Dynamic Range Control Scheme Applied to GNSS Signal

Acquisition Using Evolutionary Concept

ABSTRACT This paper presents a novel application of dynamic range control method to the signal

acquisition in Global Navigation Satellite System (GNSS). Typical signal search methods conduct

operation of multiplication and addition with regards to each Doppler frequency and code delay trial

point step by step on the basis of two dimension (2D) signal search plane and finally employ

hypothesis test to observe the existence of signal. Though traditional acquisition method can adjust

frequency search step size to obtain more accurate parameters for the initial use of tracking loop, it

takes longer signal search time. The proposed method not only utilizes the evolutionary characteristics

to find the optimal signal parameters but also employs an adaptive scheme to control the search range

and avoid the evolution method converging to local optimal value so as to acquire the desired signal

rapidly. Simulations and experiment results show that the proposed method can improve the precision

of signal parameters and shorten signal acquisition time for coarse/acquire (C/A)-code, precision code

(P-code), and binary offset carrier (BOC) signals.

Keywords —GNSS receivers, Synchronization, Evolutionary algorithm

Subject index: EL2 (Communication Systems) EL7 (Signal Processing) EN1 (Aerospace Electronics)
Introduction

The positioning and time information provided by the global navigation satellite system (GNSS) has

been extensively applied to navigation, military and civil currently, among which the global

positioning system (GPS) developed in America from the early 1970’s has been utilized in military.

Each GPS satellite simultaneously transmits on two L-band frequencies, denoted by L1 and L2, which

are 1575.42 and 1227.60 MHz, respectively, and the PRN code modulates with the L-band frequency.

The PRN code consists of C/A (coarse acquisition) code on L1 carrier and P code on L1 and L2

carriers. The P-code is encrypted and the resulting code is termed P(Y)-code. The encrypted
P(Y)-code is only accessible to authorized users with cryptographic keys. In GNSS receiver design,

code and frequency synchronization plays a crucial role, including signal acquisition and signal

tracking. The signal acquisition process is a two dimensional search which lies in acquiring signal

parameters (coarse code delay and Doppler frequency) for the use of phase locked loop (PLL) and

delay locked loop (DLL) in the tracking loop (Parkinson and Spilker 1996). Conventional signal

search method (Braasch and Van Dierendonck 1999) proceeds with acquisition every 500 Hz on the

basis of specific frequency search range (generally ± 10KHz). Meanwhile, this method conducts the

search for coarse code delay. After two dimensional search (code search and frequency search domain),

the “pull in” process initiates in order to estimate more accurate signal parameters. At this stage,

coarse signal parameters have been acquired. Then, the estimated parameters are sent to the tracking

loop to proceed with signal synchronization. Though the method can effectively acquire signal

parameters, it also consumes large amounts of signal search time. In fact, such a GNSS signal

acquisition problem has been dealt with by many authors (Van Nee and Coenen 1991; Namgoong and

Meng 2001; Starzyk and Zhu 2001; Pang et al. 2003; Akopian 2005; Wilde et al. 2006). These so

called parallel code phase search methods compute the correlation from time domain to frequency

domain by utilizing a circular convolution in order to fast acquire C/A-code and P-code signals and

have been implemented in software define radio techniques (Brown et al. 2000). Recently, a reduced

fast Fourier Transform (FFT)-based limited code correlation and multi-C/A code acquisition method

have also been proposed to further improve the acquisition performance (Sagiraju et al. 2008; Jan and

Lin 2009).

Though the above methods can effectively reduce system complexity and acquire signal

parameters, these methods still need hypothesis of different search frequency to test. As result, the

number of frequency bin plays a crucial factor in signal acquisition process. Chang et al. (2008) have

proposed an adaptive frequency step size control method to solve this problem. This paper observes

that signal correlation value (search power) and its variation van reduce signal search times on 2D

plane using adaptive logic control method to shorten search time and the result also yields
high-precision acquisition parameters. Based on such a concept, the signal search power and its

variation serve as the parameter of adaptive adjustment logic in this paper along with evolution

method to search for signal parameter. In contrast to the signal acquisition of FFT-based methods, the

proposed method can shorten search speed and acquire more accurate parameters.

Evolution method has been extensively applied in nature science and artificial intelligence. This

concept originated from J. D. Bagley’s dissertation in 1967 (Bagley 1967). Afterwards, J. H. Holland

(1975) was influenced by this concept and proposed related application research to become the

precursor of genetic algorithm. This method has been proven to be an effective optimization method

and a robust search technique as well (De Jong 1975; Goldberg 1989; Fogel 1994). Within certain

frequency and code search range, several sets of initial signal parameters (code and frequency search

value) are generated in local replica. Each set of initial parameter consists of both code delay and

Doppler frequency. These signal parameters are coded to serve as the initial individuals (code delay

individual and Doppler frequency individual, respectively) in proposed method. Then, each set of

initial individual (including code delay and Doppler frequency individuals) processed through

correlation generates an output value, which serves as the fitness value of that set of initial individual.

Afterwards, the evolutionary process of selection, reproduction, crossover and mutation begins.

During the search process, once the fitness value exceeds the initial threshold (set at noise floor level),

the desired signal may fall within the vicinity of the corresponding parameters of that set of individual.

In next iteration, the signal search range is narrowed down, the sets of individual are decreased and the

threshold is increased. If the fitness value exceeds the maximum threshold, the result of acquisition is

successful. Otherwise, the acquisition result may simply contain noise, which calls for a repeated run

of evolution process. The adaptive scheme guides the solution to its ultimate global level by adapting

the search space boundaries of the signal parameters during the optimization process. Simulations and

experiment results reveal that this method not only saves signal search time but also finds more

accurate signal acquisition parameters for C/A-code, P-code, and binary offset carrier (BOC) signals.

Comparison results of the published methods are also demonstrated in this paper.
 The organization of this paper is as follows. Section “Problem formulation” describes the signal

acquisition process and the possible problems to be encountered. Section “Methodology” depicts how

to apply the proposed method to GNSS signal acquisition. Section “Acquisition result and

performance analysis” explains the simulation procedure of the proposed method and provides the

performance evaluation and comparison of published methods with proposed method. An example is

listed to analyze the feasibility of this method. Finally, Section “Conclusions” summarizes the key

points of this paper.

Problem Formulation

Signal acquisition process

The major purpose of this paper is to employ evolution method to enhance the performance of

GNSS signal acquisition. Under the scenario of ignoring data modulation and interference, the

incoming signal is down-converted and then transferred to digital Intermediate Frequency (IF) signal.

The signal at sampling time tk can be described as:

r (tk ) = PG (tk - τ )exp(− j 2π ( f c + f d ) tk +θ ) + w(tk ) , (1)

where P is the power of the direct line-of-sight signal, f c is the IF, and j = −1 . f d , θ , and τ

denote the Doppler frequency, carrier phase, and code delay, respectively. The noise components

w(tk ) are referred to as white Gaussian noise distribution, in which the power spectrum density is

N 0 / 2 . The N 0 stands for single sideband power spectrum density of noise. G (⋅) is the filtered code

sequence expressed by C/A-code, P-code or BOC signal (GPS OS SIS ICD 2000; Galileo OS SIS ICD

2008). The C/A-code has a rate of 1.023 Mchip/s with a period of 1023 chips (1ms) and the P-code

truncated period is one week with chipping rate of 10.23 Mchip/s. BOC modulated signals are

employed for the Galileo SIS and modernized GPS currently. The baseband signal r (tk ) multiplies a

locally generated replica and the result is processed through correlation operation. The resultant output

is expressed as:
 N k,c
1
Ik =
N k,c
∑ s(tk )G (tk − v) cos(2π( f c − u )tk )
tk = 0

N k,c
, (2)
1
Qk =
N k,c
∑ s(tk )G (tk − v) sin(2π( fc − u )tk )
tk = 0

where N k,c is number of samples per coherent accumulation segment, equal to Tc / Ts where Ts is

the sampling period and Tc is code chip rate. u and v are the guessed value of the Doppler

frequency and code delay, respectively, in frequency search range ⎡⎣ f c − umax,lo , f c + umax,up ⎤⎦ and code

search range ⎡⎣ vmax,lo , vmax,up ⎤⎦ . umax,up and umax,lo are the upper boundary and lower boundary of

Doppler frequency, respectively. Similarly, vmax,up and vmax,lo are the upper boundary and lower

boundary of code delay, respectively. G (⋅) is unfiltered local replica. The in-phase I k and

quadrature-phase Qk are squared respectively and then summed altogether. The result is accumulated

after K non-coherent integration time. The correlator output is shown as follows:

y (u , v)
K
= ∑ ( I k2 + Qk2 ) , (3)
k =1

K [( P N k,c / 2) R (ετ ,k )sinc(ε f,k Tc )cos(ε θ ,k ) + wI,k ]2

=∑
k =1 + [( P N k,c / 2) R(ετ ,k )sinc(ε f,k Tc )sin(εθ ,k ) + wQ,k ]2

where R(⋅) is the correlation function between the filtered incoming signal and local replica, and Tc

represents the coherent accumulation interval; ε f,k and ε θ ,k are the difference between incoming and

internally generated Doppler frequency estimate, and that between incoming and internally generated

carrier phase estimate, respectively. ετ ,k represents the estimated code delay error. Each of R (ετ ,k )

and sinc(ε f,k Tc ) has a maximum amplitude of 1 when ετ ,k and ε f,k take on the value of zero. If

these values are not zero, the result is a decrease in the amplitude of correlations. wI,k and wQ,k

denote the in-phase and quadrature-phase noise samples with variance σ 2 =N k,c N 0 / 2 , respectively.

The acquisition process is to find maximum y (u, v) through search on the two dimensional
(frequency and code domain) grid of trial points, frequency point u , and code point v . Two

hypothesis tests define the test statistic for signal detection (Zigangirov 2004; Torrieri 2004; Barket

2005). When the value y (u , v) exceeds the set threshold v, the signal is considered present

(hypothesis H1 ). Otherwise, the signal is absent (hypothesis H 0 ). Under hypothesis H1 , the

probability of signal detection PD is given by

∞
PD = ∫0 Pr(Ty > v , H1 ) p1 (Ty H1 )dTy
∞
, (4)
= ∫v exp(- y - ρ) I 0 (2 ρy )dy

where ρ = P / 2σ 2 , Ty is the test statistic, Pr(⋅) refers to probability, and I 0 (⋅) is the zero-order

modified Bessel function of the first kind. The false alarm rate of hypothesis H 0 to H1 is as shown

in (5)

∞
PFA = ∫0 Pr(Ty > v , H 0 ) p0 (Ty H 0 )dTy
∞ 1 y v , (5)
= ∫v 2σ 2
exp(- 2 )dy = exp(- 2 )
2σ 2σ

where p1 (⋅) and p0 (⋅) are non-central and central chi-square χ 2 distribution with two degrees of

freedom, respectively. Suppose the signal is acquired successfully, the Doppler frequency f̂ d and

code delay τ̂ can be obtained. The constant threshold v = σ 2ln( PFA

-1
) is often selected in order to

achieve a certain probability of false alarm in noise alone by employing the well-known
Neyman–Pearson criterion (Dillard 1974).
The above depiction shows that what concerns the designer is to rapidly find the required u and

v in a short time so as to have the correlation exceed the set threshold. The following chapters

demonstrate that the proposed acquisition method can effectively speed up signal acquisition speed

and maintain high estimated parameter precision.

Methodology
Standard GA acquisition procedure

The acquisition time and the precision of signal parameters lie in the signal search range and each

search step size. That is, the guessed number of trial search point each time determines the signal

search time. Though smaller step size increases the precision of estimated signal parameter, it also

increases search times, which causes longer acquisition time. On the contrary, the larger the step size,

the more the loss of correlation. In the process of GA, the search range (including frequency and code

range), and the number of acquisition trial points are adjusted to shorten signal search time. The

adaptive scheme is to enhance the accuracy of signal parameters. In the following, the whole process

of adaptive GA method applied to signal acquisition is illustrated.

1. Doppler frequency and code delay encoding

The first step in employing GA to conduct signal acquisition is to convert the locally generated

Doppler frequency and code delay to string by way of binary encoding method (Davis 1991). The

respective upper bound of Doppler frequency and code delay ( umax,up and vmax,up ) is converted to

string and the length is expressed as m and n , respectively.

encoding
umax,up ⎯⎯⎯⎯ → Sf : a m a m −1 ...a 2 a1
encoding
a , b ∈ {0,1} , (6)
vmax,up ⎯⎯⎯⎯ → Sg : b n b n −1 ...b 2 b1

where Sf and Sg are the Doppler frequency and code delay individuals, respectively. m and n are

the maximum bit number of Doppler frequency and code delay individuals. Each bit in the individual

is termed as genetic gene.

2. Fitness function determination

After the parameters are encoded, it is necessary to define the fitness function to evaluate the

quality of each individual. To put it simply, fitness function is the performance index of GA, which

determines the quality of individual, retains the superior gene and gets rid of inferior gene to gradually

upgrade the overall performance index through the evolution by generation. During the signal
acquisition process, the signal detection margin serves as fitness function, which is defined by
E[ y (uz , vz )]
Qz = 10 ⋅ log10 { }, z = 1, 2,3,...,Z . , (7)
E[ wk ]

where wk = wI,k + j wQ,k is the noise term as depicted in (3). E[⋅] is the expectation value, and Z is

the initial population number depicted in step 3. The output y (uz , vz ) can be obtained through

correlation of coherent integration and non-coherent integration. The goal of GA is to find the optimal

uz and vz parameters in order to maximize Qz value.

3. Initial population size selection

In step 1, the search parameters have been encoded as two individuals, which are one set of

solution in signal acquisition result. Before utilizing GA, Z sets of parameters must be produced

randomly, which represents initial population indicated by

U = {(u1 , v1 ), (u2 , v2 ), ..., (ui , vi ),..., (uZ −1 , vZ −1 ), (uZ , vZ )} , (8)

The evolution of initial population is adopted to obtain optimal solution. The number of initial

population depends on the complexity of search problem. The solution derived from the adoption of

initial population may be poor in performance. However, through the evolution of GA, the subsequent

generation outperforms the previous one and finally achieves the optimal value.

4. Selection and reproduction

Similar to the cell division of organism, the probability of reproduction is based on the rule of

survival of the fitness. Selection is the process that prefers individuals with high fitness over low-fitted

ones. Thus, the probability of choosing certain individual is in proportion to its fitness. Suppose the

fitness value of i-th individual (ui , vi ) is Qi , the probability of reproduction is represented by

Z
Δ i = Qi / ∑ Qi . (9)
i =1

The most prevalent natural selection method of reproduction is roulette wheel game, the slot

division of which is proportional to the fitness value of individual. The high-fitted ones stand a higher

chance to be selected in roulette game.

5. Crossover

Biologically, crossover is the exchange of genes between the chromosomes of two parents.

One-point crossover is a method for GA that operates on binary strings. Here, crossover contributes

to the information exchange between individuals. The site of point-to-point crossover is random.

The individual after crossover consists of part of genes from each of parental individuals. Thus, this

new individual is distinct from its previous one. The purpose of crossover is to combine superior

genes from different individuals in order to generate new individuals with higher performance index.

However, not every chosen new individual has to go through crossover. Suppose in (6), the

exchange point between the i-th Doppler frequency individual Sif , code delay individual Sig and

the next set Sif+1 , and Sig+1 is k, correspondingly. The new individuals after crossover are illustrated

in (10) and (11):

Sif : a im a im −1 ... a ik ...a i2 a1i Sif′ : a im a im −1 ... a ik+1 ...a i2+1a1i +1

Si+1 i +1 i +1 i +1 i +1 i +1
f : a m a m −1 ...a k ...a 2 a1 S(fi+1)′ : a im+1a im+−11 ...a ik ...a i2 a1i
(10)

Sig : bin bin −1 ... bik ...bi2 b1i Sig′ : bin bin −1 ... bik+1 ...bi2+1b1i +1
Si+1 i +1 i +1 i +1 i +1 i +1
g : b n b n −1 ...b k ...b 2 b1
S(gi+1)′ : bin+1bin+−11 ...bik ...bi2 b1i
(11)

6. Mutation

In biological system, besides natural selection, organism goes through mutation to adapt itself to

surroundings for survival. The above-depicted GA merely maintains original individuals and combines

superior genes from different individuals with high fitness. However, no matter how many times the

reproduction and crossover are, the GA does not produce characteristics that are inherently absent in

individuals. The function of mutation lies in producing characteristics which are originally absent in

high-fitted individuals. The k -th bit of i-th Doppler frequency individual Sif and code delay

individual Sig is changed from a ik Æ a ki (1Æ0), shown as follows:

 Sif : a im a im −1 ... a ik ...a i2 a1i Sif′ : a im a im −1 ... a ki ...a i2 a1i
Sig : bin bin −1 ... bik ...bi2 b1i Sig′ : bin bin −1 ... bki ...bi2 b1i
(12)

Though the above method can acquire the optimal value, large population size is required during

search process, which often leads to longer computation time. To reduce search time, the population

size and search range boundary are automatically adjusted during each iteration process to speed up

search time. The following is the process of adaptive GA.

GA with adaptive scheme

In each iteration, fixed and large numbers of initial population result in longer signal search time.

As a result, the initial detection of the presence of desired signal (the signal parameters have not been

accurately acquired yet) allows us to decrease the number of initial population in next search. The

( p + 1) − th iteration of initial population number is as follows:

( p +1) ( p)
Z p +1 = Z p / (Qmax / Qmax ), (13)

( p)
where Qmax ∈ max{Q1( p ) , Q2( p ) , Q3( p ) ,...Qz( p ) } . When the Qmax
( p)
exceeds noise floor, GA initiates two

( p)
procedures. The first procedure is that when Qmax exceeds the set detection threshold v (depicted in

section “Problem Formulation”), it indicates the desired signal has been acquired and the GA process

( p)
terminates. The second procedure is when the Qmax as between detection threshold and initial

detection threshold, the adaptive scheme begins to control the number of initial population. The initial

detection margin is set at 1.4 Q and the definition of Q is the same as that in (7). The only

difference is that Q is the fitness value acquired under the scenario of no desired signal. Meanwhile,

the adaptive scheme also controls and adjusts the search range of Doppler frequency and code delay,

which is depicted in the following:

( p)
f c − umax,lo ≤ u ( p +1) ≤ f c + umax,up
( p)
, (14)
 ( p)
vmax,lo ≤ v p +1 ≤ vmax,up
( p)
, (15)

where

⎧⎪ f c + fˆd( p ) + α ( umax,up
( p) ( p)
− f c / 2), umax,up (0)
< umax,up
( p +1)
u max,up = ⎨ (0) , (16)
⎪⎩ umax,up , otherwise

⎧⎪ f + fˆ ( p ) − α ( umax,lo
( p) ( p)
− f c / 2), umax,lo (0)
>umax,lo
( p +1)
umax,lo = ⎨ c (0) d , (17)
u
⎪⎩ max,lo , otherwise

⎪⎧τˆ + β (vmax,up / 2),

( p) ( p) ( p) (0)
( p +1)
vmax,up < vmax,up
v max,up = ⎨ (0) , (18)
⎪⎩ vmax,up , otherwise

( p +1)
⎧⎪τˆ ( p ) − β (vmax,lo
( p) ( p)
/ 2), vmax,lo (0)
> vmax,lo
v max,lo = ⎨ (0) , (19)
⎪⎩ vmax,lo , otherwise

where α and β are reduced factors, which mainly control the convergence rate during signal

acquisition process.

The adaptive scheme adjusts the Doppler frequency and code delay boundaries with the increase of

acquisition iteration. Thus, this adjustment results in faster convergence and more accurate parameter

values.

Termination criterion

The termination criterion of GA normally regulates reproduction generation or detects the

difference between generations. If there is no sign of evolution after several generations, the evolution

has terminated, which indicates the convergence to optimal value. The Err in (20) approximates zero,

which means the GA process comes to end.

( M +1) (M )
Err = Qmax − Qmax , (20)

where M is the final iteration number after GA converge. When the GA process finishes, the chosen

optimal individual is decoded so as to obtain the Doppler frequency and code delay of desired signal.
Acquisition Results and Performance Analysis

This section compares the proposed GA method with the other published methods in acquiring

GNSS signals (C/A-code, P-code, and BOC) under the same simulation parameter. Three types of

signal are presented for signal acquisition analysis under no consideration for mutual jamming

between signals. Suppose signals are down converted to 20MHz and the sampling frequency is 64

MHz, which fulfills Nyquist’s sampling requirement due to the fact that P-code has a bandwidth of

20.46 MHz. To verify the proposed method as capable of acquiring GNSS signals, the signal search

power margin (SPM) is adopted to evaluate the performance of different methods and is calculated as

follows:

Cp
SPM = 10 ⋅ log10 ( ), (21)
E[Cn ]

where Cp is the highest peak of search power during correlation process. E[Cn ] depicts the mean

power of noise. The proposed method has evaluated by using a developed program under MATLAB

environment.

Simulation of proposed method

The process of applying GA with adaptive scheme to signal acquisition is shown in Fig. 1. For

each parameter value, all individuals of the current population are evaluated and a local optimal value

is selected. This operation is repeated as many times as required with a diversified population. This

diversity is assured by the application of GA operators in binary coding. Once a set of local optimal

value is obtained, a global one is selected. Similarly, this process is carried out for all search values to

produce the solution per iteration. At this stage, if the acquisition process is not terminated, the

maximum search space boundaries (including Doppler frequency and code delay) are adjusted by

subjecting them to an evolution cadence (variation of fitness function). This process is repeatedly

carried out in this manner as long as it has not entered into an offspring stagnation; otherwise, the
program is interrupted by a fixed maximum iteration number. The initial population number of

C/A-code, P-code, and BOC is set at 1500, 7000, and 5750, respectively, with C/No=45 dB-Hz,

crossover probability 0.85 , mutation probability 0.008 and false alarm rate PFA = 0.01 in GA

process. The selection criteria of initial population number is calculated as follows:

Z = 2μ × (chip length) , (22)

where μ is a scale factor and the value between 0.8 and 1. The chip length of C/A-code and BOC is

1023 and 4096, respectively. It assumes a segment of the P-code for several ms or 0.5s and generates

the population size.

A large number of initial population can speed up signal search time, but is also takes higher hardware

cost. The initial population number, crossover, and mutation probability value are set based on

experience rule to be successfully utilized in GA to acquire desired signals. Figs. 2–3 show the

examples for the evolution values of two parameters (Doppler frequency and code delay) for C/A-code.

The protruding “○” in Fig. 3 is caused by mutation in GA. The two figures show that in utilizing

adaptive scheme, the signal search range will gradually narrow down its range after each iteration to

ensure the estimated signal parameter to converge to global optimal value and also reduce the iteration

times. On the contrary, it takes longer acquisition time without using scheme.

The proposed method is compared with other existing methods (Borre et al. 2007; Chang 2008)

with regards to performance of signal acquisition. Figs. 4−6 illustrate the application of various signal

acquisition methods to acquire C/A-code, P-code, and BOC signals, respectively, with various C/No

values, the result of which is each different calculated maximum search power margin. Such an

illustration is to focus on the relation between known C/No and calculated SPM. Fig. 4 shows that the

proposed method has higher SPM as opposed to other methods. With the increase of C/No, the

difference of SPM between each method becomes smaller. Fig. 5 depicts that under low C/No, the

proposed method has higher SPM. With the increase of C/No, the curve of SPM approximates stable

as in Fig. 4. Fig. 6 demonstrates that with the increase of C/No, the difference of SPM between each
method becomes larger, which is caused by the fact that serial and parallel frequency search methods

are under greater influence by noise regarding the acquisition of BOC. Figs. 4−6 reveal that when the

curve od SPM approximates stable (C/No as 50 dB-Hz) and also under the same integration time,

P-code is under greater influence by noise in correlation process with regards to its signal acquisition.

As a result, the SPM of P-code is lower than that of other signal types in correlation process.

Fig. 7 shows the detection probability of the C/A-code, P-code, and BOC signals, respectively,

assuming that neither Doppler frequency nor code delay error is present, for a coherent integration

time of 1 ms and for 2ms non-coherent summations. The search region is often very large owing to the

long period and high chip rate of the P-code (This paper is limiting the searching segment to 2ms);

hence, serial search method which tests one resolution cell at a time is too slow to search the P-code.

Fig. 8 depicts the relation between C/No and maximum number of initial population (0-th generation

of GA) under the signal detection probability as below 90%. The figure indicates that the number of

initial population tends to decrease with the increase of C/No. With respect to BOC signal acquisition,

the decrease speed of the number of initial population is slower than that of other two types of signal.

On the other hand, Fig. 9 analyzes the relation between the number of initial population and search

times regarding C/A-code. The result demonstrates that when the number of initial population adopts

roughly 1.5 times code period length of C/A-code (1023 chip), the signal can be accurately acquired

within fewer search times under C/No= 45 dB-Hz. When the number of initial population is below 0.5

times code period length, the proposed method can not acquire the signal. Such a result is the same

with P-code and BOC. Figs. 10−12 illustrate the zoom-in signal acquisition results of C/A-code,

P-code, and BOC, respectively, and these results can show the GA evolutionary process.

In addition, the system computation time of adaptive GA depends on the maximum initial

population number Z and iteration number M , which takes O( ZM ) . For example, regarding the

hardware implementation, the standard GA method takes 2× 1500 plus 1500 multiplications and

2× 1500 additions for C/A-code. Therefore, the total required number of operations equals 4,500

multiplications and 3,000 additions. The adoption of adaptive GA takes additional memory for the
adaptive adjustment. Though adaptive GA is higher in hardware cost than serial method, the former

outperforms the latter in the improvement of signal acquisition performance. Table 1 compares the

simulated performances of our proposed method with those of the existing methods. The execution

time is measured using the tic and toc functions in MATLAB. An average personal computer (PC) is

adopted for 20 execution time measurements, with the mean computed as well (Borre et al. 2007). The

table indicates that the proposed method is superior in time and more precise in parameter than other

methods in acquiring three different types of signal. It is noteworthy that because P-code has longer

period time, code length of only that segment (about 2ms) is utilized for signal acquisition during

simulation. Hence, its acquisition time is the longest among different signal types. Serial search

method is superior in its easy implementation of hardware, but is inferior in parameter precision than

other methods. Parallel code phase search method is higher in parameter precision rapidly, but is also

higher in hardware cost. Though parallel frequency search method is rapid in acquisition time, but is

limited by the length of FFT ans sampling rate. Thus, this method can not be efficiently promoted and

is seldom applied to realistic circuit. Though Chang’s method is low in hardware implementation

complexity, high in estimated parameter precision and parallels the proposed method, the GA with

adaptive scheme is more rapid and efficient in acquisition time. Despite the fact that the proposed

method is somewhat higher in hardware implementation complexity (count on number of initial

population), it has potential to be implemented in hardware under the gradual and continuous

enhancement of integration circuit design technique.

Conclusions

In this paper, a novel GA with adaptive scheme is proposed, verified, and applied to GPS signal

acquisition through simulation and experiment. This paper utilizes the evolutionary characteristics of

GA in the acquisition of GNSS signals. It combines an adaptive scheme that guides and controls the

signal parameter boundaries and population number. Indeed, this proposed method can reduce the

acquisition time and yield to more accurate signal parameters using larger number of initial population.
Though the hardware implementation complexity and extra memory of this system increase in

acquiring P-code and BOC signals, it is shown form the result that it indeed performs well in the

enhancement of signal acquisition performance. In the future, the goal is oriented towards the

reduction of hardware implementation complexity to meet the demand of low cost.

Acknowledgements

References

Akopian D (2005) Fast FFT based GPS satellite acquisition methods. IEE Proc Radar Sonar Navig
152(4):277–286
Bagley JD (1967) The behavior of adaptive systems which employ genetic and correlative algorithms.
PhD thesis, University of Michigan, Ann Arbor
Barket M (2005) Signal detection and estimation, 2nd ed., Artech House, Massachusetts.
ISBN:1580530702
Borre K, Akos DM, Bertelsen N, Rinder P, Jensen SH (2007) A software-defined GPS and Galileo
receiver: A signal-frequency approach, Birkhauser Boston. New York. ISBN: 978-0-8176-4390-4
Braasch MS, Van Dierendonck AJ (1999) GPS receiver architectures and measurements. Proc IEEE
87(1):48–64
Brown A, May M, Tanju B (2000) Benefits of Software GPS Receivers for Enhanced Signal
Processing. GPS Solut 4(1): 56–66 (DOI: 10.1007/PL00012829)
Chang CL (2008) Using Fuzzy Logic Controller with Adaptive Detection Scheme for Fast Acquisition
of Satellite Navigation Signals. J. Chin. Inst. Eng. Spec. Issue Intel. Sens., Actu., and Control,
33(3), pp. 367−378, April 2010.
Davis L (1991) Handbook of genetic algorithms. New York: Van Nostrand Reinhold
De Jong KA (1975) An analysis of the behavior of a class of genetic adaptive systems, Ph.D. thesis,
University of Michigan, Ann Arbor
Dillard GM (1974) Mean-level detection of nonfluctuating signals. IEEE Trans. Aerospace Electron
Syst, AES-10(6):795–799
European Space Agency/Galileo Joint Undertaking (2008) Galileo open service signal in space
interface control document. Draft GAL OS SIS ICD/D.1, Feb.
Fogel DB (1994) An introduction to simulated evolutionary optimization IEEE Trans Neural
Networks 5(1):3–14
Goldberg DE (1989) Genetic algorithms in search, optimization and machine learning,
Addison-Wesley, Reading, MA
Holland JH (1975) Adaptation in natural and artificial systems, University of Michigan Press, Ann
Arbor, MI
Jan SS, Lin YC (2009) A new multi-C/A code acquisition method for GPS. GPS Solut 13(4): 293–303
(DOI: 10.1007/s10291-009-0122-7)
Namgoong W, Meng TH (2001) Minimizing power consumption in direct sequence spread spectrum
correlators by resampling IF samples-part I: performance analysis. IEEE Trans. Circuits Syst. II,
Analog Digit. Signal Process. 48(5):450–459
Pang J, Van Graas F, Starzyk J, Zhu Z (2003) Fast direct GPS P-code acquisition. GPS Solut
7(3):168–175
Parkinson BW, Spilker JJ (eds.) (1996) Global Positioning System: theory and applications, vol I.
American Institute of Aeronautics and Astronautics, Washington, DC
Rockwell International Corporation (1991) Interface Control Document ICD-GPS-200, Downey,
California, 9–86
Sagiraju PK, Raju GVS, Akopian D (2008) Fast acquisition implementation for high sensitivity global
positioning systems receivers based on joint and reduced space search. IET Radar Sonar
Navigation 2(5):376–387
Starzyk J, Zhu Z (2001) Averaging correlation for C/A code acquisition and tracking in frequency
domain. Proc. IEEE Midwest Symp. on Circuits and Syst. 2:905–908
Torrieri D (2004) Principles of Spread-Spectrum Communication Systems. Springer. ISBN:
0387227822
Tsui JBY (2004) Fundamentals of global positioning system receivers—a software approach, 2nd ed,
Wiley, Hoboken. ISBN: 978-0-471-70647-2
Van Nee DJR, Coenen AJRM (1991) New fast GPS code-acquisition technique using FFT. Electron
Lett 27(2):158–160
Wilde WDE, Sleewaegen JM, Simsky A et al. (2006) New fast signal acquisition unit for GPS/Galileo
receivers. Proc. ENC GNSS.
Zigangirov KS (2004) Theory of code division multiple access communication, IEEE Press, Wiley,
New York. ISBN: 978-0-471-45712-1
Table captions:
Table 1 Comparative performance of different acquisition methods.

Figure captions:

Fig. 1 Flowchart of GNSS acquisition procedure with adaptive GA.

Fig. 2 Evolution of Doppler frequency. (C/A-code)
Fig. 3 Evolution of code delay. (C/A-code)
Fig. 4 Maximum search power margin versus C/No (C/A-code)
Fig. 5 Maximum search power margin versus C/No (P-code)
Fig. 6 Maximum search power margin versus C/No (BOC).
Fig. 7 Signal detection probability versus C/No.
Fig. 8 Maximum number of initial population versus C/No.
Fig. 9 Number of initial population versus iteration (C/No=45 dB-Hz)
Fig. 10 Acquisition result of C/A-code with adaptive GA [zoom]. (1500 population size; 13 iteration
times)
Fig. 11 Acquisition result of P-code with adaptive GA [zoom]. (7000 population size; 25 iteration
times)
Fig. 12 Acquisition result of BOC with adaptive GA [zoom]. (5750 population size; 19 iteration times)
 Table 1 Comparative performance of different acquisition methods.
Parallel code GA with
Serial Parallel frequency Chang’s method
phase adaptive
P-code/ P-code/ P-code/ P-code/
C/A-code C/A code P-code/BOC C/A-code C/A-code C/A-code
BOC BOC BOC BOC
Execute 2202.3/ 1392.3/ 1793.8/ 1226.1/ 1183.2/
1505.2 720.5 901.2 700.4 650.4
time 1704.3 983,2 1103.6 960.4 850.9
Repetitions 1ms 0.5s/4ms 140/78
401 ( 50 Hz step 401 ( 50 Hz step 401 401 34 (10~
(Without code chip code chip (10~ 39 145/96
size) size) ( 50 Hz step size) ( 50 Hz step size) 1KHz)
Pull-in) length length 280Hz)
Poor Poor Fair Fair
Parameters
(Count on number of (Count on number of (Count on number of (Count on number of Poor Poor High Fair High High
Precision
bins) bins) bins) bins)
High
Fair High
Fair High (Count on
Complexity Low Fair (Count on FFT (Count on Fair High Fair
(Count on FFT size) (Count on FFT size) number of
size) FFT size)
population)
Fig. 1 Flowchart of GNSS acquisition procedure with adaptive GA.
Fig. 2 Evolution of Doppler frequency. (C/A-code)

Fig. 3 Evolution of code delay. (C/A-code)

Fig. 4 Maximum search power margin versus C/No (C/A-code)

Fig. 5 Maximum search power margin versus C/No (P-code)

Fig. 6 Maximum search power margin versus C/No (BOC).

Fig. 7 Signal detection probability versus C/No.

 Fig. 8 Maximum number of initial population versus C/No.

Fig. 9 Number of initial population versus iteration (C/A-code; C/No= 45 dB-Hz)

Fig. 10 Acquisition result of C/A-code with adaptive GA [zoom]. (1500 population size; 13 iteration
times)

Fig. 11 Acquisition result of P-code with adaptive GA [zoom]. (7000 population size; 25 iteration
times)
Fig. 12 Acquisition result of BOC with adaptive GA [zoom]. (5750 population size; 19 iteration times)