You are on page 1of 6

Research on Mutual Interference Mitigation for

Automotive Radar Systems


YU Xuyao
2018011213020

1 Background
1.1 Wide Application of Advanced Driver Assistance System (ADAS)
Assisted driving systems are becoming embedded in more vehicles as the technology base
matures (as shown in Figure ??). These systems promise greater safety, reduced congestion
and greater situation awareness for drivers.

1.2 New Extension of Spectrum


The United States Federal Communications Commission has expanded the spectrum
available for vehicular radars, to include the entire 76-81 GHz band.

1.3 The Presence of Mutual Interference


Radar is inherently a wave sensing measurement and subject to constructive and destruc-
tive combinations of radar transmissions. Therefore, interference is expected in environments
dense with reflective targets, as well as environments with similar transmitters.

1.4 Typical Radar Parameters

Figure 1: Surrounding field monitoring technologies for driver assistance systems

1
Figure 2: Surrounding field monitoring technologies for driver assistance systems

2
2 Source of Mutual Interference (Mutual Interference Mod-
eling)
The operation of multiple active systems can result in an environment where each sensor
is subject to energy emitted by other transmitters, as well as its own. This situation results
in mutual interference for the sensing systems, and degrades their performance[1].

2.1 Interference Power Model[1]


The Interference model is to reveal the probability of intercept, (POI) of the victim
radar in the frequency spectrum and time, and the magnitude of the energy of the interference
in various scenarios.
The spectral POI for a pair of radar is based on the amount of the available band they
occupy, or channel fraction, CF That is, a 200 MHz system, operating in the 76-to 77 GHz
band, has a channel fraction of 0.2. The general equation for spectral POI in a population of
K radars, ωK , is shown below.


K−1
ωK = 1 − (1 − CFk )
k=1

Similarly, the temporal POI for a pair of radar is based on the amount of the available
time they transmit, or duty factor, DF . The general equation for spectral POI in a population
of K radars, TK is shown below.


K−1
TK = 1 − (1 − DFk )
k=1

The total system overlap ξK , is the product of the temporal and spectral POI.

ξK = ωK TK

The product of these, ξ2 , is the pairwise overlap used in the interference model.
The method begins by modeling the interfering transmitters at-large as a point process, a
way of modeling a random set. There are several varieties of point processes, but the author in
[1] will use a simple and natural process, the Poisson point process. It has a single parameter,
λ, called the intensity, which is the average density of points in an interval. In our context,
modeling interferers with a Poisson point process, Θ, with intensity λ = 1/25 m would mean
that there is, on average, an interfering car every 25 meters along a road.
From here, the quantity of interest is the cumulated interference I at a point y from all
interfering nodes in the Poisson point process Θ,

I(y) = Px hx l (|y − x|2 )
x∈Θ

Here, Px denotes the power received from the transmitter at x, hx denotes the (random) power
fading coefficient, and l denotes the path loss, which is assumed only to be a function of the
distance between y and x. We will assume that E [hx ] = 1. Next, because we are using a

3
Figure 3: Surrounding field monitoring technologies for driver assistance systems

Figure 4: Pulsed Radars as Victim

Poisson point process, we can use Campbell’s Theorem to calculate the mean of I(y) with
respect to the point process:
[ ] ∫ ∫

EΘ [I(y)] = EΘ Px hx l (|y − x|2 ) = λE [hx ] Px l (|y − x|2 ) dx = λ Px l (|y − x|2 ) dx
x∈Θ

From here, we need only adjust Px and l to each scenario’s geometry to compute the inter-
ference power.

2.2 Perspective of Scenario

2.3 Perspective of Waveform


There is no need for us to discuss all the combination Interference between Pulsed, CS
and FMCW Radars, since interference effects between CS and FMCW radars are shown to
be equivalent[4].

2.3.1 Interference Between Pulsed and FMCW Radars[5]

2.3.2 Pulsed Radars as Victim

Pulsed-radar output for partially in-band FMCW interference with a magnitude 8 dB


smaller than target return. The interference appears as a slowly varying change in the noise

4
Figure 5: FMCW Radars as Victim

Figure 6: FMCW Radars as Victim

floor

2.3.3 FMCW Radars as Victim

Impact on the FMCW radar will be similar to the transient interference. The magnitude
of the interfering signal will be much larger, but its duration will be even shorter. Such inter-
ference is easily eliminated by either clipping or masking out the offending pulse. However,
we need some robust signal interpolation methods to better reconstruct the IF signal.

2.3.4 Interference Between FMCW Radars[4]

Interference between FMCW, CS and CW radar.


Interference that generates ghost targets. (a) CS radar case. (b) FMCW radar case.
Interference with different chirp slope but identical chirp duration
Interference with different chirp slope and duration but identical bandwidth. The upper
part shows the instantaneous frequency of the dechirped beat signal.

5
Figure 7: FMCW Radars as Victim

Figure 8: FMCW Radars as Victim

Figure 9: FMCW Radars as Victim

You might also like