1. Trang chủ
  2. » Thể loại khác

DSpace at VNU: Amplitude and phase adaptive nulling with a genetic algorithm for array antennas

4 116 0

Đang tải... (xem toàn văn)

THÔNG TIN TÀI LIỆU

Thông tin cơ bản

Định dạng
Số trang 4
Dung lượng 432,41 KB

Các công cụ chuyển đổi và chỉnh sửa cho tài liệu này

Nội dung

Amplitude and Phase Adaptive Nulling with a Genetic Algorithm for Array Antennas Le Quang Thao, Nguyen Ngoc Dinh, Dam Trung Thong Radio Physics Department, Physics Faculty Hanoi Univers

Trang 1

Amplitude and Phase Adaptive Nulling with a

Genetic Algorithm for Array Antennas

Le Quang Thao, Nguyen Ngoc Dinh, Dam Trung Thong

Radio Physics Department, Physics Faculty Hanoi University of Science, VNU

Hanoi, Vietnam thaolq@vnu.edu.vn Tel: 84.4.858.2254 Mobile: 84.983.712.941

Abstract - This paper researchs an efficient by using the genetic

algorithm apply in amplitude and phase to adaptive nulling for

phase array antennas A genetic algorithm adjusts some of the

least significant bits of the beam steering phase shifters and

amplitude weights in order to minimize the total output power

and place nulling in the jammes Using a few bits for nulling

speeds convergence of the algorithm and limits pattern

distortions Various results are presented to show the advantages

of this approach

Keywords: phase array antenna, genetic algorithm, adaptive

nulling

I INTRODUCTION

Almost radar system while the frequency spectrum

becomes more crowded, they become more vulnerable to

interference When the mainbeam gain times the desired signal

is less than the sidelobe gain times the interference signal, the

desired signal is overwhelmed by the interference A smart or

adaptive antenna is one alternative for recovering desirable

signals A smart antenna adapts its receive and/or transmit

pattern characteristics in order to improve the antenna’s

performance It may place a null in the direction of an

interference source or steer the mainbeam in the direction of a

desired signal At least two different antennas constitute a

smart antenna, they known as an antenna array The amplitude

and phase of the received signals are weighted and summed in

such a way as to meet some desired performance expectation

MIMO (multiple input /multiple output) communications

systems have adaptive transmit and receive antennas

Adaptive nulling complements the low sidelobe antenna

by placing nulls in a few low sidelobes to reject the strongest

interfering sources Almost adaptive algorithm possesses some

of these characteristics, but none of them meets all the

characteristics, as a result, they do not find the optimum

weights to reject the interference at hand Some common

adaptive algorithms include Least Mean Square Algorithm and

Howells-Applebaum adaptive processor, and examples can be

found in references [1] and [2] These methods are very fast but

the difficulties mentioned prohibit their widespread use,

particularly for arrays with more than a handful of elements

Another class of algorithms adjusts the phase shifter

settings in order to reduce the total output power from the array [3] and [4] These algorithms are cheap to implement because they use the existing array architecture without expensive additions, such as adjustable amplitude weights or more receivers Their drawbacks include slow convergence and possibly high pattern distortions This class has two approaches The fisrt approach to phase only nulling is to use a random search algorithm [5] Random search algorithms check

a small number of all possible phase settings in search of the minimum output power The search space for the current algorithm iteration can be narrowed around the regions of the best weights of the previous iteration A second approach forms an approximate numerical gradient and uses a steepest descent algorithm to find the minimum output power [6] This approach has been implemented experimentally but is slow and gets trapped in local minima As a result, the best phase settings to achieve appropriate nulls are usually not found This paper describes a simple technique suitable for minimizing the total output power and place nulling in interference source by making small adjustments amplitude and phase perturbations at each element of phase arrays antennas The approach combines a genetic algorithm with the hardware limitations of the array to place nulls in the directions of interference with small perturbations to the far field pattern Excellent nulling results are possible for most interference experiment

II THE ARRAY ANTENNA AND THE ADAPTIVE ALGORITHM

A Theory of the array antennas

A linear array antenna is a group of equally spaced antennas arranged along a line and whose outputs are added together to provide a single output

Figure 1 shows a diagram of such an array Mathematically, the array far field pattern is given by:

¦

=

Ψ

= N

n

j

ne n w AF

1

Where:

wn = anejĮn : amplitude complex weight at element n

Trang 2

Ȍn : phase due to element position and observation direction

N: number of elements in the array

Figure 1 Diagram of a phase array antennas

In order to minimize total output power, which consists of

the interference signal and possibly the desired signal, without

sacrificing the desired signal requires the antenna to place nulls

in the sidelobes and not the mainbeam It’s reasonable to

assume that the desired signal enters the mainbeam and the

interference enters the sidelobes Large phase shifts are

required to place a null in the mainbeam Large reductions in

the amplitude weights are required in order to reduce the

mainbeam Consequently, if only small amplitude and phase

perturbations are allowed, then a null can’t be placed in the

mainbeam but can be placed in the sidelobes Lower sidelobes

require smaller perturbations to the weights in order to place

the null Using only a few least significant bits (LSBs) of the

digital phase shifter and attenuator bits prevents the GA from

placing nulls in the mainbeam

B Theory of the adaptive algorithm

Amplitude and phase adaptive nulling algorithm modifies

the quantized phase and amlitude weights based on the total

output power of the array If no interference is present, then

the algorithm tries to minimize the desired signal To prevent

desired signal degradation, the algorithm should only be turned

on when the desired signal becomes swamped by the

interference or the nulling phase shifts and amplitudes weights

should be small This potential problem is discussed in more

detail in the next section The disadvantages of the phase-only

algorithms make them unlikely candidates for use with

antenna arrays This section presents a method that is as fast

as the beam space algorithm, doesn’t easily get stuck in

local minima, and limits pattern distortion

The adaptive algorithm is based on a genetic algorithm

and uses a limited number of bits of the digital phase shifters

and a few least significant attenator bits in amplitude weights

A genetic algorithm is a computer program that finds an

optimum solution by simulating evolution in nature In this

application the phase shifter settings evolve until the antenna

pattern has nulls in the directions of jammers and amplitude

weights setting reduce the meanbeam A genetic algorithm was

chosen for this application because it is an efficient method

to perform a search of a very large, discrete space of phase

settings to achieve the minimum output power of the array

A genetic algorithm (GA) offers an alternative to traditional

local search algorithms It is an optimization algorithm inspired

by the well known biological processes of genetics and

evolution Genetics is the study of the inheritance and variation

of biological traits Manipulation of the forces behind genetics

is found in breeding animals and genetic engineering Evolution is closely intertwined with genetics It results in genetic changes through natural selection, genetic drift, mutation, and migration Genetics and evolution result in a population that is adapted to succeed in its environment In other words, the population is optimized for its environment

Figure 2 Genetic algorithm flowchart

The algorithm has the following steps as showed in figure 2

1 Create an initial population

2 Evaluate the fitness of each population member

3 Invoke natural selection

4 Select population members for mating

5 Generate offspring

6 Mutate selected members of the population

7 Terminate run or go to step 2

III EXPERIMENT RESULTS

We consider with a 20 element adaptive linear array with elements spaced half a wavelength apart Assume the amplitude and phase weights have 6 bits quantization If the only source enters the mainbeam, then the adaptive algorithm tries to reduce the mainbeam in order to reduce the total output power Figure 3 shows the mainbeam reduction when 0-4 least significant bits (LSBs) in the amplitude weights are used for nulling Zero bits corresponds to the quiescent pattern A maximum reduction of 1 dB is possible using 4 bits of amplitude

21 28 35

Theta (degree)

4 3 2 1 0

Figure 3 Mainbeam reduction depend on amplitude weight

Trang 3

In contrast, Figure 4 shows the mainbeam reduction when

0-4 LSBs least in the phase weights are used for nulling Using

1-3 LSBs results in very little perturbation to the mainbeam

Unlike amplitude-only nulling, phase-only nulling causes beam

squint Also, 4 bits of phase had more effects on the mainbeam

than did 4 bits of amplitude This experiment demonstrates that

adaptive nulling with the LSBs would not cause significant

degradation to the mainbeam

21

28

35

Theta (degree)

4 3 2 1 0

Figure 4 Mainbeam reduction depend on phase weight

We consider the array has 6 bits amplitude and phase

weights Two LSBs of the amplitude weights and three of the

phase weights are used for nulling [7] show in Figure 5 The

desired signal is incident on the peak of the mainbeam and is

normalized to 14 or 15 dB One 0dB jammers enter the

sidelobes at -300

Figure 5 The least signifi cant bits of the amplitude and phase weights

The applied GA parameters are a population size of 20, a

chromosome length of 18 bits and a 50% selection rate, uses

roulette wheel selection and uniform crossover, and has a

mutation rate of 10% This is summarized in Table 1

TABLE I DESIGN PARAMETER FOR GA OPTIMIZATION

Item Definition Value

Subarray row number N 20

Population size M 20

Gene size P=N-1 3

Gene bits Q 6

Choromosome bits L=PxQ 18

Item Definition Value

Crossover probability Pc 50%

Mutation probability Pm 10%

Convergence of the algorithm is shown in Figure 6 Generation 1 represents the received signals from the quiescent pattern Generation 2 is the best of the initial random population of the GA Performance levels off in about 9 generations

0 30

Generation

Figure 6 Genetic algorithm convergence

Sometimes decreasing the null at one jammer location is done at the expense of the null at the other jammer location The total power output decreases monotonically, but the reduction in an individual jammer’s contribution to the total output power can go up or down The output power due to the desired signal remains relatively constant, because the main-beam remains virtually unperturbed Nulls appear in the array factor at the angles of the jammers as shown in Figure 7 The nulls come at a cost of increased average sidelobe level

2 4 6 8 10 12 14 16

Theta (degree)

Initial pattern Optimized pattern -30 degree

Figure 7 Null are adaptively placed at -30 0

The next example has the same array with two 0dB jammers enter the sidelobes at -200 and 200 Nulls appear in the array factor at the angles of the two jammers as shown in Figure 8

Trang 4

-1000 -80 -60 -40 -20 0 20 40 60 80 100

5

10

Theta (degree)

Initial pattern Optimized pattern

Figure 8 Null are adaptively placed at -20 0 and 20 0

IV CONCLUSIONS

The genetic algorithm performed well for one and two

jammers that were :

Separated by an angular width of at least one sidelobe and

were not symmetric in angular distance about the main beam

Both the one or two jammers nulling showed fast convergence,

deep nulling capability, and small pattern distortions Using

only a few least significant bits and small phase values and

amplitude weights for the MSB are keys to the algorithm’s

performance Its important advantage is that it is easy to

implement on existing phased arrays

Amplitude and phase adaptive nulling works on existing phased-array antenna designs, unlike signal processing based adaptive arrays that require receivers at every element Its main disadvantage is slow convergence time The genetic algorithm approach amplitude and phase adaptive nulling is significantly faster than the previous approaches of random search and gradient methods Thus, it advances the state of amplitude and phase adaptive nulling

REFERENCES

[1] R A Monzingo and T W Miller, “Introduction to Adaptive Antennas”, New York: Wiley,1980

[2] R T Compton, “Adaptive Antennas Concepts and Performance”, Englewood Cliffs, NJ: Prentice Hall, 1988

[3] C A Baird and G G Rassweiler, “Adaptive sidelobe nulling using digitally controlled phase-shifters”, IEEE AP Trans 24(5):638–649 (Sept 1976)

[4] H Steyskal, “Simple method for pattern nulling by phase perturbation”, IEEE AP-S Trans 31(1):163–166 (Jan 1983)

[5] R L Haupt, “Phase-only adaptive nulling with genetic algorithms”, IEEE AP-S Trans 45(5):1009–1015 (June 1997)

[6] R L Haupt, “Adaptive nulling in monopulse antennas”, IEEE AP-S Trans 36(2):202–208 (Feb 1988)

[7] R L Haupt and Douglas H Werner, “Genetic Algorithms in electromagnetics”, pp 137–141.I A John Wiley & Sons, Inc., Publication 2007.

Ngày đăng: 16/12/2017, 10:37