Sunghwa Son
2011.09.28
Introduction
Time-varying wireless channel
 Large-scale attenuation
•
Due to changing
distance
SNR (dB)
30
25
20
15
10
5
0
0
2000
Due to multipath
•
Unpredictable
Bit errors, frame losses
8000
10000
1700
1750
25
SNR (dB)
 Interference
10 s
6000
Time (Milliseconds)
 Small-scale fading
•
4000
20
15
10
5
0
1500
1550
1600
1650
250 ms
Time (Milliseconds)
Online Bit Rate Adaptation
 Varying modulation & coding
(redundancy)
 Better channel  Higher rate
 Huge gains possible (802.11g has 1-54
Mbps)
Requirements
Estimate channel
accurately
Responsive
Do not react
to interference
Related Work
Existing Algorithms
Frame-based
SNR-based
• RRAA, Wong et. al., 2006.
• SampleRate, Bicket, 2005.
• RBAR, Holland et. al., 2001.
• CHARM, Judd et. al., 2008.
Data
Data
ACK
Estimate frame loss rate at
each bit rate
SNR using preamble
Lookup table
SNR  best rate
Problems With Existing Algorithms
Indoor
Channel
Quality
Outdoor
SNR-based
“SNR  bit rate” table specific to environment
Which table to use when?
X
Frame-based
Too slow
Reacts to interference
Design
SoftRate: Key Insight
Channel
Quality
SoftPHY
Hints
Per-bit
Confidences
Estimate BER
SoftRate
Interference-free BER
Detect Interference
SoftRate: Contributions
• Adapts to channel accurately and quickly
• Robust to collision losses
• Feedback: interference-free BER from SoftPHY hints
• 2X gain over frame-based and SNR-based
Computing SoftPHY Hints
TX
Bits
Encoder
Modulator
RX
Demodulator
Symbols
Symbols
Coded bits
Soft Output
Decoder
Coded Bits
Error vectors
(rcvd symbol – closest valid symbol)
Bits
SoftPHY
Hints
BER from SoftPHY Hints
Soft Output Viterbi
(or) BCJR decoder
decoded bits
log
For linear block or
convolutional code
Pr(correctly decoded)
Pr(incorrectly decoded)
Log Likelihood Ratio
SoftPHY hint of a bit
s =
Probability of bit error
p =
1-p
log p
1
1 + es
BER = Average p over all bits in the packet
Computing Interference-free BER
25
Compute Interference-free BER
SoftPHY Hint
20
15
10
Interference
5
0
0
500
1000
1500
2000
2500
Bit position
3000
3500
4000
SoftPHY Hints With Weak
Signal
35
SoftPHY Hint
30
25
20
15
10
5
0
0
100
200
300
400
Bit position
500
600
700
800
The SoftRate Protocol
Data
Receiver
Interference-free BER
BER
Interference Detection
Sender
Pick rate with max throughput
SoftPHY Hints
Rate Selection at the Sender
BER
24 Mbps
18 Mbps
10-7 10-6 10-5 10-4 10-3
12 Mbps
Adjacent rates have an order of
magnitude difference in BER
Throughput
When is the current rate optimal?
BER
24 Mbps
Optimality range for 18 Mbps
18 Mbps
10-7 10-6 10-5 10-4 10-3
12 Mbps
Throughput
When to lower the rate?
BER
24 Mbps
18 Mbps
10-7 10-6 10-5 10-4 10-3
12 Mbps
Throughput
When to increase the rate?
BER
24 Mbps
18 Mbps
10-7 10-6 10-5 10-4 10-3
12 Mbps
Throughput
The SoftRate Protocol
Data
Receiver
Interference-free BER
BER
Interference Detection
Sender
Precompute optimality ranges
 If BER below optimality range,
increase rate.
 If above range, decrease rate.
 Otherwise, continue at current rate.
SoftPHY Hints
Implementation
Evaluation Method
TCP
ns-3 simulations
Rate Adaptation (SoftRate,
SNR-based, Frame-based)
SoftPHY Traces
PHY: GNU Radio + USRP
Experiments
Channel Simulator
Evaluation Questions
SoftPHY
• Can SoftPHY hints estimate channel BER?
• Can SoftPHY hints identify interference?
SoftRate
• Gains of SoftRate in mobile channels?
• SoftRate robust to interference?
SoftPHY Evaluation
SoftPHY Hints Predict BER
SNR vs. BER: Varying Mobility
Interference detection accuracy:
various bit rate
Softrate Evaluation
SoftRate Evaluation
in Mobile Channels
• Compare with
– StaticBest: omniscient—best for each pkt
– SNR-based: RBAR and CHARM
– Frame-based: RRAA and SampleRate
TCP
AP
Clients
Wired
LAN
Traces
Is SoftRate close to optimal?
(Walking Speed)
Aggregate Throughput
12
10
8
StaticBest
6
SoftRate
4
2
0
1
2
3
Number of clients
Within 10% of the optimal
4
5
SoftRate vs. Frame-based:
Walking speed
Aggregate Throughput
12
10
8
SoftRate
6
RRAA
SampleRate
4
2
0
1
2
3
4
Number of clients
Up to 2X over best frame-based algorithm
5
SoftRate vs. SNR-based:
Varying Mobility
Rate selection accuracy
Evaluation Answers
SoftPHY
• Can SoftPHY hints estimate channel BER? YES
~80% of
• Can SoftPHY hints identify interference?
the time
SoftRate
• Gains of SoftRate in mobile channels? 2X - 4X
• SoftRate robust to interference? Almost always
Conclusion
SoftRate: Summary
• Accurate, responsive, robust to collision losses
• Feedback: interference-free BER from SoftPHY hints
• 2X over frame-based, 4X over untrained SNR-based
Looking ahead
• BER computation from SoftPHY hints useful for other
cross-layer protocols