Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency - - PowerPoint PPT Presentation

simulation of 802 11 phy mac the quest for accuracy and
SMART_READER_LITE
LIVE PREVIEW

Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency - - PowerPoint PPT Presentation

Research Motivation Proposed Approach First Implementation Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency Michele Segata Renato Lo Cigno 9th Annual Conference on Wireless On-demand Network Systems and Services January


slide-1
SLIDE 1

Research Motivation Proposed Approach First Implementation

Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency

Michele Segata Renato Lo Cigno

9th Annual Conference on Wireless On-demand Network Systems and Services January 9-11, 2012, Courmayeur, Italy

Michele Segata, Renato Lo Cigno Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency

slide-2
SLIDE 2

Research Motivation Proposed Approach First Implementation Realism vs. Scalability YANS PhySim Shadowing

Reasons Of This Work

Need of realistic and scalable simulations for VANETs ns-3 choices:

ns-3 default PHY layer (YANS)

Stochastic Scalable Lack of realism

PhySim implementation by DSN Research Group (KIT)1

Emulative Not scalable Highly realistic

Other popular simulators:

ns-2 Omnet++

None consider shadowing due to obstacles Goal: provide a scalable model accurate enough for VANET simulations

Michele Segata, Renato Lo Cigno Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency

slide-3
SLIDE 3

Research Motivation Proposed Approach First Implementation Realism vs. Scalability YANS PhySim Shadowing

ns-3 Models’ Description

Michele Segata, Renato Lo Cigno Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency

slide-4
SLIDE 4

Research Motivation Proposed Approach First Implementation Realism vs. Scalability YANS PhySim Shadowing

YANS - Stochastic model

Chunk based with BER/PER approach

Frame 1 Frame 2 Frame 3 Signal C1 C2 C3 C4 ED threshold Energy

Frame received with probability Pr(f) =

  • ci∈f

1 − Pe(ci).

Michele Segata, Renato Lo Cigno Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency

slide-5
SLIDE 5

Research Motivation Proposed Approach First Implementation Realism vs. Scalability YANS PhySim Shadowing

YANS - Stochastic model

Optimistic (recently, error rate model updated by NIST) Preamble / header decoding phases missing No capture effects Fading model (i.e., Nakagami) does not consider relative speed Frame 1 Frame 2 Signal

Michele Segata, Renato Lo Cigno Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency

slide-6
SLIDE 6

Research Motivation Proposed Approach First Implementation Realism vs. Scalability YANS PhySim Shadowing

PhySim - Emulative model

Emulative - DSP oriented approach Bits -> Scrambling -> Conv. encoding -> Interleaving -> Modulation -> IFFT -> GI -> Samples Signal represented as complex time samples Channel represented through tapped delay line TDL setup using data from real traces for realistic fading Drawback: traces are relative to a fixed scenario

Michele Segata, Renato Lo Cigno Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency

slide-7
SLIDE 7

Research Motivation Proposed Approach First Implementation Realism vs. Scalability YANS PhySim Shadowing

PhySim - Emulative model

Reception = reverse send procedure:

Try to detect preamble and estimate freq. offset Try to decode the PLCP header Try to decode the payload

Natural reproduction of real phenomena High realism Huge computational load

Michele Segata, Renato Lo Cigno Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency

slide-8
SLIDE 8

Research Motivation Proposed Approach First Implementation Realism vs. Scalability YANS PhySim Shadowing

A note on shadowing

Shadowing: additional attenuation caused by obstacles Usually modelled using random fluctuations of signal energy What about this case?

A B

A single truck can cause 20 dB of attenuation (Meireles et. al., "Experimental study on the impact of vehicular

  • bstructions in VANETs", VNC 2010)

Michele Segata, Renato Lo Cigno Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency

slide-9
SLIDE 9

Research Motivation Proposed Approach First Implementation

Proposed Approach

Michele Segata, Renato Lo Cigno Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency

slide-10
SLIDE 10

Research Motivation Proposed Approach First Implementation

Idea: Markov Decision Process

Create a MDP for the PHY receive procedure Tune it with results obtained through PhySim Important parameters:

Current reception phase: RP = {Preamble, Header, payLoad} Vector of interfering frames I F ∈ I = (ts, te, PW, B, ∆f, MC, ∆v) Frame under reception (described as any other frame F)

The state S of the MDP is S = {FS; FR; FD; (RP, I), E} where FS = initial state, FR/FD = absorbing states for receive/discard decision, E = environment

Michele Segata, Renato Lo Cigno Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency

slide-11
SLIDE 11

Research Motivation Proposed Approach First Implementation

MDP Graphical Representation

Michele Segata, Renato Lo Cigno Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency

slide-12
SLIDE 12

Research Motivation Proposed Approach First Implementation Captures Shadowing Fading

First implementation

Features: PHY state machine with captures Simple environment description (cars and trucks) for shadowing effects Uses the NIST BER model

Michele Segata, Renato Lo Cigno Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency

slide-13
SLIDE 13

Research Motivation Proposed Approach First Implementation Captures Shadowing Fading

Fraction of frames generating a capture, 5 dB thr.

2 4 6 8 10 100 200 300 400 500 Percentage of capture frames (%) Number of vehicles (#) Percentage of PHY captures, 8 lanes, 802.11p, 6Mbps, 500 Bytes,10Hz, 5dB thr. Total Preamble Header Payload 2 4 6 8 10 100 200 300 400 500 Percentage of capture frames (%) Number of vehicles (#) Percentage of PHY captures, 8 lanes, 802.11p, 6Mbps, 500 Bytes,10Hz, 5dB thr. Total Preamble Header Payload

ED thr. = -104 dBm, Preamble BUSY over -65 dBm ED thr. = -85 dBm, Preamble BUSY over -85 dBm Michele Segata, Renato Lo Cigno Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency

slide-14
SLIDE 14

Research Motivation Proposed Approach First Implementation Captures Shadowing Fading

Impact of trucks on frame reception

A B

20 40 60 80 100 1 2 3 4 5 6 7 8 9 Percentage of frames received (%) Number of obstructing trucks (#) Without shadowing Shadowing 20 dB, linear Shadowing 20 dB, geometric Shadowing 10 dB, linear Shadowing 10 dB, geometric

Figure: Payload 500 bytes, data rate 6 Mbps

Michele Segata, Renato Lo Cigno Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency

slide-15
SLIDE 15

Research Motivation Proposed Approach First Implementation Captures Shadowing Fading

Impact of relative speed

Work in progress. Can take 1 hour to process 100-200 frames

0.2 0.4 0.6 0.8 1 5 10 15 20 25 30 Fraction of frames received SINR (dB) NIST model ∆v 14 m/s ∆v 28 m/s ∆v 50 m/s ∆v 72 m/s

Figure: Payload 500 bytes, data rate 6 Mbps

Michele Segata, Renato Lo Cigno Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency

slide-16
SLIDE 16

Conclusion The End Appendix Conclusion

Conclusion

Currently available stochastic models are not precise enough for VANETs (PHY, fading, shadowing) A DSP-like approach harms scalability, but is useful for understanding and model derivation An MDP-like approach with enough information can improve precision

Michele Segata, Renato Lo Cigno Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency

slide-17
SLIDE 17

Conclusion The End Appendix

That’s all! Thanks for listening! Questions?

Contacts: {msegata,locigno}@disi.unitn.it

Michele Segata, Renato Lo Cigno Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency

slide-18
SLIDE 18

Conclusion The End Appendix Frequency Offset PHY layer behavior

∆f effects on preamble, header and payload

  • 5

5 10 15 20 SINR (dB)

  • 80
  • 60
  • 40
  • 20

20 40 60 80 Frequency offset (ppm) 0.2 0.4 0.6 0.8 1 Fraction of preamble drops (#)

  • 5

5 10 15 20 SINR (dB)

  • 80
  • 60
  • 40
  • 20

20 40 60 80 Frequency offset (ppm) 0.2 0.4 0.6 0.8 1 Fraction of header drops (#)

  • 5

5 10 15 20 SINR (dB)

  • 80
  • 60
  • 40
  • 20

20 40 60 80 Frequency offset (ppm) 0.2 0.4 0.6 0.8 1 Fraction of payload drops (#)

Michele Segata, Renato Lo Cigno Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency

slide-19
SLIDE 19

Conclusion The End Appendix Frequency Offset PHY layer behavior

PHY layer behavior – Noise floor only

0.2 0.4 0.6 0.8 1

  • 5

5 10 15 20 Fraction of frames (#) SINR (dB) Successfully received Drop at preamble Drop at header Drop at payload

Michele Segata, Renato Lo Cigno Simulation of 802.11 PHY/MAC: the Quest for Accuracy and Efficiency