ES2 : Managing Link Level Parameters for Elevating Data Rate and - - PowerPoint PPT Presentation

ES2: Managing Link Level Parameters for Elevating Data Rate and Stability in High Throughput WLAN

Sandip Chakraborty, Subhrendu Chattopadhyay

Department of Computer Science and Engineering Indian Institute of Technology Kharagpur Department of Computer Science and Engineering Indian Institute of Technology Guwahati

January 8, 2016

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 1 / 16

Preface

High-throughput extension for wireless local area networks (WLAN): IEEE 802.11n (600 Mbps), IEEE 802.11ax (Gbps wireless) Supports a large set of link control parameters:

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 2 / 16

Preface

High-throughput extension for wireless local area networks (WLAN): IEEE 802.11n (600 Mbps), IEEE 802.11ax (Gbps wireless) Supports a large set of link control parameters:

Number of MIMO Spatial Streams

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 2 / 16

Preface

High-throughput extension for wireless local area networks (WLAN): IEEE 802.11n (600 Mbps), IEEE 802.11ax (Gbps wireless) Supports a large set of link control parameters:

Number of MIMO Spatial Streams Channel Bonding: IEEE 802.11n supports 20 and 40 MHz, IEEE 802.11ac supports 20, 40, 80, 120 MHz

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 2 / 16

Preface

High-throughput extension for wireless local area networks (WLAN): IEEE 802.11n (600 Mbps), IEEE 802.11ax (Gbps wireless) Supports a large set of link control parameters:

Number of MIMO Spatial Streams Channel Bonding: IEEE 802.11n supports 20 and 40 MHz, IEEE 802.11ac supports 20, 40, 80, 120 MHz Advanced Modulation and Coding Schemes – Supports higher physical data rates

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 2 / 16

Preface

High-throughput extension for wireless local area networks (WLAN): IEEE 802.11n (600 Mbps), IEEE 802.11ax (Gbps wireless) Supports a large set of link control parameters:

Number of MIMO Spatial Streams Channel Bonding: IEEE 802.11n supports 20 and 40 MHz, IEEE 802.11ac supports 20, 40, 80, 120 MHz Advanced Modulation and Coding Schemes – Supports higher physical data rates Frame aggregation and Block Acknowledgement – Reduce channel access overhead

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 2 / 16

Preface

High-throughput extension for wireless local area networks (WLAN): IEEE 802.11n (600 Mbps), IEEE 802.11ax (Gbps wireless) Supports a large set of link control parameters:

Number of MIMO Spatial Streams Channel Bonding: IEEE 802.11n supports 20 and 40 MHz, IEEE 802.11ac supports 20, 40, 80, 120 MHz Advanced Modulation and Coding Schemes – Supports higher physical data rates Frame aggregation and Block Acknowledgement – Reduce channel access overhead Short Guard Intervals – Saves guard time when interference is less

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 2 / 16

Link Adaptation: What is the Best Link Configuration?

Every feature has its own pros and cons:

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 3 / 16

Link Adaptation: What is the Best Link Configuration?

Every feature has its own pros and cons:

Channel Bonding: Data loss is more for wider channels – channel bonding is not good when channel error rate is high

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 3 / 16

Link Adaptation: What is the Best Link Configuration?

Every feature has its own pros and cons:

Channel Bonding: Data loss is more for wider channels – channel bonding is not good when channel error rate is high Modulation and Coding: Higher modulation and coding requires higher signal strength to sustain – high modulation and coding rate may not be suitable when SINR is low

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 3 / 16

Link Adaptation: What is the Best Link Configuration?

Every feature has its own pros and cons:

Channel Bonding: Data loss is more for wider channels – channel bonding is not good when channel error rate is high Modulation and Coding: Higher modulation and coding requires higher signal strength to sustain – high modulation and coding rate may not be suitable when SINR is low Frame Aggregation: Frame aggregation may result in high data loss due to channel interference

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 3 / 16

Some Observations from a Practical Testbed

Testbed Configurations:

Ralink RT-3352 Wireless Router-on-chip – supports IEEE 802.11n 2 × 2 MIMO Supports 20 MHz and 40 MHz at 5 GHz band – 300 Mbps physical data rate Linux Kernel 2.4.12 – openwrt supported

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 4 / 16

Observation 1: Impact of Modulation and Coding

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 6 8 10 12 14 16 PSR Average SNR (dB) (a) Channel Width = 20 MHz MCS 4 MCS 7 MCS 11 MCS 15 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 6 8 10 12 14 16 PSR Average SNR (dB) (b) Channel Width = 40 MHz MCS 4 MCS 7 MCS 11 MCS 15

Observations:

Low modulation and coding levels can sustain at low SNR region – provide better PSR compared to high modulation and coding values High modulation and coding levels provide good PSR at high SNR region PSR variation in significantly more in 40 MHz communication compared to 20 MHz communication.

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 5 / 16

Observation 2: Impact of Channel Bonding and Guard Intervals

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 6 8 10 12 14 16 18 PSR Average SNR (dB) (a) Impact of Channel Bonding 20 MHz 40 MHz 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 6 8 10 12 14 16 18 PSR Average SNR (dB) (b) Impact of Guard Interval 800 ns 400 ns

Observations:

40 MHz gets more affected due to external noise and interference Short guard interval (400 ns) is effective for low interference scenario

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 6 / 16

Observation 2: Impact of Channel Bonding and Guard Intervals

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 6 8 10 12 14 16 18 PSR Average SNR (dB) (a) Impact of Channel Bonding 20 MHz 40 MHz 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 6 8 10 12 14 16 18 PSR Average SNR (dB) (b) Impact of Guard Interval 800 ns 400 ns

Observations:

40 MHz gets more affected due to external noise and interference Short guard interval (400 ns) is effective for low interference scenario

We need to develop an adaptive link parameter selection mechanism based on channel condition...

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 6 / 16

Solution Approach: Estimate, Sample and Select (ES2)

A three step iterative process:

Estimate the SNR at transmitter from the measured received signal strength (RSS) at the receiver Sample the feature sets based on the estimated SNR thresholds Select the final data rate from the filtered samples

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 7 / 16

Estimation of SNR

Estimation of SNR is non-trivial, because,

The noise level significantly depends on parametric settings (like number of spatial streams, channel width etc) → Simple subtraction of noise level from signal level does not work. The transmitter needs to figure out link parameter settings, whereas SNR is measured at the receiver → SNR prediction is required at transmitter.

Piggybacking SNR does not work because the link parameter settings may change!

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 8 / 16

Estimation of SNR: A Kalman Filtering Approach

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 9 / 16

Sample Feature Set

SNR is quite fluctuating and may not be a good choice for feature sampling. We use diffESNR – diffESNRt = SNRt(|SNR2

t − SNR2 t−1|)

This captures SNR fluctuation as well – if fluctuation is more, some link parameter settings become unstable.

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 10 / 16

Impact of diffESNR

10 20 30 40 50 60 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 10 20 30 40 Goodput (Mbps) diffESNR (dB) MCS (a) Effect of Spatial Streams Low DiffESNR High DiffESNR diffESNR 1 2 3 4 5 6 7 5 10 15 20 25 30 Goodput (Mbps) diffESNR (dB) (b) Effect of Channel Bonding 20 MHz 40 MHz COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 11 / 16

Estimate, Sample and Select (ES2)

A three step iterative process:

Estimate the SNR at transmitter from the measured received signal strength (RSS) at the receiver Sample the feature sets based on the estimated SNR thresholds Select the final data rate from the filtered samples – We are left with

• nly few options! Apply standard rate adaptation...

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 12 / 16

Protocol Performance: Static Network

30 40 50 60 70 80 90 100 110 120 4 6 8 10 12 Goodput (Mbps) Number of Nodes (a) Goodput ES2 SampleLite Minstrel HT 10 20 30 40 50 60 4 6 8 10 12 Samples per Second Number of Nodes (b) Sampling Overhead ES2 SampleLite Minstrel HT

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 13 / 16

Protocol Performance: Mobile Network

20 30 40 50 60 70 80 90 100 4 6 8 10 12 Goodput (Mbps) Number of Nodes (a) Goodput ES2 SampleLite Minstrel HT 10 20 30 40 50 60 70 80 90 4 6 8 10 12 Samples per Second Number of Nodes (b) Sampling Overhead ES2 SampleLite Minstrel HT

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 14 / 16

Fairness: Average Link Goodput Variation

5 10 15 20 25 4 6 8 10 12 Goodput Variation (Mbps) Number of Nodes (a) Static Scenario ES2 SampleLite Minstrel HT 5 10 15 20 25 4 6 8 10 12 Goodput Variation (Mbps) Number of Nodes (b) Mobile scenario ES2 SampleLite Minstrel HT

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 15 / 16

Concluding Remarks

ES2 works well in pure IEEE 802.11n network, but the sampling does not work sometime in a mixed network (IEEE 802.11n + IEEE 802.11b/g)

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 16 / 16

Concluding Remarks

ES2 works well in pure IEEE 802.11n network, but the sampling does not work sometime in a mixed network (IEEE 802.11n + IEEE 802.11b/g)

10 20 30 40 50 60 70 80 100 200 300 400 500 600 Goodput (Mbps) Time (Sec) ES2 SampleLite COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 16 / 16

Concluding Remarks

ES2 works well in pure IEEE 802.11n network, but the sampling does not work sometime in a mixed network (IEEE 802.11n + IEEE 802.11b/g)

10 20 30 40 50 60 70 80 100 200 300 400 500 600 Goodput (Mbps) Time (Sec) ES2 SampleLite

Can we say something about the interoperability or backward compatibility?

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 16 / 16

Thank You

COMSNETS 2016 (Bangalore, India) ES2: Managing Link Level Parameters... January 8, 2016 16 / 16