1. An Investigation on SYNC Detector False Alarms
Month Year
doc.: IEEE yy/1634r0
An Investigation on SYNC Detector False Alarms
Date:
Authors:
Kome Oteri(InterDigital)

2. SYNC templates (HDR/LDR)
Introduction
In [1], the responses of the HDR and LDR SYNC detectors to the HDR and LDR data fields have been studied:
It was shown that, the correlation between the WUR payload and SYNC can be high, which may create a false alarm in some WURx implementations
In this study, we discuss an alternative SYNC detector to lower the false alarm rate while maintaining the same miss detection
Legacy
SYNC
Data
TX:
RX window
SYNC detection
SYNC templates (HDR/LDR)
Slide
e.g., WURx wakes up
False alarms?

3. SYNC templates (HDR/LDR)
SYNC Detector
In [1], the SYNC detector compares the ratio of the correlation output (𝑥) and the sum samples (𝑦) with a threshold between 0 and 1
e.g., WURx wakes up
RX window
Legacy
TX:
SYNC
Data
Slide 𝑦
Sum of abs samples
False alarms due to the high correlation?
x/y≷threshold?
Correlate
SYNC templates (HDR/LDR) 𝑥

4. Threshold Selection
If the threshold is set based on the statistics of the maximum of the detector output within a period for AWGN (e.g., ), it has been observed that the false alarm rate is 100% for high SNR [1]
A higher threshold can increase the miss detection rate. This issue can be addressed by using a different detector design
100% false alarm!
Threshold

5. An Alternative SYNC Detector
An alternative SYNC detector may estimate the noise floor (arising due to the WUR payload) by getting 𝑁 taps separated by 𝑇 𝜇𝑠 from the output of correlator (e.g., 𝑁=15, 𝑇=0.8 𝜇𝑠) and compare it with the correlator output
If the SYNC is detected, LDR and HDR can be differentiated by checking the sign of correlator output
e.g., WURx wakes up
Legacy
SYNC
Signal 𝜏
𝑁 1 ×𝑇 𝜇𝑠
𝑁 2 ×𝑇 𝜇𝑠
Correlate 𝜏
SYNC template … … 𝑦
Mean( ⋅ 2 )
x/y≷th 𝑥
⋅ 2

6. Simulation Results (HDR)
When the alternative SYNC detector (i.e., detector 2) is employed, the false alarm rate issue is remedied significantly
For HDR data, there is a trade-off between miss detection and false alarm rate as a function of the threshold
𝑁 1 = 𝑁 2 =15 and 𝑇=0.8 𝜇s

7. Simulation Results (LDR)
Since 𝑇=0.8 𝜇𝑠 and the ON duration for LDR (i.e, 4 𝜇𝑠) is larger than ON duration for SYNC field (i.e., 2 𝜇𝑠), the detector calculates 𝑦 with the correlated samples and the false alarm rate decreases as a function of the threshold for LDR data
𝑁 1 = 𝑁 2 =15 and 𝑇=0.8 𝜇s

8. Conclusion
In this study, we show that a SYNC detector which takes the noise floor into account can handle false alarm rate issues (i.e., detection of HDR and LDR data fields as SYNC field) without degrading the missed detection rate

9. References
[1] S. Shellhammer and B. Tian, “Concerns about Sync Detector False Alarms” IEEE /1201r0, Jul. 2018