multimedia stream
• Network to application layer feedback (with
avail bandwidth information)
• End to end bandwidth estimation
• We will discuss a video adaptation example
Adaptive , renegotiable QoS support
• If nodes move, the end to end path
capacity can change dynamically:
– The connectivity and topology change
– The channel propagation characteristics
change (eg, multipath fading)
– The external interference can change
• A possible answer:
– Renegotiable QoS
– rate adaptation
– preemption of lower priority users
• Case study: adaptive video streaming
Scalable, Adaptive Video Streaming
FLIR
server
client
Why Adaptive Video?
• To prevent traffic congestion
– Adjust the stream rate so that it “fairly”
shares the available bandwidth
• To deal with channel random interference,
propagation, jamming
– Adjust (reduce) video packet size,
strengthen the channel encoding to combat
random/burst errors
• Challenge:
– Distinguish between congestion (must
reduce rate) and random errors (keep same
rate but strengthen the code)
The concept
channel fading
mobility
internal interference
MPEG
server H263
server
LOSS FROM
CONGESTION
LOSS FROM ERROR
client
client
external interference
(jamming, environment noise)
• End to End Feedback Adaptation Approach
– Traditional approach
– Transport and Application employ end-to-end
measurements to perform flow control
• Network Feedback Adaptation Approach
– Network layer propagates channel measurements to
source
– Ability to detect cause of degradation and react
properly
• Implemented in simulation, testbed and hybrid simulatn
Adaptation Techniques
• RTP Loss rate Adaptation (Trial And Error)
– It constantly ‘tries’ to support higher rates
– Backs up when loss is detected
• Available Bandwidth (AB Probe) eg. TCPW
– Av. Bdw. estimated from inter-packet
intervals
– Can distinguish error loss from congestion
• Network Feedback
– Link channel quality and bdw info
piggybacked on routing pkts
– Gives accurate picture of network path
state to source
Network feedback
• Goal: dynamic estimation of available
bandwidth AB from source to destination
• Approach: MAC and network layer cooperation
• At the MAC layer:
– estimate the permissible throughput from a
node to each of its neighbors
– estimate the “available bandwidth” AvBd for
a node
• At the network layer:
– Finally, propagate the AvBd estimate to
source using routing control packets
MAC layer computation
(a) Estimate link throughput:
Throughput = S / (Tack - Ttx)
where:
S = packet size (bits)
Ttx : time when packet tx is first attempted
Tack : time when ACK is received
This estimate accounts for collision avoidance,
retransmissions, backoffs etc
The estimate is averaged over a measurement
window
(b) Measure node utilization u = fraction of time
node is busy (ie, it has packets in the queue)
Available Bdw = (1 – u) x Throughput
Measurement Example
11Mbps 802.11 links 10Mbps
Throughput
Permissible
Throughput
A
Node
utilization=80%
6Mbps
AB
AC
AD
10Mbps(<11)
7Mbps (<11)
6Mbps (<11)
B
Hidden
Terminals
7Mbps
2Mbps
1.4Mbps
1.2Mbps
D
Hidden
Terminals
C
Hidden
Terminals
Link Assisted/Network Feedback
channel fading
server
MPEG internal interference
H263
Multi-hop noise
.....
Network
Link
server
Application
external interference
client
(jamming, environment noise)
Network
Link
Measure()
client
LOSS FROM ERROR
LOSS FROM
CONGESTION
Propagate()
mobility
Network
Link
Measure() Network
Link
Network
Link
Measure()
Measure()
API
Network
Link
Network QoS Feedback
Accurate Available Bandwidth
Measurement and Advertising
How to propagate to sources? QM-AODV
• Q-AODV – extension of AODV for QoS
1 RREQ
routing
Update Reverse
Route with AB(1)
• Finds routes given: minimum
bandwidth and maximum delay
MREP(AB(4))
2
constraint
If Bhop=2 or
minimum
• Does not deal with measurement
3 Update
aggregation
RREP
(min(inf,AB(5),0 or 1)
• Does not have to propagate
Significant
measurements
AB Change
• AvBd propagation - QM-AODV
4
– Use Q-AODV packet format
RREP
– Add bottleneck hop information
(inf,0)
– Use an MREP special reply packet
5
– Use Measurement Relay Factor to
trigger Measurement updates
802.11 Multi-hop – Loss Rates
Loss Rates - Network of 64
100%
80%
%
60%
40%
20%
0
10
20
30
40
50
60
70
0%
Connections
Net
• AB-probe
AB
RTP
High L
PP
Low L.
performs best initially, then worse
than RTP when load is high
• RTP adaptation works better than direct PP
(Packet Pair) method
• Network Feedback is best
Real-Time Protocol (RTP)
• Provides standard packet format for real-time
applications - a sublayer of the transport layer
• Typically runs over UDP
• Specifies header fields (see below)
• Payload Type: 7 bits, providing 128 possible different
types of encoding; eg PCM, MPEG2 video, etc.
• Sequence Number: 16 bits; used to detect packet loss
• Timestamp: 32 bytes; gives the sampling instant of the
first audio/video byte in the packet; used to remove
jitter introduced by the network
• Synchronization Source identifier (SSRC): 32 bits; an ID
for the stream; assigned randomly by the source