Upload
others
View
11
Download
0
Embed Size (px)
Citation preview
Accurate and Precise Distance Estimationfrom Phase-based Ranging DataIPIN 2018, Nantes, FranceYannic Schröder, Dennis Reimers and Lars WolfInstitute of Operating Systems and Computer Networks, TU Braunschweig, Germany
Institute of Operating Systemsand Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Overview
Data Acquisition→ Phase-based Ranging
Distance Computation→ CDE AlgorithmAlgorithm Speedup→ InterpolationComparison to two other algorithms→ Evaluation
Yannic Schröder, Dennis Reimers and Lars Wolf Page 2Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Overview
Data Acquisition→ Phase-based RangingDistance Computation→ CDE Algorithm
Algorithm Speedup→ InterpolationComparison to two other algorithms→ Evaluation
Yannic Schröder, Dennis Reimers and Lars Wolf Page 2Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Overview
Data Acquisition→ Phase-based RangingDistance Computation→ CDE AlgorithmAlgorithm Speedup→ Interpolation
Comparison to two other algorithms→ Evaluation
Yannic Schröder, Dennis Reimers and Lars Wolf Page 2Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Overview
Data Acquisition→ Phase-based RangingDistance Computation→ CDE AlgorithmAlgorithm Speedup→ InterpolationComparison to two other algorithms→ Evaluation
Yannic Schröder, Dennis Reimers and Lars Wolf Page 2Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Phase-based Ranging
Obtain distance in meters between two wireless sensor nodes
Radio transceivers with Phase Measurement Units (e.g. AT86RF233)Measure phase response of radio channel between nodes
Exemplary phase data for meter distance:
Distance is proportional to slope/frequency of phase response
Yannic Schröder, Dennis Reimers and Lars Wolf Page 3Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Phase-based Ranging
Obtain distance in meters between two wireless sensor nodesRadio transceivers with Phase Measurement Units (e.g. AT86RF233)
Measure phase response of radio channel between nodes
Exemplary phase data for meter distance:
Distance is proportional to slope/frequency of phase response
Yannic Schröder, Dennis Reimers and Lars Wolf Page 3Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Phase-based Ranging
Obtain distance in meters between two wireless sensor nodesRadio transceivers with Phase Measurement Units (e.g. AT86RF233)Measure phase response of radio channel between nodes
Exemplary phase data for meter distance:
Distance is proportional to slope/frequency of phase response
Yannic Schröder, Dennis Reimers and Lars Wolf Page 3Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Phase-based Ranging
Obtain distance in meters between two wireless sensor nodesRadio transceivers with Phase Measurement Units (e.g. AT86RF233)Measure phase response of radio channel between nodes
Exemplary phase data for 5 meter distance:
2400 2420 2440 2460 2480 2500Frequency [MHz]
−π
0
π
Phas
ean
gle
ϕ
Distance is proportional to slope/frequency of phase response
Yannic Schröder, Dennis Reimers and Lars Wolf Page 3Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Phase-based Ranging
Obtain distance in meters between two wireless sensor nodesRadio transceivers with Phase Measurement Units (e.g. AT86RF233)Measure phase response of radio channel between nodes
Exemplary phase data for 10 meter distance:
2400 2420 2440 2460 2480 2500Frequency [MHz]
−π
0
π
Phas
ean
gle
ϕ
Distance is proportional to slope/frequency of phase response
Yannic Schröder, Dennis Reimers and Lars Wolf Page 3Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Phase-based Ranging
Obtain distance in meters between two wireless sensor nodesRadio transceivers with Phase Measurement Units (e.g. AT86RF233)Measure phase response of radio channel between nodes
Exemplary phase data for 10 meter distance:
2400 2420 2440 2460 2480 2500Frequency [MHz]
−π
0
π
Phas
ean
gle
ϕ
Distance is proportional to slope/frequency of phase response
Yannic Schröder, Dennis Reimers and Lars Wolf Page 3Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Complex-valued Distance Estimation (CDE)
Compute Fast Fourier Transform (FFT) from complex signal
Result is the time-domain impulse response of the radio channelMaximum peak indicates the distance (via propagation speed c)
Peak height = Distance Quality Indicator (DQI)
Exemplary impulse response for 5 meter distance:
Yannic Schröder, Dennis Reimers and Lars Wolf Page 4Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Complex-valued Distance Estimation (CDE)
Compute Fast Fourier Transform (FFT) from complex signalResult is the time-domain impulse response of the radio channel
Maximum peak indicates the distance (via propagation speed c)
Peak height = Distance Quality Indicator (DQI)
Exemplary impulse response for 5 meter distance:
0 50 100 150 200 250 300dPMU [m]
0.0
0.5
0.7
1.0
DQ
I
Yannic Schröder, Dennis Reimers and Lars Wolf Page 4Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Complex-valued Distance Estimation (CDE)
Compute Fast Fourier Transform (FFT) from complex signalResult is the time-domain impulse response of the radio channelMaximum peak indicates the distance (via propagation speed c)
Peak height = Distance Quality Indicator (DQI)
Exemplary impulse response for 5 meter distance:
0 50 100 150 200 250 300dPMU [m]
0.0
0.5
0.7
1.0
DQ
I
Yannic Schröder, Dennis Reimers and Lars Wolf Page 4Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Complex-valued Distance Estimation (CDE)
Compute Fast Fourier Transform (FFT) from complex signalResult is the time-domain impulse response of the radio channelMaximum peak indicates the distance (via propagation speed c)
Peak height = Distance Quality Indicator (DQI)
Exemplary impulse response for 5 meter distance:
0 50 100 150 200 250 300dPMU [m]
0.0
0.5
0.7
1.0
DQ
I
Yannic Schröder, Dennis Reimers and Lars Wolf Page 4Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Complex-valued Distance Estimation (CDE)
Compute Fast Fourier Transform (FFT) from complex signalResult is the time-domain impulse response of the radio channelMaximum peak indicates the distance (via propagation speed c)
Peak height = Distance Quality Indicator (DQI)
Exemplary impulse response for 5 and 10 meter distance:
0 50 100 150 200 250 300dPMU [m]
0.0
0.5
0.7
1.0
DQ
I
Yannic Schröder, Dennis Reimers and Lars Wolf Page 4Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
FFT Bin Count
What is the optimal FFT bin count?
What happens if the phase response is noisy?212 = 4096 bins reach lower bound on accuracy
Yannic Schröder, Dennis Reimers and Lars Wolf Page 5Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
FFT Bin Count
What is the optimal FFT bin count?
What happens if the phase response is noisy?
212 = 4096 bins reach lower bound on accuracy
0 50 100 150 200 250 300dPMU [m]
0.0
0.5
0.7
1.0
DQ
I
Yannic Schröder, Dennis Reimers and Lars Wolf Page 5Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
FFT Bin Count
What is the optimal FFT bin count?
What happens if the phase response is noisy?212 = 4096 bins reach lower bound on accuracy
5.4 5.6 5.8 6.0 6.2 6.4 6.6 6.8 7.0dPMU [m]
0.62
0.64
0.66
0.68
0.70
0.72
DQ
I
1024 bins
Yannic Schröder, Dennis Reimers and Lars Wolf Page 5Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
FFT Bin Count
What is the optimal FFT bin count?
What happens if the phase response is noisy?212 = 4096 bins reach lower bound on accuracy
5.4 5.6 5.8 6.0 6.2 6.4 6.6 6.8 7.0dPMU [m]
0.62
0.64
0.66
0.68
0.70
0.72
DQ
I
1024 bins4096 bins
Yannic Schröder, Dennis Reimers and Lars Wolf Page 5Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
FFT Bin Count
What is the optimal FFT bin count?
What happens if the phase response is noisy?212 = 4096 bins reach lower bound on accuracy
27 28 29 210 211 212 213 214 215
FFT bin count B
10−3
10−2
10−1
100
Mea
nA
bsol
ute
Err
or[m
]
CDE w/o noise
Yannic Schröder, Dennis Reimers and Lars Wolf Page 5Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
FFT Bin Count
What is the optimal FFT bin count?What happens if the phase response is noisy?
212 = 4096 bins reach lower bound on accuracy
27 28 29 210 211 212 213 214 215
FFT bin count B
10−3
10−2
10−1
100
Mea
nA
bsol
ute
Err
or[m
]
CDE w/o noise
Yannic Schröder, Dennis Reimers and Lars Wolf Page 5Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
FFT Bin Count
What is the optimal FFT bin count?What happens if the phase response is noisy?
212 = 4096 bins reach lower bound on accuracy
27 28 29 210 211 212 213 214 215
FFT bin count B
10−3
10−2
10−1
100
Mea
nA
bsol
ute
Err
or[m
]
CDE w/o noiseCDE w/ noise
Yannic Schröder, Dennis Reimers and Lars Wolf Page 5Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
FFT Bin Count
What is the optimal FFT bin count?What happens if the phase response is noisy?212 = 4096 bins reach lower bound on accuracy
27 28 29 210 211 212 213 214 215
FFT bin count B
10−3
10−2
10−1
100
Mea
nA
bsol
ute
Err
or[m
]
CDE w/o noiseCDE w/ noise
Yannic Schröder, Dennis Reimers and Lars Wolf Page 5Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Interpolation
Can we reduce the FFT bin count by interpolating the peak?
29 = 512 bins suffice with interpolation
Yannic Schröder, Dennis Reimers and Lars Wolf Page 6Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Interpolation
Can we reduce the FFT bin count by interpolating the peak?
29 = 512 bins suffice with interpolation
5.4 5.6 5.8 6.0 6.2 6.4 6.6 6.8 7.0dPMU [m]
0.62
0.64
0.66
0.68
0.70
0.72
DQ
I
1024 bins
Yannic Schröder, Dennis Reimers and Lars Wolf Page 6Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Interpolation
Can we reduce the FFT bin count by interpolating the peak?
29 = 512 bins suffice with interpolation
5.4 5.6 5.8 6.0 6.2 6.4 6.6 6.8 7.0dPMU [m]
0.62
0.64
0.66
0.68
0.70
0.72
DQ
I
1024 binsinterpolated
Yannic Schröder, Dennis Reimers and Lars Wolf Page 6Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Interpolation
Can we reduce the FFT bin count by interpolating the peak?
29 = 512 bins suffice with interpolation
27 28 29 210 211 212 213 214 215
FFT bin count B
10−3
10−2
10−1
100
Mea
nA
bsol
ute
Err
or[m
]
CDE w/o noiseCDE w/ noiseinterpolated CDE w/ noise
Yannic Schröder, Dennis Reimers and Lars Wolf Page 6Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Interpolation
Can we reduce the FFT bin count by interpolating the peak?29 = 512 bins suffice with interpolation
27 28 29 210 211 212 213 214 215
FFT bin count B
10−3
10−2
10−1
100
Mea
nA
bsol
ute
Err
or[m
]
CDE w/o noiseCDE w/ noiseinterpolated CDE w/ noise
Yannic Schröder, Dennis Reimers and Lars Wolf Page 6Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Evaluation Setup
Phase-based data from 4 scenarios
Comparison of regular CDE with interpolated version (iCDE)Comparison with two other algorithms
RDE: Real-valued Distance EstimationESSR: Efficient Slope Sampling Ranging by Oshiga et al.
Accuracy: Mean Absolute Error (MAE)Precision: Standard Deviation (σ)All algorithms use the exact same data set
Yannic Schröder, Dennis Reimers and Lars Wolf Page 7Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Evaluation Setup
Phase-based data from 4 scenariosComparison of regular CDE with interpolated version (iCDE)
Comparison with two other algorithms
RDE: Real-valued Distance EstimationESSR: Efficient Slope Sampling Ranging by Oshiga et al.
Accuracy: Mean Absolute Error (MAE)Precision: Standard Deviation (σ)All algorithms use the exact same data set
Yannic Schröder, Dennis Reimers and Lars Wolf Page 7Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Evaluation Setup
Phase-based data from 4 scenariosComparison of regular CDE with interpolated version (iCDE)Comparison with two other algorithms
RDE: Real-valued Distance EstimationESSR: Efficient Slope Sampling Ranging by Oshiga et al.
Accuracy: Mean Absolute Error (MAE)Precision: Standard Deviation (σ)All algorithms use the exact same data set
Yannic Schröder, Dennis Reimers and Lars Wolf Page 7Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Evaluation Setup
Phase-based data from 4 scenariosComparison of regular CDE with interpolated version (iCDE)Comparison with two other algorithms
RDE: Real-valued Distance Estimation
ESSR: Efficient Slope Sampling Ranging by Oshiga et al.Accuracy: Mean Absolute Error (MAE)Precision: Standard Deviation (σ)All algorithms use the exact same data set
Yannic Schröder, Dennis Reimers and Lars Wolf Page 7Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Evaluation Setup
Phase-based data from 4 scenariosComparison of regular CDE with interpolated version (iCDE)Comparison with two other algorithms
RDE: Real-valued Distance EstimationESSR: Efficient Slope Sampling Ranging by Oshiga et al.
Accuracy: Mean Absolute Error (MAE)Precision: Standard Deviation (σ)All algorithms use the exact same data set
Yannic Schröder, Dennis Reimers and Lars Wolf Page 7Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Evaluation Setup
Phase-based data from 4 scenariosComparison of regular CDE with interpolated version (iCDE)Comparison with two other algorithms
RDE: Real-valued Distance EstimationESSR: Efficient Slope Sampling Ranging by Oshiga et al.
Accuracy: Mean Absolute Error (MAE)
Precision: Standard Deviation (σ)All algorithms use the exact same data set
Yannic Schröder, Dennis Reimers and Lars Wolf Page 7Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Evaluation Setup
Phase-based data from 4 scenariosComparison of regular CDE with interpolated version (iCDE)Comparison with two other algorithms
RDE: Real-valued Distance EstimationESSR: Efficient Slope Sampling Ranging by Oshiga et al.
Accuracy: Mean Absolute Error (MAE)Precision: Standard Deviation (σ)
All algorithms use the exact same data set
Yannic Schröder, Dennis Reimers and Lars Wolf Page 7Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Evaluation Setup
Phase-based data from 4 scenariosComparison of regular CDE with interpolated version (iCDE)Comparison with two other algorithms
RDE: Real-valued Distance EstimationESSR: Efficient Slope Sampling Ranging by Oshiga et al.
Accuracy: Mean Absolute Error (MAE)Precision: Standard Deviation (σ)All algorithms use the exact same data set
Yannic Schröder, Dennis Reimers and Lars Wolf Page 7Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Park Scenario
−4 −3 −2 −1 0 1 2 3 4Error [m]
0.00
0.25
0.50
0.75
1.00C
DF
Park
CDEiCDERDEESSR
Interpolation has no influence on the resultRDE and ESSR are accurate but not precise
Yannic Schröder, Dennis Reimers and Lars Wolf Page 8Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Park Scenario
−4 −3 −2 −1 0 1 2 3 4Error [m]
0.00
0.25
0.50
0.75
1.00C
DF
Park
CDEiCDERDEESSR
Interpolation has no influence on the result
RDE and ESSR are accurate but not precise
Yannic Schröder, Dennis Reimers and Lars Wolf Page 8Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Park Scenario
−4 −3 −2 −1 0 1 2 3 4Error [m]
0.00
0.25
0.50
0.75
1.00C
DF
Park
CDEiCDERDEESSR
Interpolation has no influence on the resultRDE and ESSR are accurate but not precise
Yannic Schröder, Dennis Reimers and Lars Wolf Page 8Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Errors of the Park Scenario
min. [m] max. [m] median [m] MAE [m] σ [m]CDE -0.493 0.638 0.131 0.149 0.104
iCDE -0.507 0.652 0.139 0.151 0.103RDE -69.187 3.648 0.331 0.706 4.292
ESSR -40.668 282.824 0.626 2.555 19.653
High accuracy: MAE = 0.149 mHigh precision: σ = 0.104 mCDE and iCDE perform almost identicalLarge outliers for RDE and ESSR
Yannic Schröder, Dennis Reimers and Lars Wolf Page 9Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Errors of the Park Scenario
min. [m] max. [m] median [m] MAE [m] σ [m]CDE -0.493 0.638 0.131 0.149 0.104
iCDE -0.507 0.652 0.139 0.151 0.103RDE -69.187 3.648 0.331 0.706 4.292
ESSR -40.668 282.824 0.626 2.555 19.653
High accuracy: MAE = 0.149 m
High precision: σ = 0.104 mCDE and iCDE perform almost identicalLarge outliers for RDE and ESSR
Yannic Schröder, Dennis Reimers and Lars Wolf Page 9Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Errors of the Park Scenario
min. [m] max. [m] median [m] MAE [m] σ [m]CDE -0.493 0.638 0.131 0.149 0.104
iCDE -0.507 0.652 0.139 0.151 0.103RDE -69.187 3.648 0.331 0.706 4.292
ESSR -40.668 282.824 0.626 2.555 19.653
High accuracy: MAE = 0.149 mHigh precision: σ = 0.104 m
CDE and iCDE perform almost identicalLarge outliers for RDE and ESSR
Yannic Schröder, Dennis Reimers and Lars Wolf Page 9Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Errors of the Park Scenario
min. [m] max. [m] median [m] MAE [m] σ [m]CDE -0.493 0.638 0.131 0.149 0.104
iCDE -0.507 0.652 0.139 0.151 0.103RDE -69.187 3.648 0.331 0.706 4.292
ESSR -40.668 282.824 0.626 2.555 19.653
High accuracy: MAE = 0.149 mHigh precision: σ = 0.104 mCDE and iCDE perform almost identical
Large outliers for RDE and ESSR
Yannic Schröder, Dennis Reimers and Lars Wolf Page 9Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Errors of the Park Scenario
min. [m] max. [m] median [m] MAE [m] σ [m]CDE -0.493 0.638 0.131 0.149 0.104
iCDE -0.507 0.652 0.139 0.151 0.103RDE -69.187 3.648 0.331 0.706 4.292
ESSR -40.668 282.824 0.626 2.555 19.653
High accuracy: MAE = 0.149 mHigh precision: σ = 0.104 mCDE and iCDE perform almost identicalLarge outliers for RDE and ESSR
Yannic Schröder, Dennis Reimers and Lars Wolf Page 9Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Office corridor Scenario
−4 −3 −2 −1 0 1 2 3 4Error [m]
0.00
0.25
0.50
0.75
1.00C
DF
Office corridor
CDEiCDERDEESSR
RDE and ESSR overestimate the distanceMultipath effects likely lead to wrong measurements (plateaus)
Yannic Schröder, Dennis Reimers and Lars Wolf Page 10Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Office corridor Scenario
−4 −3 −2 −1 0 1 2 3 4Error [m]
0.00
0.25
0.50
0.75
1.00C
DF
Office corridor
CDEiCDERDEESSR
RDE and ESSR overestimate the distance
Multipath effects likely lead to wrong measurements (plateaus)
Yannic Schröder, Dennis Reimers and Lars Wolf Page 10Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Office corridor Scenario
−4 −3 −2 −1 0 1 2 3 4Error [m]
0.00
0.25
0.50
0.75
1.00C
DF
Office corridor
CDEiCDERDEESSR
RDE and ESSR overestimate the distanceMultipath effects likely lead to wrong measurements (plateaus)
Yannic Schröder, Dennis Reimers and Lars Wolf Page 10Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Errors of the Office corridor Scenario
min. [m] max. [m] median [m] MAE [m] σ [m]CDE -0.810 3.199 0.299 0.550 0.738
iCDE -0.807 3.169 0.297 0.542 0.730RDE -29.686 24.734 0.453 0.850 2.218
ESSR -2.567 296.961 0.660 1.468 13.012
Scenario is more challengingCDE/iCDE with much higher precision than RDE and ESSRAgain large outliers for RDE and ESSR
Yannic Schröder, Dennis Reimers and Lars Wolf Page 11Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Errors of the Office corridor Scenario
min. [m] max. [m] median [m] MAE [m] σ [m]CDE -0.810 3.199 0.299 0.550 0.738
iCDE -0.807 3.169 0.297 0.542 0.730RDE -29.686 24.734 0.453 0.850 2.218
ESSR -2.567 296.961 0.660 1.468 13.012
Scenario is more challenging
CDE/iCDE with much higher precision than RDE and ESSRAgain large outliers for RDE and ESSR
Yannic Schröder, Dennis Reimers and Lars Wolf Page 11Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Errors of the Office corridor Scenario
min. [m] max. [m] median [m] MAE [m] σ [m]CDE -0.810 3.199 0.299 0.550 0.738
iCDE -0.807 3.169 0.297 0.542 0.730RDE -29.686 24.734 0.453 0.850 2.218
ESSR -2.567 296.961 0.660 1.468 13.012
Scenario is more challengingCDE/iCDE with much higher precision than RDE and ESSR
Again large outliers for RDE and ESSR
Yannic Schröder, Dennis Reimers and Lars Wolf Page 11Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Errors of the Office corridor Scenario
min. [m] max. [m] median [m] MAE [m] σ [m]CDE -0.810 3.199 0.299 0.550 0.738
iCDE -0.807 3.169 0.297 0.542 0.730RDE -29.686 24.734 0.453 0.850 2.218
ESSR -2.567 296.961 0.660 1.468 13.012
Scenario is more challengingCDE/iCDE with much higher precision than RDE and ESSRAgain large outliers for RDE and ESSR
Yannic Schröder, Dennis Reimers and Lars Wolf Page 11Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Conclusion
CDE algorithm to compute distance from phase response
Interpolation to reduce number of FFT binsComparison to two other state-of-the-art algorithmsEvaluation in real-world environmentsHigh accuracy: MAE = 14.9 cmHigh precision: σ = 10.4 cm
Whats more in the paper?
MathFilter bad measurements via Distance Quality IndicatorTwo more scenarios (Apartment, Basement)
Thank you for your attention!
Yannic Schröder, Dennis Reimers and Lars Wolf Page 12Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Conclusion
CDE algorithm to compute distance from phase responseInterpolation to reduce number of FFT bins
Comparison to two other state-of-the-art algorithmsEvaluation in real-world environmentsHigh accuracy: MAE = 14.9 cmHigh precision: σ = 10.4 cm
Whats more in the paper?
MathFilter bad measurements via Distance Quality IndicatorTwo more scenarios (Apartment, Basement)
Thank you for your attention!
Yannic Schröder, Dennis Reimers and Lars Wolf Page 12Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Conclusion
CDE algorithm to compute distance from phase responseInterpolation to reduce number of FFT binsComparison to two other state-of-the-art algorithms
Evaluation in real-world environmentsHigh accuracy: MAE = 14.9 cmHigh precision: σ = 10.4 cm
Whats more in the paper?
MathFilter bad measurements via Distance Quality IndicatorTwo more scenarios (Apartment, Basement)
Thank you for your attention!
Yannic Schröder, Dennis Reimers and Lars Wolf Page 12Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Conclusion
CDE algorithm to compute distance from phase responseInterpolation to reduce number of FFT binsComparison to two other state-of-the-art algorithmsEvaluation in real-world environments
High accuracy: MAE = 14.9 cmHigh precision: σ = 10.4 cm
Whats more in the paper?
MathFilter bad measurements via Distance Quality IndicatorTwo more scenarios (Apartment, Basement)
Thank you for your attention!
Yannic Schröder, Dennis Reimers and Lars Wolf Page 12Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Conclusion
CDE algorithm to compute distance from phase responseInterpolation to reduce number of FFT binsComparison to two other state-of-the-art algorithmsEvaluation in real-world environmentsHigh accuracy: MAE = 14.9 cm
High precision: σ = 10.4 cm
Whats more in the paper?
MathFilter bad measurements via Distance Quality IndicatorTwo more scenarios (Apartment, Basement)
Thank you for your attention!
Yannic Schröder, Dennis Reimers and Lars Wolf Page 12Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Conclusion
CDE algorithm to compute distance from phase responseInterpolation to reduce number of FFT binsComparison to two other state-of-the-art algorithmsEvaluation in real-world environmentsHigh accuracy: MAE = 14.9 cmHigh precision: σ = 10.4 cm
Whats more in the paper?
MathFilter bad measurements via Distance Quality IndicatorTwo more scenarios (Apartment, Basement)
Thank you for your attention!
Yannic Schröder, Dennis Reimers and Lars Wolf Page 12Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Conclusion
CDE algorithm to compute distance from phase responseInterpolation to reduce number of FFT binsComparison to two other state-of-the-art algorithmsEvaluation in real-world environmentsHigh accuracy: MAE = 14.9 cmHigh precision: σ = 10.4 cm
Whats more in the paper?
MathFilter bad measurements via Distance Quality IndicatorTwo more scenarios (Apartment, Basement)
Thank you for your attention!
Yannic Schröder, Dennis Reimers and Lars Wolf Page 12Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Conclusion
CDE algorithm to compute distance from phase responseInterpolation to reduce number of FFT binsComparison to two other state-of-the-art algorithmsEvaluation in real-world environmentsHigh accuracy: MAE = 14.9 cmHigh precision: σ = 10.4 cm
Whats more in the paper?Math
Filter bad measurements via Distance Quality IndicatorTwo more scenarios (Apartment, Basement)
Thank you for your attention!
Yannic Schröder, Dennis Reimers and Lars Wolf Page 12Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Conclusion
CDE algorithm to compute distance from phase responseInterpolation to reduce number of FFT binsComparison to two other state-of-the-art algorithmsEvaluation in real-world environmentsHigh accuracy: MAE = 14.9 cmHigh precision: σ = 10.4 cm
Whats more in the paper?MathFilter bad measurements via Distance Quality Indicator
Two more scenarios (Apartment, Basement)
Thank you for your attention!
Yannic Schröder, Dennis Reimers and Lars Wolf Page 12Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Conclusion
CDE algorithm to compute distance from phase responseInterpolation to reduce number of FFT binsComparison to two other state-of-the-art algorithmsEvaluation in real-world environmentsHigh accuracy: MAE = 14.9 cmHigh precision: σ = 10.4 cm
Whats more in the paper?MathFilter bad measurements via Distance Quality IndicatorTwo more scenarios (Apartment, Basement)
Thank you for your attention!
Yannic Schröder, Dennis Reimers and Lars Wolf Page 12Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
Phase-based Ranging CDE Algorithm Interpolation Evaluation Conclusion
Conclusion
CDE algorithm to compute distance from phase responseInterpolation to reduce number of FFT binsComparison to two other state-of-the-art algorithmsEvaluation in real-world environmentsHigh accuracy: MAE = 14.9 cmHigh precision: σ = 10.4 cm
Whats more in the paper?MathFilter bad measurements via Distance Quality IndicatorTwo more scenarios (Apartment, Basement)
Thank you for your attention!Yannic Schröder, Dennis Reimers and Lars Wolf Page 12Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
CDE Algorithm Results Interpolation
Complex-valued Distance Estimation
Compute complex signal from phase responseAssume amplitudes as 1
NExemplary complex signal for 5 meter distance:
− 1N
0
1N
Re{
Z}
2400 2420 2440 2460 2480 2500Frequency [MHz]
− 1N
0
1N
Im{Z}
Yannic Schröder, Dennis Reimers and Lars Wolf Page 13Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
CDE Algorithm Results Interpolation
Apartment Scenario
−4 −3 −2 −1 0 1 2 3 4Error [m]
0.00
0.25
0.50
0.75
1.00
CD
F
Apartment
CDEiCDERDEESSR
Yannic Schröder, Dennis Reimers and Lars Wolf Page 14Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
CDE Algorithm Results Interpolation
Errors of the Apartment Scenario
min. [m] max. [m] median [m] MAE [m] σ [m]CDE -0.701 0.140 0.372 0.376 0.146
iCDE -0.711 0.145 0.384 0.377 0.147RDE -1.035 41.818 0.346 0.862 3.794
ESSR -1.799 294.078 0.615 3.313 26.353
Yannic Schröder, Dennis Reimers and Lars Wolf Page 15Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
CDE Algorithm Results Interpolation
Basement Scenario
−4 −3 −2 −1 0 1 2 3 4Error [m]
0.00
0.25
0.50
0.75
1.00
CD
F
Basement
CDEiCDERDEESSR
Yannic Schröder, Dennis Reimers and Lars Wolf Page 16Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
CDE Algorithm Results Interpolation
Errors of the Basement Scenario
min. [m] max. [m] median [m] MAE [m] σ [m]CDE -2.136 2.329 0.204 0.414 0.548
iCDE -2.144 2.350 0.216 0.409 0.532RDE -2.172 29.598 0.293 0.890 2.850
ESSR -2.355 3.740 0.652 0.766 0.601
Yannic Schröder, Dennis Reimers and Lars Wolf Page 17Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
CDE Algorithm Results Interpolation
Parameters of the Scenarios
# meas. min. dist. [m] max. dist. [m]Basement 350 0.5 20.0
Office corridor 650 0.5 49.0Apartment 350 1.0 7.0
Park 950 0.5 100.0
Yannic Schröder, Dennis Reimers and Lars Wolf Page 18Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks
CDE Algorithm Results Interpolation
Interpolation without Noise
27 28 29 210 211 212 213 214 215
FFT bin count B
10−3
10−2
10−1
100
Mea
nA
bsol
ute
Err
or[m
]
CDE w/o noiseCDE w/ noiseinterpolated CDE w/o noiseinterpolated CDE w/ noise
Yannic Schröder, Dennis Reimers and Lars Wolf Page 19Accurate and Precise Distance Estimation from Phase-based Ranging Data Institute of Operating Systems
and Computer Networks