RFbeam Microwave GmbH

RSP1 Radar Processor

Datasheet

6 Detection Algorithm

Detection algorithms are based on the complex FFT (Fast Fourier Transform) of the I and Q analog

signal inputs.FFT output logarithmic in order to get good signal processing conditions for both large and

small signals.

FFT represents in fact many narrowband filters that reduce noise amplitude. RSP1 uses 256 point FFT

resulting in 128 bins (filters) for each forward and backward movements.

This kind of detection results in a much better sensitivity than simple comparator solutions.

Approximative gain in S/N ratio by using a 256pt FFT is 10 * log(128) = 21dB. In reality, more than

double detection range can be reached compared to comparator solution.

Signals and processing behaviour may be explored by the RFbeam tool RSP1_Scope,

connected at the serial debug port (see 7.5 Serial Debug Interface)

6.1 Adaptive noise detection

The advanced noise detection technique leads to the outstanding sensitivity of the RSP1 solution.

Noise is measured separately for each frequency represented by the FFT results. Two stages in noise

measurement exist:

1. After power-on, an initial noise curve is built by measuring the mean of each FFT frequency bin.

Number of means (measuring time) can be selected by Parameter S04.

2. Adaptive mean is continuously built during operation. Adaptation time constant ca nbe selected by

Parameter S0C

If using I/Q stereo sensors (like K-LC2, K-LC5 etc.), noise cancellation is very efficient, because

movement Doppler signals can be efficiently distinguished from noise signals.

Trigger level results from the sum of adaptive noise

threshold (Grey plot) and the sensitivity selected by

parameter A02.

Signal in the center is the DC offset caused by the

amplifier and ADC converter.

Noise and small signals looks very strong in logarithmic form of the FFT magnitude. Remember

that a signal difference of 1mVrms to 2mVrms produces same logarithmic magnitude difference

as a difference from 500mVrms to 1Vrms.

© RFbeam Microwave GmbH www.rfbeam.ch

December 11, 2015

7/25

Fig. 6: Signal levels used for object detection. Example: forward moving object

FFT

Trig level

I/Q signal

interference

interference

noise

mean

noise level

min.trigger margin (param S0A)

sensitivity (param A01)

Trigger threshold

signal level

backward

forward