IMDEA Networks Institute Hans Suys and Bjrn Debaillie Imec Belgium - - PowerPoint PPT Presentation

▶
imdea networks institute
SMART_READER_LITE
LIVE PREVIEW

IMDEA Networks Institute Hans Suys and Bjrn Debaillie Imec Belgium - - PowerPoint PPT Presentation

Apr 07, 2023 369 likes •782 views

Adrian Loch, Hany Assasa, Joan Palacios, and Joerg Widmer IMDEA Networks Institute Hans Suys and Bjrn Debaillie Imec Belgium 1 2 December 14, 2017 Zero Overhead Device Tracking Paper Lamp Searchlight Omnidirectional Wireless

slide-1
SLIDE 1

1

Adrian Loch, Hany Assasa, Joan Palacios, and Joerg Widmer IMDEA Networks Institute Hans Suys and Björn Debaillie Imec Belgium

slide-2
SLIDE 2

2 Zero Overhead Device Tracking

December 14, 2017

Omnidirectional Wireless Communication < 6 GHz Paper Lamp Searchlight

slide-3
SLIDE 3

2 Zero Overhead Device Tracking

December 14, 2017

Omnidirectional Wireless Communication < 6 GHz

Directional Wireless Communication > 6 GHz (mmWave)

Paper Lamp Searchlight

slide-4
SLIDE 4

3 Zero Overhead Device Tracking

December 14, 2017

Node A Node B

Mechanism as in IEEE 802.11ad

slide-5
SLIDE 5

3 Zero Overhead Device Tracking

December 14, 2017

Node A Node B

Mechanism as in IEEE 802.11ad

slide-6
SLIDE 6

3 Zero Overhead Device Tracking

December 14, 2017

Node A Node B

Mechanism as in IEEE 802.11ad

slide-7
SLIDE 7

3 Zero Overhead Device Tracking

December 14, 2017

Node A Node B

Approaches in Related Work

slide-8
SLIDE 8

3 Zero Overhead Device Tracking

December 14, 2017

Node A Node B

Approaches in Related Work

slide-9
SLIDE 9

3 Zero Overhead Device Tracking

December 14, 2017

Node A Node B

Approaches in Related Work

slide-10
SLIDE 10

4 Zero Overhead Device Tracking

December 14, 2017

Node B

Our multi-lobe approach

Node A

slide-11
SLIDE 11

4 Zero Overhead Device Tracking

December 14, 2017

Node B

Our multi-lobe approach

Node A

slide-12
SLIDE 12

4 Zero Overhead Device Tracking

December 14, 2017

Node B

Our multi-lobe approach

Node A

slide-13
SLIDE 13

4 Zero Overhead Device Tracking

December 14, 2017

Node B

Our multi-lobe approach

Node A

slide-14
SLIDE 14

4 Zero Overhead Device Tracking

December 14, 2017

Our multi-lobe approach

Node A

slide-15
SLIDE 15

5

  • Zero Overhead Device Tracking

 Exploiting Preambles in 802.11ad  Golay Sequences for Phase Detection

  • Practical Challenges

 Preamble Robustness  Handling Obstacles

  • Evaluation

 Practical Results  Simulation Results

  • Conclusions

Zero Overhead Device Tracking

December 14, 2017

slide-16
SLIDE 16

www.networks.imdea.org

The core of our mechanism

6

RX TX +1

  • 1
slide-17
SLIDE 17

Data STF Acknowledgment STF

Zero Overhead Device Tracking 7

Receiving half of the preamble using multilobe pattern

  • No modification to the transmitter nor the operation of IEEE 802.11ad
  • Modified receiver fully backward compatible with regular devices
  • Receiver uses misalignment information for both TX and RX steering

December 14, 2017

CEF 1.2 µs 3.63 µs CEF Payload Payload

slide-18
SLIDE 18

Data STF Acknowledgment STF

Zero Overhead Device Tracking 7

Receiving half of the preamble using multilobe pattern

  • No modification to the transmitter nor the operation of IEEE 802.11ad
  • Modified receiver fully backward compatible with regular devices
  • Receiver uses misalignment information for both TX and RX steering

December 14, 2017

CEF 1.2 µs 3.63 µs CEF Payload Payload RX TX RX TX

slide-19
SLIDE 19

Data STF Acknowledgment STF

Zero Overhead Device Tracking 7

Receiving half of the preamble using multilobe pattern

  • No modification to the transmitter nor the operation of IEEE 802.11ad
  • Modified receiver fully backward compatible with regular devices
  • Receiver uses misalignment information for both TX and RX steering

December 14, 2017

CEF 1.2 µs 3.63 µs CEF Payload Payload RX TX RX TX Pattern switch time is below 50 ns

  • n state-of-the-art antennas and

can be as low as 50 ps

slide-20
SLIDE 20

Zero Overhead Device Tracking 8

Designing two-lobe beampatterns to track devices

  • Relative phase-shift among the two lobes is designed to be 180º
  • Receiver receives part of the preamble using the two-lobe beampattern
  • Comparing first and second half of preamble reveals orientation

Two-lobe beampatterns are feasible using analog beamforming

December 14, 2017

slide-21
SLIDE 21

Zero Overhead Device Tracking 9

Detecting phase shift among preamble halves

  • Direct phase comparison is challenging due to beam pattern change
  • Properties of Golay sequences allow for robust phase shift detection
  • Correlation results in a positive (0º shift) or negative (180º shift) spike

Obtain Golay sequence via one-lobe and two-lobe beam pattern Compute cross-correlation of both sequences Rotation to the left Rotation to the right Well aligned

December 14, 2017

slide-22
SLIDE 22

www.networks.imdea.org

Making our approach work on practical hardware

10

slide-23
SLIDE 23

Zero Overhead Device Tracking 11

Packet detection with only part of the preamble STF

  • A receiver may only receive half of the STF in the worst case
  • We show in practice that an 802.11ad decoder works with such an STF
  • Receiver can equalize packet since CEF is received with regular pattern

December 14, 2017

TX: COTS 802.11ad RX: Keysight Wideband Waveform Center

slide-24
SLIDE 24

Zero Overhead Device Tracking 12

Obstacles and misalignments are radically different

  • In case of antenna misalignment, the current path is still available
  • In case of blockage, the nodes must find an entirely new path
  • 802.11ad beacon sweeps address both but we focus on misalignment

December 14, 2017

slide-25
SLIDE 25

Zero Overhead Device Tracking 12

Obstacles and misalignments are radically different

  • In case of antenna misalignment, the current path is still available
  • In case of blockage, the nodes must find an entirely new path
  • 802.11ad beacon sweeps address both but we focus on misalignment

December 14, 2017

 

Our mechanism can track both paths individually

slide-26
SLIDE 26

Zero Overhead Device Tracking 12

Obstacles and misalignments are radically different

  • In case of antenna misalignment, the current path is still available
  • In case of blockage, the nodes must find an entirely new path
  • 802.11ad beacon sweeps address both but we focus on misalignment

December 14, 2017

 

Our mechanism can track both paths individually Finding the reflected path is a separate problem

slide-27
SLIDE 27

www.networks.imdea.org

Practical and simulative results

13

slide-28
SLIDE 28

Zero Overhead Device Tracking 14

Implementation on phased antenna array at IMEC

  • Highly flexible 60 GHz frontend featuring 2x8 antenna elements
  • Antenna allows for phase and amplitude control of each element
  • Testbed is fully controlable from Matlab, allowing for rapid prototyping

Control PC and TX Equipment RX Equipment

TX Antenna

Signal Generator

Control PC

Differential IQ Differential IQ Oscilloscope

RX Antenna

December 14, 2017

slide-29
SLIDE 29

Zero Overhead Device Tracking 15

Implementation on phased antenna array at IMEC

  • One side of the link rotates according to real-world gyroscope traces
  • Automatic beam-steering adjustment based on correlation output
  • Steering error always below 5º which results in up to 2x throughput gain

Seamless and fast error recovery Walking movement at indoor speed

December 14, 2017

slide-30
SLIDE 30

16 Zero Overhead Device Tracking

December 14, 2017

We achieve significant performance improvements

  • We measure performance both in terms of throughput and angle error
  • Device tracking can maintain a high rate in spite of movement/rotation
  • The angle error is below the outage threshold even for strong rotations

Periodic controlled rotation of one of the ends of the link

slide-31
SLIDE 31

Zero Overhead Device Tracking 17

Full protocol stack implementation on NS-3 and Matlab

  • Matlab symbol-level simulation including raytracer and 60 GHz model
  • NS-3 packet-level simulation in direct execution mode with TCP Cubic
  • Large number of randomly generated scenarios using gyroscope traces

802.11ad suffers massive SNR drops due to rotation whereas our approach continuously adapts to such rotations Our approach reduces the angle error to below 3º in almost all cases

December 14, 2017

slide-32
SLIDE 32

Zero Overhead Device Tracking 17

Full protocol stack implementation on NS-3 and Matlab

  • Matlab symbol-level simulation including raytracer and 60 GHz model
  • NS-3 packet-level simulation in direct execution mode with TCP Cubic
  • Large number of randomly generated scenarios using gyroscope traces

802.11ad suffers massive SNR drops due to rotation whereas our approach continuously adapts to such rotations Our approach reduces the angle error to below 3º in almost all cases

December 14, 2017

slide-33
SLIDE 33

www.networks.imdea.org

Summary of our insights

18

slide-34
SLIDE 34

Zero Overhead Device Tracking 19

December 14, 2017

slide-35
SLIDE 35

Zero Overhead Device Tracking 19

December 14, 2017

  • Goal: track movement and rotation of IEEE 802.11ad devices with

zero overhead and no changes to the operation of the standard

slide-36
SLIDE 36

Zero Overhead Device Tracking 19

December 14, 2017

  • Goal: track movement and rotation of IEEE 802.11ad devices with

zero overhead and no changes to the operation of the standard

  • Challenges:

 Designing multi-lobe beam patterns that allow for misalignment tracking  Detecting the phase shift among non-equalized preamble halves

slide-37
SLIDE 37

Zero Overhead Device Tracking 19

December 14, 2017

  • Goal: track movement and rotation of IEEE 802.11ad devices with

zero overhead and no changes to the operation of the standard

  • Challenges:

 Designing multi-lobe beam patterns that allow for misalignment tracking  Detecting the phase shift among non-equalized preamble halves  Accurate per-packet movement and rotation tracking of a path  Mechanism is fully backward compatible with IEEE 802.11ad devices

  • 1. No change to the operation nor frame format of IEEE 802.11ad
  • 2. Mechanism works even when communicating with legacy nodes
  • Contributions and results:
slide-38
SLIDE 38

Zero Overhead Device Tracking 19

December 14, 2017

  • Goal: track movement and rotation of IEEE 802.11ad devices with

zero overhead and no changes to the operation of the standard

  • Challenges:

 Designing multi-lobe beam patterns that allow for misalignment tracking  Detecting the phase shift among non-equalized preamble halves  Practical implementation on a full bandwidth 60 GHz testbed

  • 1. Testbed features an electronically steerable phased antenna array
  • 2. We achieve an angle error below 5º for most cases

 Extensive simulation campaign achieving up to 2.38x higher throughput

  • Contributions and results:
slide-39
SLIDE 39

Zero Overhead Device Tracking 20

December 14, 2017

Implemented and tested on the PHARA4 60 GHz WiFi solution

slide-40
SLIDE 40

21

December 14, 2017

Zero Overhead Device Tracking