Enabling Wireless LAN Troubleshooting
Contents
1. Detection Accuracy gt 85 for varying Airtime Utilization of Hidden Li Framework Evaluation Capture Effect 9 locations exposed to Capture Effect e e N A Oo oc o o6 o6 6 Detection Accuracy 9 10 20 30 40 50 Airtime Utilization Low Airtime Utilization leads to similar impact as of Hidden Terminal Failure in detection Outline vvvv Vv Conclusion and Future Work Conclusion and Future Work Based on MAC layer statistics exposed to user level Defined the key metrics able to characterize common 802 11 pathologies Developed our application level framework for identifying trends of metrics in presence of a pathology Achieved high accuracy of detection Conclusion and Future Work Extension of our framework for detection in presence of multiple pathologies gt Large scale evaluation in real world environments Passive detection for reducing overhead Thank you2. Enabling Wireless LAN Troubleshooting Ilias Syrigos Stratos Keranidis Thanasis Korakis and Constantine Dovrolis Outline Introduction Motivation IEEE 802 11 Pathologies Detection Methodology Framework Evaluation vv vv Vv Conclusion and Future Work Introduction Motivation Poor performance in home WLANs An everyday phenomenon Various causes often unknown to home administrators Troubleshooting hard even to the experts Introduction Motivation Two approaches for diagnosing WLAN pathologies Application layer frameworks running over commercial WLAN devices Lack of accuracy Better applicability Driver modifications or even custom hardware for diagnosing in PHY MAC Better accuracy Lack of applicability Introduction Motivation Our proposal Bridge the gap Take advantage of default driver level information Rate control algorithm statistics exported to user level for debugging Define the metrics able to characterize each considered pathology Extensive experimentation in controlled environments Incorporate our findings in a user level detection framework Evaluate its performance by quantifying the detection accuracy Outline IEEE 802 11 Pathologies Detection Methodology Framework Evaluation vv vv Vv Conclusion and Future Work IEEE 802 11 Pathologies The pathologies catego
3. Camera 2 MW Oven p Bluetooth Se D 50 SNR 15 dB O z 9 12 18 24 36 48 54 6 9 12 18 PHY layer bitrate of SUT Mbps PHY layer bitrate of SUT Mbps Decrease in NCA caused of CW doubling Decrease in FDR in complex bitrates Detection Methodology Hidden Terminal 100 100 zi TR 4 PHY 24 90 7 90 TR 5 PHY 12 80 80 TTR 5 PHY 36 70 _ 70 TR 5 PHY 48 S eo 9 go TR 1 PHY 36 50 c 50 2 40 TR 1 PHY 24 Q 40 30 TR 5 PHY 12 30 20 TR 5 PHY 36 20 TR 5 PHY 48 L 10 TR 10 PHY 36 p 9 12 18 24 36 48 54 9 142 18 24 436 48 PHY layer bitrate of SUT Mbps PHY layer bitrate of SUT Mbps NCA decreases due to Low SNR coexistence Asmall increase due to shorter duration of frames followed by a decrease in FDR No Trend Detection Methodology Capture Effect 100 mfa TR 1 PHY 24 100 90 TR 5 PHY 12 90 30 WF TR 5 PHY 36 80 TR 5 PHY 48 3 is m TR 10 PHY 36 9 d peli i aa d 50 ETR 5 PHY 36 50 Q 40 TR5 PHY 48 9 u TR 10 PHY 36 30 30 20 20 10 10 0 48 54 9 12 18 24 36 12 18 24 36 PHY layer bitrate of SUT Mbps PHY layer bitrate of SUT Mbps Similar to Hidden Terminal but heavier impact leads to no trend in both NCA and FDR Detection Methodology Summariz
4. 00 po l 90 o 9 9 9 9 9 9 80 s0 70 4 70 60 S 60 p 50 50 Q 40 1 STA TR 5 PHY 6 9 40 7 1 STA TR 5 PHY 6 u 1 STA TR 5 PHY 54 30 91 STA TR 5 PHY 54 1 STA TR 20 PHY 24 7 2 STAs TR 5 PHY 24 10 4 3 STAs TR 5 PHY 24 29 1 STA TR 20 PHY 24 2 STAs TR 5 PHY 24 10 3 STAs TR 5 PHY 24 48 54 6 9 12 18 24 36 9 12 18 24 36 48 54 f PHY layer bitrate of SUT Mbps Ht i Bitrate diversity leads to decrease in NCAs while FDR remains constant Detection Methodology Contention with non 802 11 devices 100 zi MW Oven 1 1007 90 9 MW Oven 2 90 80 F Surv Camera 1 80 S a Surv Camera 2 S A lt 50 I e X K _ __ x 50 a LL 2e zi MW Oven 1 9 MW Oven 2 20 J 20 F Surv Camera 1 1 al 1 Di Surv Camera 2 48 54 48 54 18 24 36 9 12 18 24 36 PHY layer bitrate of SUT Mbps PHY layer bitrate of SUT Mbps Constant performance of NCA metric Increasing FDR in case of MW Fluctuation in case of Camera due to almost zero transmission attempts Detection Methodology Low SNR Low Signal and High Noise 100 100 90 90 80 80 70 70 3 50 SNR 15 dB OQ 40 4 SNR 10 dB 3 MERSNR 5 dB 20 Surv Camera 35 MW Oven 1 Bluetooth 40 4 SNR 10 dB 30 SNR 5 dB 20 Surv
5. ing Normalized Channel Frame Delivery Rate FDR Constant No Trend Increasing Decreasing 802 11 Hidden Non 802 11 Decreasing l i Contention Low SNR Contention Terminal NoTrend Capture Effet Non 802 11 Contention Accesses NCA Outline gt gt d Framework Evaluation Conclusion and Future Work Framework Evaluation Contention One two and three contending stations Varying PHY bitrates Varying traffic loads 4 100 D gt 80 60 lt E 40 o 433 STAs g 20 2 2 STAs a L 1 STA 5 15 25 75 85 95 35 45 55 65 Airtime Utilization Detection accuracy of 100 in cases of performance degradatio Framework Evaluation Frame Loss Evaluation Link 20 different locations 4 different levels of transmission power Resulting in 80 different scenarios Interfering Link Fixed location Varying PHY rate Varying traffic loads Framework Evaluation Low SNR Evaluation when Interfering Link is off ceo o N jf o o o6 o6 Detection Accuracy 9 5 oe 10 15 20 Link SNR 100 accuracy until SNR is not considered Low Framework Evaluation Hidden Terminal 4 locations exposed to Hidden Terminal B Dm oO A N e Detection Accuracy 9 10 20 30 40 50 Airtime Utilization
6. rization that we followed is based on the way 802 11 protocol functions Carrier Sense Backoff Retransmissions policy CW Medium Contention Multiple 802 11 devices competing for channel access Non 802 11 devices Microwave ovens Wireless Cameras etc operating in 2 4 GHz band Frame Loss Low SNR conditions due to Low Signal Power or due to High Noise Symmetric and Asymmetric Capture Effect Hidden Terminal 802 11 Performance Pathologies Medium Contention 802 11 Non 802 11 802 11 Contention Contention Impairments Hidden Capture Terminal Effect Low Signal High Noise MAC Layer Statistics Our approach is based on two key metrics evaluated across bitrates Normalized Channel Accesses NCA CA MCA CA Channel Accesses per sec MCA Model Based Channel Accesses per sec Frame Delivery Ratio FDR ST CA ST Successful Transmissions per sec Outline Detection Methodology Framework Evaluation vv vv Vv Conclusion and Future Work Detection Methodology Initial throughput test for performance estimation Throughput under 80 of max gt Triggers detection mechanism Characterize evolution of key metrics across bitrates NCA and FDR Identification of trends across bitrates Theil Sen Estimator Increasing Decreasing No Trend and Constant Detection Methodology gt Contention with 802 11 devices 1
Download Pdf Manuals
Related Search