Weak beacon detection for air-to-ground optical wireless link establishment

Yaoqiang Han; Anhong Dang; Junxiong Tang; Hong Guo

doi:10.1364/OE.18.001841

1. Introduction

Free-space optical communication (FSOC) presents a lot of promising features in the next generation wireless communication system because of its advantages over radio frequency technology, such as high data rates, high transmission security, and fast deployment [1,2]. Due to the limitation of current microwave links, it receives special attention in the field of mass information exchange between the airborne platform and the ground station [3,4].

FSOC system is often a point-to-point system, and one challenge for FSOC system is fast establishment of the optical link. The laser pointing, acquisition and tracking (PAT) technology is critical for the optical link establishment. And one of the most important steps of PAT is to determine exact positions of communication terminals, which is the so-called beacon acquisition. A typical beacon acquisition process is usually as below: First, the transmitter and the receiver point their antennas to each other based on the priori knowledge, then the transmitter emits a beacon. The receiver detects the beacon spot from received images and determines the transmitter’s exact position, after that the receiver sends a laser beam back to the transmitter. Beacon acquisition is accomplished after the transmitter successfully determines the receiver’s position by detecting the feedback laser beam.

One big problem of beacon acquisition is that the beacon will be attenuated and disrupted by atmosphere, and become weak and distorted when it arrives at the receiver. Therefore, fast detection of the beacon hidden in the background is critical for beacon acquisition [5]. When the background interference is relatively weak and uniform, such as in inter-satellite laser communications, the brightest spot is simply chosen as the beacon [6]. However, in the case of strong and nonuniform background interference, such as in air-to-ground optical communications, mirror reflections of sunlight and other bright speckles in the image might be brighter than the beacon, thus the performance of the traditional detection scheme is degraded seriously. Although the detection scheme based on the beacon shape extraction has been proposed, its performance is also seriously affected by atmosphere [7,8]. To the best of our knowledge, almost all beacon detection schemes in practical PAT systems are traditional detection schemes, because both the computation power of the onboard signal processing system and the acquisition time are limited [9,10].

New beacon detection schemes utilizing signal processing algorithm become feasible due to the development of electronic devices, especially the field programmable gate array (FPGA). Many practical techniques are available for weak signal detection, such as matched filtering [11], Fourier filtering [12], and wavelet transform [13]. Compared with other techniques, matched filtering features low complexity and fast detection, therefore we utilize matched filtering in the proposed scheme. Matched filter detects the presence of the designated signal by correlating the designated signal with the received signal, therefore the correlation characteristics of the designated signal determines the noise suppression ability.

In our scheme, we propose an m-sequence as the designated signal to modulate the beacon, because m-sequence has good autocorrelation and cross-correlation characteristics [14]. However, traditional matched filter design has little effect on suppressing strong interference, therefore digital differential matched filter (DDMF) is introduced, which aims to suppress strong interference. Moreover, for fast beacon detection, each pixel of the image sensor has its own DDMF of the same structure to process its received image sequence in parallel, and these DDMFs form an array at the receiving end. Theoretical analysis and outdoor experiment show that the proposed scheme can realize fast beacon detection under strong interference conditions, and consequently reduce the required beacon transmission power dramatically.

The structure of the paper is arranged as follows: The correlation beacon detection scheme introduction and theoretical analysis are presented in Section 2. The scheme performance simulation and the outdoor validation experiment are described in Section 3. A comparison between the proposed scheme and the traditional detection scheme is also presented in this section. A summarization is given in the last section.

2. The correlation beacon detection scheme

2.1 Workflow of the correlation beacon detection scheme

The PAT workflow utilizing the proposed beacon detection scheme is demonstrated in Fig. 1 . Compared with the traditional PAT workflow, beacon detection on the airborne platform utilizes the proposed scheme. But applying this scheme at the ground station is unnecessary, since the sky background in the image is relatively weak and uniform.

Fig. 1 PAT process of the air-to-ground optical wireless link.

Download Full Size | PDF

An m-sequence generator is modulating the beacon at the ground station. The beacon travels through the atmosphere and illuminates the area where the airborne platform is located. The camera on the airborne platform sends the received image sequence to the DDMF array and the outputs of the DDMF array form a new image, from which the beacon is detected. After the exact position of the ground station is determined, another laser beam is emitted from the airborne platform to the ground station. And beacon acquisition is accomplished after the PAT system of the ground station determines the airborne-platform’s position by detecting the feedback laser beam.

2.2 Atmospheric turbulence effect

From Fig. 1, it can be seen that beacon will travel through atmosphere before it arrives at the receiver, and atmosphere turbulence will impact big influence on the propagation of the beacon. Atmospheric turbulence is a common problem for FSOC system, which leads to the laser beam irradiance fluctuation, and it not only degrades the PAT module’s performance, but also affects the communication performance. In order to quantify the irradiance fluctuation, scintillation index (SI) $σ_{I}^{2}$ is introduced, and its definition is as follow [15]:

σ_{I}^{2} = \frac{< I^{2} >}{< I >^{2}} - 1,

where I denotes the received laser beam irradiance.

σ_{I}^{2}

is related to the weather condition, propagation distance and the FSOC system design parameters.

Since atmospheric turbulence affects FSOC system seriously, system design parameters such as receiver aperture size have to be optimized to weaken the atmospheric turbulence effect in practical FSOC systems. In [16], an analysis of the laser beam irradiance fluctuation in stratospheric optical payload experiment (STROPEX) is presented. In this experiment, the altitude of the airborne platform is 22 km, and the maximum distance between the airborne platform and the ground station is 61 km. In order to weaken the irradiance fluctuation, the aperture size of the antenna is 40 cm in this experiment, and the experiment result shows that SI is decreased to ten percent of that in the case of without antenna. Due to the aperture averaging, SI can be kept below 0.1 in most situations [16].

Based on the discussion above, it can be concluded that SI is small after the FSOC system design parameters have been optimized. And in the case of weak fluctuation, lognormal distribution is always used to describe the probability density function (PDF) of the laser beam irradiance, and it is expressed as below [15]:

p (I_{n}) = \frac{1}{I_{n} \sqrt{2 π σ_{I}^{2}}} \exp {- \frac{{[\ln (I_{n}) + \frac{1}{2} σ_{I}^{2}]}^{2}}{2 σ_{I}^{2}}},

where

I_{n}

is the normalized laser beam irradiance. The PDF of the normalized laser beam irradiance is shown in Fig. 2 , and SI takes 0.005, 0.01, 0.05 and 0.1 respectively. It can be seen that the PDF of the laser beam irradiance get close to the normal distribution when SI is relative small.

Fig. 2 The PDF of the normalized laser beam irradiance.

Download Full Size | PDF

The laser beam irradiance fluctuation will affect the beacon detection performance, and it should be taken into account. Since SI can be decreased to a small value by optimizing the FSOC system parameters, we focus on the discussion of the atmospheric turbulence effect in the case of small SI in this paper. When SI is relative small, the PDF of the normalized laser beam irradiance approximately follows the normal distribution and the irradiance fluctuation is weak. Therefore the effect of irradiance fluctuation would be treated approximately as Gaussian noise in this paper. However, when SI is big, the above approximation is obviously not suitable.

2.3 Background interference and noise

Beacon detection is affected by the interference and the noise. In the received images, there are bright speckles which might be brighter than the beacon, and all bright speckles except the beacon are treated as the interference. There are also various kinds of noise, which can be divided into two classes. The first class is intrinsic noise, such as dark current noise of the image sensor (CCD or CMOS), and the electronic noise of circuits. The second is related to the received light power of each pixel, such as shot noise and the noise which is an approximation of the laser beam irradiance fluctuation as discussed above. For simplicity, all kinds of noise are treated as additive white Gaussian noise (AWGN) [17,18]. $σ_{e l e}^{2}$ denotes the first class noise power variance, and $σ_{s h o t}^{2} (i)$ denotes the second class noise power variance of the i^th pixel in the image. Thus, the standard deviation of the noise power of the i^th pixel is:

σ_{n} (i) = \sqrt{σ_{e l e}^{2} + σ_{s h o t}^{2} (i)} .

For quantitative analysis of the interference and the noise, two parameters are introduced: image signal-to-interference ratio (ISIR) and image signal-to-noise ratio (ISNR). And the definition of ISIR is:

I S I R = \frac{P_{b}}{P_{intmax}},

where P_b denotes the beacon power, and P_intmax denotes the maximum interference power in the image. And the definition of ISNR is:

I S N R = \frac{P_{b e f}}{σ_{nmax}^{}},

where P_bef denotes the beacon power component when the interference has been eliminated. In traditional detection schemes,

P_{b e f} = P_{b} - P_{intmax}

. And we use

σ_{nmax}^{}

denoting the maximum standard deviation of the noise power of the pixels in the image.

2.4 DDMF array design

In order to realize fast beacon detection, DDMF array is introduced, and each pixel of the image sensor has its own DDMF of the same structure to process its received image sequence in parallel. The structure of DDMF is shown in Fig. 3 .

Fig. 3 The structure of DDMF.

Download Full Size | PDF

As shown in Fig. 3, DDMF consists of two matched filters, which have the same structure and different scaling factors: $a (n)$ and $b (n)$ . We assume that $S (n)$ is the binary m-sequence of length N( $N = 2^{l} - 1$ ), where l is the shift register number of the sequence generator. According to the digital matched filter theory, the scaling factor vector is the conjugated time-reversed version of the m-sequence [14]. Therefore, scaling factor vectors $a (n)$ and $b (n)$ are expressed as below:

a (n) = S (n), b (n) = S (n + q), n \in (0, 1, \dots N ​ - 1),

where q denotes the phase difference between a (n) and b (n), and

q \in (1, 2, \dots N - 1)

. The DDMF output

U_{0} (m, i)

is the subtraction of these two filter outputs.

The image sequence of each pixel is sent into its own DDMF. Since the gray value of the image denotes the received power in the integration time, we use gray value denoting the signal power in the calculation below, which is an integer that ranges from 0 to 255 in this paper. I(m,i) denotes the gray value of the i^th pixel in the m^th image, and it can be expressed as follow:

I (m, i) = {\begin{cases} I_{b a c k} (m, i) only background \\ I_{b e a c o n} (m, i) + I_{b a c k} (m, i) background with b e a c o n, \end{cases}

where

I_{b a c k} (m, i)

denotes the gray value of the background component, and

I_{b e a c o n} (m, i)

denotes the gray value of the beacon component modulated by the binary m-sequence.

In order to sample the proper value of the beacon, over sampling has to be utilized, and $N_{s}$ denotes the over sampling ratio. In one calculation period, onlyNsamples are utilized, and other samples are shifted to next registers which are denoted by $T_{d}$ in Fig. 3. There are $N_{s} - 1$ registers between the adjacent scaling factors. $U_{o} (m, i)$ denotes the $m_{}^{t h}$ filter output of the $i_{}^{t h}$ pixel, and it is expressed as:

\begin{matrix} U_{o} (m, i) = a (n) * I (m, i) - b (n) * I (m, i) \\ = \sum_{k = 0}^{N - 1} a (k) I (m + k, i) - \sum_{k = 0}^{N - 1} b (k) I (m + k, i), \end{matrix}

where ∗represents the convolution operator.

2.5 Theoretical calculation

In our proposed scheme, beacon detection is performed in the processed image formed by the outputs of the DDMF array, therefore, the output distribution function of each DDMF is first deduced. For the sake of discussion, the beacon emission state at the transmitter is defined, $H_{0}$ denotes the beacon is not emitted, and $H_{1}$ denotes the beacon is emitted. And we assume the background interference power is time independent during the limited acquisition time.

From Eq. (8), it can be seen that the DDMF is a linear filter, therefore, the filter output also follows the Gauss distribution the same as the noise power distribution. Take the i^th pixel for instance, the average filter output under H₀ is:

\begin{matrix} {E (U_{0} (m, i)) |}_{H_{0}} = E [\sum_{k = 0}^{N - 1} a (k) I (m + k, i) - \sum_{k = 0}^{N - 1} b (k) I (m + k, i)] \\ = (\sum_{k = 0}^{N - 1} a (k) - \sum_{k = 0}^{N - 1} b (k)) \cdot E (I_{b a c k} (m, i)) \\ = 0. \end{matrix}

Equation (9) indicates that the interference can be eliminated utilizing the differential structure. And the average filter output under $H_{1}$ is:

\begin{matrix} {E (U_{0} (m, i)) |}_{H_{1}} \\ = E [\sum_{k = 0}^{N - 1} a (k) I (m + k, i) - \sum_{k = 0}^{N - 1} b (k) I (m + k, i)] \\ = E [\sum_{k = 0}^{N - 1} a (k) (I_{b e a c o n} (m + k) + I_{b a c k} (m + k)) - \sum_{k = 0}^{N - 1} b (k) (I_{b e a c o n} (m + k) + I_{b a c k} (m + k))] \\ = {\begin{cases} \frac{N + 1}{4} P_{b p}, when the image sequence matches the above part of DDMF; \\ 0 � when the image sequence does not match the DDMF; \\ - \frac{N + 1}{4} P_{b p}, when the  image sequence matches the below part of DDMF; \end{cases} \end{matrix}

where

P_{b p}

denotes the power of each modulated beacon pulse. Equation (10) indicates that the filter output under

H_{1}

changes among three states periodically.

According to the characteristics of m-sequence, there are $(N + 1) / 2$ different scaling factors between $a (n)$ and $b (n)$ [14]. And the variance value of the filter output is:

\begin{matrix} D (U_{0} (m, i)) = D [\sum_{k = 0}^{N - 1} a (k) I (m + k, i) - \sum_{k = 0}^{N - 1} b (k) I (m + k, i)] \\ = \sum_{k = 0}^{N - 1} {(a (k) - b (k))}^{2} \cdot D (I (m + k, i)) \\ = \frac{N + 1}{2} σ_{n}^{2} (i) . \end{matrix}

Equation (11) shows that noise power variance increases with growth of the m-sequence length (N).

Based on the calculation above, the distribution function of the filter output $U_{0} (i)$ is:

p (U_{o} (i)) |_{H_{0}} = \frac{1}{\sqrt{(N + 1) π} \cdot σ_{n} (i)} \exp (- \frac{U_{o}^{2}}{(N + 1) σ_{n}^{2} (i)}),

p (U_{o} (i)) |_{H_{1}} = \frac{1}{\sqrt{(N + 1) π} \cdot σ_{n} (i)} \exp (- \frac{{(U_{o} - P)}^{2}}{(N + 1) σ_{n}^{2} (i)}),

where

P = (N + 1) \cdot P_{b p} / 4

. According to Eq. (10), though the filter output has three possible values under

H_{1}

, only the positive output is used for beacon detection in our scheme due to the positive detection threshold.

In our proposed scheme, beacon detection is to find the brightest spot in the processed image, and the maximum value must be above the threshold. For simplicity, we assume that the beacon only illuminates one pixel, or one unite of binning pixels in the image such as 2 by 2 pixels [19]. However, from the above equations, it can be seen that the filter output distribution function varies with the pixel. For simplicity, we use $σ_{n \max}$ denoting the maximum $σ_{n} (i)$ and $σ_{n \max}^{2}$ denoting the maximum $σ_{n}^{2} (i)$ , then they are inserted into the expression to obtain the upper bound of the calculation.

False alarm probability (FAP) and detection probability (DP) are key parameters for the beacon detection scheme, and we will present the calculation for FAP and DP. The FAP of the proposed scheme is expressed as follows according to Eq. (12):

\begin{matrix} P_{f a_i m a g e} = 1 - \prod_{i = 1}^{M} (1 - P_{f a_p i x e l i}) \\ = 1 - {(1 - \int_{β}^{\infty} \frac{1}{\sqrt{(N + 1) π} \cdot σ_{n \max}^{}} \exp (- \frac{U_{o}^{2}}{(N + 1) \cdot σ_{n \max}^{2}}) d U_{o})}^{M} \\ = 1 - {(\frac{1}{2} + \frac{1}{2} e r f (\frac{β}{\sqrt{N + 1} \cdot σ_{n \max}^{}}))}^{M}, \end{matrix}

where

P_{f a_p i x e l i}

is the FAP of the

i^{t h}

pixel detection in the processed image; M denotes the total pixel number in the image; β denotes the detection threshold. We define

β / σ_{n \max}^{}

as the threshold noise ratio (TNR), and Eq. (14) can be written as:

P_{f a_i m a g e} = 1 - {(\frac{1}{2} + \frac{1}{2} e r f (\frac{T N R}{\sqrt{N + 1}}))}^{M} .

The DP of the proposed scheme is expressed as follow:

\begin{matrix} P_{dt_image} = P {U_{o} (m, i) \geq β, U_{o} (m, i) > U_{o} (m, k), 1 \leq k \leq M, k \neq i} \\ = \int_{β}^{\infty} \prod_{k = 1, k \neq i}^{M} p {U_{o} (m, k) < U_{o} (m, i) | U_{o} (m, i)} p (U_{o} (m, i)) d U_{o} (m, i) \\ = \frac{1}{\sqrt{π}} \int_{\frac{T N R}{\sqrt{N + 1} \cdot}}^{\infty} {[\frac{1}{2} + \frac{1}{2} e r f (z)]}^{M - 1} \exp [- {(z - \frac{P}{\sqrt{N + 1} \cdot σ_{n \max}})}^{2}] d z . \end{matrix}

Since the beacon in the proposed scheme is modulated by the binary m-sequence, the average beacon power in our scheme is equivalent to the beacon power in the traditional detection scheme. According to the characteristics of the m-sequence, the average beacon power is:

P_{b_a v e} = \frac{P_{b p} \cdot (N + 1)}{2 N} .

According to Eq. (5), the ISNR of the proposed scheme is expressed as:

I S N R = \frac{P_{b_a v e}}{σ_{n \max}} .

Substituting Eq. (17) and Eq. (18) into Eq. (16), we can obtain the DP of the proposed scheme:

P_{dt_image} = \frac{1}{\sqrt{π}} \int_{\frac{T N R}{\sqrt{N + 1} \cdot σ_{n}}}^{\infty} {[\frac{1}{2} + \frac{1}{2} e r f (z)]}^{M - 1} \exp [- {(z - I S N R \cdot \frac{N}{2 \sqrt{N + 1}})}^{2}] d z .

3. Performance analysis

3.1 Simulation results

In this section, we will present the proposed scheme’s performance. According to Eq. (15) and Eq. (19), false alarm probability (FAP) and detection probability (DP) are related to image signal-to-noise ratio (ISNR) and threshold noise ratio (TNR), and we will present the relation among the above parameters. Moreover, based on the FAP and DP requirements of the beacon detection system, the ISNR, TNR value can be determined. In this paper, we assume that DP is above 0.99 and FAP is below 0.01, which are taken in the practical system [20].

The relationship among DP, ISNR and TNR is shown in Fig. 4 . DP increases with the growth of ISNR, and approaches 1 when ISNR exceed ISNR₀. ISNR₀ decreases with the growth of the m-sequence length (N), because more beacon pulse power $P_{b p}$ is accumulated with the growth of N according to Eq. (10). Figure 4 also shows that DP drops down quickly when TNR exceeds TNR₀. TNR₀ increases with the growth of N, because the noise power variance $σ_{^{n}}^{2}$ is proportion to N according to Eq. (11).

Fig. 4 DP versus TNR and ISNR with different N. (a) N takes 7, (b) N takes 15, (c) N takes 31, (d) N takes 63.

Download Full Size | PDF

The relationship between FAP and TNR with different N is shown in Fig. 5(a) , and N takes the values of 7, 15, 31 and 63, respectively. From Eq. (15) and Fig. 5(a), it can be seen that FAP drops down rapidly to zero when TNR exceed a certain value. The FAP versus TNR curves with different N are also presented, and TNR increase with the growth of N, because the noise power variance $σ_{^{n}}^{2}$ is proportion to N according to Eq. (11). For the sake of the comparison with the traditional detection scheme, ISNR and TNR requirements should be determined when FAP is below 0.01 and DP is above 0.99. When FAP equals to 0.01, the corresponding TNR values with different N are marked in Fig. 5(a).

Fig. 5 (a) FAP versus TNR, (b) ISNR versus TNR when DP equals to 0.99.

Download Full Size | PDF

Based on Fig. 4, the relationship between ISNR and TNR when DP equals to 0.99 is plotted in Fig. 5(b). When TNR is greater than the corresponding value marked in Fig. 5(a), ISNR has to increases to make the filter output greater than the threshold. And when TNR is less than the corresponding value, ISNR has to be above a certain value to make DP not less than 0.99. The ISNR versus TNR curves with different N are also presented. ISNR decreases with the growth of N, because the increasing speed of the peak filter output $U_{o} (i)$ is faster than that of the noise power variance $σ_{^{n}}^{2}$ with the growth of N.

Based on the discussion above,when DP is above 0.99 and FAP is below 0.01, the corresponding ISNR and TNR requirements with different N have been determined and they are summarized in Table 1 . From Table 1, we can see that the beacon can be effectively detected under low ISNR conditions. Based on the theoretical calculation and simulation, it can be concluded that this new scheme can suppress both strong interference and noise.

Table 1. The requirements for ISNR and TNR with different N (FAP<0.01, DP>0.99)

View Table | View all tables in this article

Based on the theoretical calculation and simulation, it can be concluded that this new scheme can suppress both strong interference and noise.

3.2 Outdoor experiment

An outdoor experiment has been conducted to validate the proposed scheme in April, 2009. Because of the limitation of experiment conditions, we conducted the experiment on the ground, and the experiment setup is shown in Fig. 6 . A beacon laser and an m-sequence generator are set at the transmitting terminal. The beacon laser is operating at 785 nm and its transmission power is reduced as low as possible to simulate the weak beacon. The m-sequence generator generates m-sequences with length of 7, 15, 31 and 63 respectively, and the modulation frequency is 24 Hz. A camera, an image processing board and a PC are set at the receiving end. The camera model is FFMV-03MTM by Point Grey Inc., and its image is monochrome with 8-bit resolution. In order to over sampling the beacon, the frame rate of the camera is 96 Hz, which is four times of the beacon modulation frequency. The image processing board, which is based on Virtex-II pro development system by Xilinx Inc., processes the received image sequence utilizing our proposed scheme and sends the results to PC. And the total beacon detection time in our current experiment varies from 0.6 s to 5.2 s, which is related to the m-sequence length N. The distance between the two terminals is about 300 m,

Fig. 6 Outdoor experiment setup.

Download Full Size | PDF

Experiments in the cases of different N and different ISNR have been conducted, and one of the experiment results is presented in Fig. 7 . Figure 7(a) is the original image captured by the camera. The image size is 288 by 288, and the beacon spot size is 2 by 2. The red circle indicates the location of the beacon, which is almost invisible in the image. The building in the image is much brighter than the beacon, and it is treated as background interference. Figure 7(b) is the gray scale map of the original image, and it is obvious that the beacon is weak compared with the background interference. Figure 7(c) shows the processing result of the original image, when N takes 7 and ISNR takes 4. It can be seen that the background interference is eliminated as indicated in Eq. (9), and the beacon is now obvious in the processed image.

Fig. 7 Outdoor experiment results.(a) original image captured by camera,(b) original gray scale map of the image,(c) processed image (N = 7, ISNR = 4).

Download Full Size | PDF

The comparison between the theoretical calculation and experiment results is presented in Fig. 8 . Four animations with different N are presented, and they demonstrate the changing process of experimental values of the maximum DDMF array output with the growth of ISNR. For a better presentation, a 2-D view of the processed image is shown and the location of the pixels illuminated by the beacon is placed at the center of the image. The red line in the image indicates the theoretical value of the maximum DDMF array output. Meanwhile the theoretical noise level is also in good agreement with the experiment result. From these animations, it can be seen that the maximum DDMF array output increases with the growth of ISNR, and the increasing speed get faster with the growth of N. When N takes 63, the beacon can be detected even ISNR is close to 1.

Fig. 8 Single-frame excerpts from animations of the peak filter output changes with the growth of ISNR. (a) N takes 7 (Media 1). (b) N takes 15 (Media 2). (c) N takes 31 (Media 3). (d) N takes 63 (Media 4).

Download Full Size | PDF

From the above discussion, it can be concluded that the experiment result is in good agreement with the theoretical calculation. Since the distance is short and the SI value in our experiment is only 0.009, atmospheric turbulence imposes little impact on the beacon detection. And that also indicates that the approximate treatment for the atmospheric turbulence effect in the case of small SI is reasonable in our study.

3.3 Comparison between the proposed scheme and the traditional detection scheme

Compared with the traditional detection scheme, the proposed detection scheme can detect the weak beacon under strong background interference conditions, and reduce the required beacon transmission power. Thus, a comparison of the required beacon transmission power between the two schemes is made. According to the definition of ISIR and ISNR in Subsection 2.2, the ISIR and ISNR requirements determine the required beacon transmission power. Therefore, the ISIR, ISNR requirements of the two schemes are also compared.

For the sake of comparison, DP is above 0.99 and FAP is below 0.01, which are the same as in the above section. For the traditional detection scheme, ISIR should be above 1, otherwise the interference is brighter than the beacon, and ISNR should be above 7 [20]. For the proposed detection scheme, ISNR requirement varies with N according to the calculation in Subsection 3.1. Based on the experiment results, we learn that the noise power standard deviation is of the order of the fourth root of the pixel gray value. Thus the required beacon transmission power can be obtained based on the ISIR and ISNR requirements, as are summarized in Table 2 .

Table 2. Comparison of the traditional detection scheme and the proposed scheme (FAP<0.01, DP>0.99)

View Table | View all tables in this article

From Table 2, it can be seen that the beacon transmission power of the proposed scheme is only about ten percent of that of the traditional detection scheme. It can be concluded that the beacon transmission power can be reduced dramatically. Since the beacon power requirement in the proposed scheme is much lower than that in traditional schemes, the beacon beam divergence can be enlarged to cover the field of uncertainty (FOU) of the airborne platform, and no scan is needed. While a scan of the FOU is always necessary in traditional schemes due to the small beacon beam divergence [5]. Therefore, the beacon detection speed is faster than that of the traditional detection schemes.

According to the discussion above, we can conclude that our proposed beacon detection scheme can realize fast beacon detection under strong background interference conditions. However, there are some factors that may affect the proposed scheme, such as platform vibration and beam wander [15,21]. The impacts of the above factors and the parameter optimization of our scheme are being studied.

4. Conclusion

In this paper, we propose a correlation beacon detection scheme for weak beacon detection under strong interference conditions. This scheme can suppress the interference by correlation reception of the m-sequence, which modulates the beacon at the transmitting terminal, and a DDMF array is introduced to realize fast beacon detection. The performance analysis of the proposed scheme has been demonstrated, and the outdoor experiment results are in good agreement with the theoretical analysis. Based on the discussion above, it can be concluded that the proposed scheme can realize fast beacon detection under strong interference conditions, and reduces the required beacon transmission power dramatically.

Acknowledgement

This work is supported by the National Nature Science Foundation of China (60572002 and 60837004).

References and links

1. D. O'Brien and M. Katz, “Optical wireless communications within fourth-generation wireless systems [Invited],” J. Opt. Netw. 4(6), 312–322 (2005). [CrossRef]

2. V. W. S. Chan, “Free-Space Optical Communications,” J. Lightwave Technol. 24(12), 4750–4762 (2006). [CrossRef]

3. J. A. Louthain and J. D. Schmidt, “Anisoplanatism in airborne laser communication,” Opt. Express 16(14), 10769–10785 (2008). [CrossRef] [PubMed]

4. T. C. Tozer and D. Grace, “High-altitude platforms for wireless communications,” IEE Electron. Commun. Eng. J. 13(3), 127–137 (2001). [CrossRef]

5. B. Epple and H. Henniger, “Discussion on design aspects for free-space optical communication terminals,” IEEE Commun. Mag. 45(10), 62–69 (2007). [CrossRef]

6. J. Juarez, A. Dwivedi, A. Hammons, S. Jones, V. Weerackody, and R. Nichols, “Free-Space Optical Communications for Next-generation Military Networks,” IEEE Commun. Mag. 44(11), 46–51 (2006). [CrossRef]

7. B. Epple, “Development and Implementation of a Pointing, Acquisition and Tracking System for Optical Free-Space Communication Systems on High Altitude Platforms,” Degree Thesis (Ludwig Maximilians Universität, 2005).

8. T. H. Ho, S. D. Milner, and C. C. Davis, “Fully optical real-time pointing, acquisition, and tracking system for free space optical link,” Proc. SPIE 5712, 81–92 (2005). [CrossRef]

9. V. W. S. Chan, “Optical space communications,” IEEE J. Sel. Top. Quantum Electron. 6(6), 959–975 (2000). [CrossRef]

10. G. G. Ortiz, S. Lee, S. Monacos, M. Wright, and A. Biswas, “Design and development of a robust atp subsystem for the altair uav-to-ground lasercomm 2.5 gbps demonstration,” Proc. SPIE 4975, 103–114 (2003). [CrossRef]

11. C. C. Funk, J. Theiler, D. A. Roberts, and C. C. Borel, “Clustering to improve matched filter detection of weak gas plumes in hyperspectral thermal imagery,” IEEE Trans. Geosci. Rem. Sens. 39(7), 1410–1420 (2001). [CrossRef]

12. G. Saavedra, R. Martínez-Cuenca, M. Martínez-Corral, H. Navarro, M. Daneshpanah, and B. Javidi, “Digital slicing of 3D scenes by Fourier filtering of integral images,” Opt. Express 16(22), 17154–17160 (2008). [CrossRef] [PubMed]

13. X. G. Xia, C. Boncelet, and G. Arce, “Wavelet transform based watermark for digital images,” Opt. Express 3(12), 497–511 (1998). [CrossRef] [PubMed]

14. S. W. Golomb, and G. Gong, Signal Design for Good Correlation: For Wireless Communication, Cryptography, and Radar, (Cambridge University Press, Cambridge, 2005).

15. L. C. Andrews, and R. L. Phillips, Laser Beam Propagation through Random Media, Second edition, (SPIE Optical Engineering Press, Bellingham, 2005), Chap.6.

16. J. Horwath, N. Perlot, M. Knapek, and F. Moll, “Experimental verification of optical backhaul links for high-altitude platform networks: Atmospheric turbulence and downlink availability,” Int. J. Satell. Commun. Network. 25(5), 501–528 (2007). [CrossRef]

17. H. Faraji and W. J. MacLean, “CCD noise removal in digital images,” IEEE Trans. Image Process. 15(9), 2676–2685 (2006). [CrossRef] [PubMed]

18. Y. Degerli, F. Lavernhe, P. Magnan, and J. A. Farre, “Analysis and reduction of signal readout circuitry temporal noise in CMOS image sensors for low-light levels,” IEEE Trans. Electron. Dev. 47(5), 949–962 (2000). [CrossRef]

19. Z. Zhimin, B. Pain, and E. R. Fossum, “Frame-transfer CMOS active pixel sensor with pixel binning,” IEEE Trans. Electron. Dev. 44(10), 1764–1768 (1997). [CrossRef]

20. E. J. Korevaar, S. H. Bloom, K. D. Slatnick, V. J. Chan, I. H. Chen, M. D. Rivers, C. Foster, K. Choi, and C. S. Liu, “Status of SDIO/IS&T lasercom testbed program,” Proc. SPIE 1866, 116–127 (1993). [CrossRef]

21. S. Lee, J. W. Alexander, and M. Jeganathan, “Pointing and tracking subsystem design for optical communications link between the International Space Station and ground,” Proc. SPIE 3932, 150–157 (2000). [CrossRef]

	N = 7	N = 15	N = 31	N = 63
ISNR	4.2	2.8	2.0	1.4
TNR	10.0	14.3	20.1	28.5

	ISIR	ISNR	Beacon transmission power
Traditional detection scheme	>1	>7	1 (normalized)
Proposed scheme	$\approx 0.1$	$1.4 \sim 4.2$	$\approx 0.1$ (normalized)

	N = 7	N = 15	N = 31	N = 63
ISNR	4.2	2.8	2.0	1.4
TNR	10.0	14.3	20.1	28.5

	ISIR	ISNR	Beacon transmission power
Traditional detection scheme	>1	>7	1 (normalized)
Proposed scheme	$\approx 0.1$	$1.4 \sim 4.2$	$\approx 0.1$ (normalized)

Weak beacon detection for air-to-ground optical wireless link establishment

Abstract

1. Introduction

2. The correlation beacon detection scheme

2.1 Workflow of the correlation beacon detection scheme

2.2 Atmospheric turbulence effect

2.3 Background interference and noise

2.4 DDMF array design

2.5 Theoretical calculation

3. Performance analysis

3.1 Simulation results

3.2 Outdoor experiment

3.3 Comparison between the proposed scheme and the traditional detection scheme

4. Conclusion

Acknowledgement

References and links

Supplementary Material (4)

Cited By

Figures (8)

Tables (2)

Equations (19)

Optics Express