International Journal

on Marine Navigation

and Safety of Sea Transportation

Volume 5

Number 3

September 2011

325

1 INTRODUCTION

There are two basic kinds of interferences which de-

crease a performance of FMCW radar. The first

kind are noise and distortion of the own radar, se-

cond one are interferences coming from nearby

buildings, sea clutter and other electronic equipment

working in the same frequency band. There are some

ways to couple with such a problem. Most of them

are based on digital signal processing algorithms

implemented in DSP processor code. Developing of

CFAR (Constant False Alarm Ratio) detector looks

as the best way of suppressing analog interferences.

The radar works in X band. In DPU (Digital Pro-

cessing Unit) 8192 point real FFT was applied as

well as binary integration, correlation and a few

kinds of CFAR including clutter-map. The antenna

and transceiver of FMCW radar are shown in the

figure 1.

Figure 1. FMCW radar

2 INTERNAL INTERFERENCES

Although manufacturers of radars try to achieve per-

fect spectrum performance, there is always small

leakage of interferences coming from the transmitter

and the receiver. Some unwanted products of digital

synthesis and power converters appear in the IF (In-

termediate Frequency) signal. Although the signal

sampled in DPU (Digital Processing Unit) is band

limited, some of the mentioned products remain in

data vector after FFT (Fast Fourier Transform) pro-

cessing. Having a level greater than background

noises, they are detected and displayed as artificial

echoes. Because interferences appear in the fixed

Impact of Internal and External Interferences

on the Performance of a FMCW Radar

P. Paprocki

Telecommunications Research Institute, Gdansk, Poland

ABSTRACT: Although FMCW technology has become by now mature, there are still some unsolved prob-

lems left. First group of them are noise and distortion from own transmitter or receiver. They are usually

caused by DDS (Direct Digital Synthesis) and power converters performance. They appear on radar display as

circles or arcs because of their isotropic properties. Second group are interferences from external world like

beams of pulsed radars working in the same frequency band, reflections from close buildings and installations

or sea clutter. Apart from creating artificial echoes of target they can cause degradation of radar tracking sys-

tem. This paper evaluates influence of internal and external sources of interferences on the FMCW radar per-

formance. DSP (Digital Signal Processing) algorithms which optimize detection of FMCW radar were pre-

sented. Means of suppressing analog interferences by digital techniques were shown as well.

326

frequency, they are seen as a circle or a part of the

circle on the radar display (Fig. 2). Such kind of arti-

fact can mislead radar operator and tracking system.

Echoes of ships which are close to the circle would

be probably undetected. Tracking of the object mov-

ing trough the interference area is frequently lost.

There are some simple ways of suppressing such in-

terferences. Firstly, we can increase the detection

level. Unfortunately, in this case we reduce the de-

tection probability of others echoes as well. It is also

possibility to change carrier frequency of transceiv-

er. Although the interference (circle) vanishes, an-

other one often appears nearby.

Figure 2. Circle-shaped interferences as a result of the own

transmitter noise and a radial interference from the pulsed radar

beam.

To analyze the problem from DSP point of view

we should consider how CFAR algorithm works [3].

Detector based on CFAR

CA-CFAR GO-CFAR SO-CFAR

Range cells in row

Σ Σ

Binary vision

after detection

SO(GO)

K X S

α /M

Comparator

Figure 3. CFAR algorithm concept

The range-cell under test is compared with its

neighborhood. If the value of the signal in the cell is

greater then the average value of the neighborhood,

detection occurs. Specific interference exists only in

a few range cells in a row. If we found the right cell

number, we could increase the CFAR level locally.

Even though real echoes in those range-cells are

suppressed as well, other range-cells are unaffected.

This simple approach suppresses the interference

just before detection. The results of applying such

approach is shown in figure 4. The circle disappears,

detection of nearby echoes is still possible.

Figure 4. Circle-shaped interferences suppressed digital algo-

rithm.

3 EXTERNAL INTERFERENCES

In order to limit the influence of other electronic de-

vices working in the same band of frequency

FMCW signal should be conditioned. The typical

example is a problem with beams of the pulsed ra-

dars which appear on the display as a radial row of

dots (Fig. 2). High level of such a radar signal

masks the nearby echoes. Data conditioning algo-

rithm limits the level of time-domain data to average

signal level in the sweep. It results in limiting un-

wanted products after FFT processing. Second

source of external interferences are reflections com-

ing from nearby buildings and sea clutter. Distin-

guishing sea clutter with real objects is often prob-

lematic. In any case some kind of filtering is needed

[1]. Possible solution is a filter based on clutter-map

implemented as a type of CFAR [2].

Figure 5. Gdynia harbor 0.25Nm range GO-CFAR

327

Komparator

-1

1-W

Binary Vision after

detection

Detection based on

Clatter map

n-1

α x y

n-1

(k)=(1-W) * y

n-1

(k) + W * z

(k)

Figure 6. Clutter map algorithm concept

Clutter-map estimate caries information about the

past and present values of particular range-angular

cell. The more up-to-date the data is, the greater con-

tribution in estimate it has. The state of the estimate

is continuous up-dated in RAM memory. The value

of the estimate can be written as:

(k)=(1-W) * y

n-1

(k) + W * Z

(k)

Figure 7. Clutter-map estimate of 32 sweeps in a row (Y-axis).

To store the estimate N x M x 2 bytes of RAM

are used (N-number of range cells, M- number of

sweeps per antenna rotation). For N=4096 and 3000

sweeps per rotation 24 MB of RAM is occupied only

by clutter-map estimate. The need to access to a

large and relatively slow external SDRAM, which

deteriorates processing speed is the main disad-

vantage of such kind of CFAR.

Detection probability is given by the following

expressions [4]:

∏

−+

])1(1[

SNR

∞→M

The W and α coefficients should be optimized in

order to achieve desired probability of detection for

maneuvering objects.

Figure 8. Probability of detection diagram.

W=0.023, K=4.

There are a few characteristic effects of clutter-

map algorithm. Any fixed echoes coming from

nearby buildings, boardwalks and buoys disappear.

In some applications radar echoes of such objects

are not needed. Moreover, they are still visible on

electronic chart. Sea clutter is suppressed by such

kind of detector as well. Figure 5 shows the radar

display of Gdynia harbor obtained using GO (Great-

est Of)-CFAR type. There are many items redundant

including unwanted echoes from nearby port facili-

ties and from the coast line. Results of the detection

based on clutter-map are shown in the figure 9. Only

maneuvering objects are still visible. Such kind of

data processing gives excellent detection perfor-

mance in case when a small object is moving along

boardwalks or quay. In the classical CFAR such ob-

ject would be overwhelmed by the strong echoes

from large fixed objects. Another advantage of clut-

ter-map in context of the tracking system, is a limita-

tion of data stream, because plots which are sent

from extractor are preliminary filtered.

Figure 9. Gdynia harbor 0.25Nm range CFAR based on clutter-

map.

328

4 CONCLUSIONS

In this paper an approach to improve FMCW radar

detection was presented. All measures of suppress-

ing internal and external interferences were based on

different kind of CFAR. It was shown how to elimi-

nate false, artificial echoes which may mislead the

operator or the tracking system. Some advantages of

detection based on clutter-map were considered as

well.

REFERENCES

[1] Keith D. Ward, Robert J.A. Tough and Simon Watts. “Sea

Clutter: Scattering, the K Distribution and Radar Perfor-

mance” . Institution pf Engineering and Technology, Lon-

don 2006

[2] Conte E. “Clutter-Map Detection for Range –Spread Tar-

gets un Non-Gaussian Clutter.” IEEE Transaction on aero-

space and electronic systems Vol. 33 April 1997

[3] Khalighi M.A. “Adaptive CFAR Processor For Nonhomo-

geneous Environments” IEEE Transaction on aerospace

and electronic systems Vol. 36 July 2000

[4] Nadav Levanon “Radar Principles” Wiley Interscience

Publication 1988