Diagnosing Wireless Diagnosing Wireless Packet Losses in 802.11: Packet Losses in 802.11: Separating Collision Separating Collision from Weak Signal from Weak Signal Shravan Rayanchu, Arunesh Mishra, Dheeraj Agrawal, Sharad Saha, Suman Banerjee
Dec 24, 2015
Diagnosing Wireless Packet Diagnosing Wireless Packet Losses in 802.11:Losses in 802.11:Separating Collision from Separating Collision from Weak SignalWeak Signal
Shravan Rayanchu, Arunesh Mishra, Dheeraj Agrawal, Sharad Saha, Suman Banerjee
MotivationMotivationPacket Loss
2 Causes
Solution Inadequate802.11
Can we determine cause of packet loss?
Packet loss in Wireless Packet loss in Wireless NetworksNetworks
A BC
While A is transmitting,
C initiates RTS to B
Packet loss in Wireless Packet loss in Wireless NetworksNetworks
A BC
Since neither A nor B knows the other is transmitting, both RTS’s
are sent and collide at B, resulting in packet loss
Packet loss in Wireless Packet loss in Wireless NetworksNetworks
A BC
Since neither A nor B knows the other is transmitting, both RTS’s
are sent and collide at B, resulting in packet loss
Packet loss in Wireless Packet loss in Wireless NetworksNetworks
A
B
C
Here A and C are in just barely in range of each other, but both are
in range of B
Packet loss in Wireless Packet loss in Wireless NetworksNetworks
A
B
C
A send its RTS to C, which is received and B is silenced
Packet loss in Wireless Packet loss in Wireless NetworksNetworks
A
B
C
C send its CTS to A, but the packet is not heard due to weak signal caused by interference by
noise
Detecting Packet LossDetecting Packet LossRecap: 2 causes of packet loss802.11 Solution
◦BEBDifferent causes lead to different
solutions
Fixing Packet LossFixing Packet Loss◦For low signal
Increase power Decrease data rate How to differentiate?
C EA
D
B
Rate = 20
Rate = 10
Introduction to COLLIEIntroduction to COLLIE802.11, CARA, and RRAA use
multiple attempts to deduce cause of packet loss
COLLIE direct approach Error packet kickbackClient analysis
COLLIE: Single APCOLLIE: Single APAP error packet kickbackClient-side analysisProblem: how can the AP
successfully re-transmit packet?
Experimental DesignExperimental Design
Two transmitters, T1 and T2Two receivers, R1 and R2Receiver R hears all signals
Experimental DesignExperimental DesignThree possibilities at R:1. Packet received without error2. Packet received in error3. No packet received
Error MetricsError MetricsThree error metrics:Bit Error Rates (BER)Symbol Error Rates (SER)Error Per Symbol (EPS)
Metrics for AnalysisMetrics for AnalysisReceived Signal Strength (RSS) =
S + IHigh RSS collisionLow RSS channel fluctuations
Bit Error Rate (BER) = total # incorrect bits
BER is higher for collisions, lower for low signal
RSS: The DetailsRSS: The Details
Of all packets lost due to low signal, 95% had an RSS less than -73dB, compared to only 10% for collisions
Metrics for AnalysisMetrics for AnalysisSymbol level errors: errors
within transmission frameMultiple tools used to analyze
symbol-level errors
FramingFraming
0011 0011 0011 0011 1101 0011
Collision
Channel Fluctuation
0011 0011 0011 0111 1011 0010
Symbol-level ErrorsSymbol-level ErrorsSymbol Error Rate (SER)- # symbols
received in errorErrors Per Symbol (EPS)- average # errors
within each symbolSymbol Error Score (S-score): calculated
as , where Bi is a burst of n bits2
1
n
ii
B
74% accuracy
S-ScoreS-Score
0011 0011 0011 0011 1101 0011
Collision
Channel Fluctuation
0011 0011 0011 0111 1011 0010
S-Score =
2 2 2 2
1
1 1 1 3n
ii
B
S-Score = 2 2 2 2
1
0 3 0 9n
ii
B
Multi-AP COLLIEMulti-AP COLLIEError packet sent to a central
COLLIE server
Most important where the capture effect is dominant
ResultsResultsStatic situation average of
30% gains in throughputFor multiple collision sources and
high mobility, throughput gains of 15-60%
ConclusionsConclusionsCOLLIE implementation achieves
increased throughput (20-60%) while optimizing channel use
Implementation can be done over standard 802.11, resulting in much lower startup costs than other protocols