Exact error rate analysis of equal gain and selection diversity for coherent free-space optical systems on strong turbulence channels

Mingbo Niu; Julian Cheng; Jonathan F. Holzman

doi:10.1364/OE.18.013915

1. Introduction

Free-space optical (FSO) communications has drawn considerable interest. Lower costs, larger bandwidths, better security, and greater flexibility are all FSO benefits [1]. FSO systems with diversity reception, in particular, achieve performance improvements by mitigating atmospheric turbulence. This has been shown for irradiance modulation and direct detection (IM/DD) system with OOK using optimal combining, equal gain combining (EGC), and selection combining [1, 2], and coherent detection has been successful in further improving the sensitivity and spectral efficiency [3].

With the benefits of coherent FSO in mind [3], exact BER expressions have been developed for differential phase-shift keying (DPSK) over K-distributed [4] and Gamma-Gamma distributed [5] turbulence. A heterodyne FSO system with pointing errors was studied by Sandalidis et al. [6]. Synchronous detection was then studied by Belmonte and Kahn [7, 8], and our recent work [9] demonstrated that maximum ratio combining (MRC) can provide excellent performance improvements at practical signal-to-noise ratios (SNRs).

In this work, both equal gain and selection diversity (SD) are analyzed for long-range and high-turbulence conditions with a K distribution [4,10]. Based on the new combiner SNR statistics, we derive an exact BER expression for binary phase-shift keying (BPSK) and an outage probability expression with EGC. For SD, we obtain outage probability and BER expressions with asynchronous DPSK and frequency-shift keying (FSK). The obtained error rates are highly accurate for all SNR regions. In particular, the asymptotic error rates are obtained based on the development of diversity order and coding gain.

2. The Coherent FSO System Model and Statistics

2.1. Channel Model

For the lth branch of a coherent FSO system using BPSK, with received and local oscillator beams mixed in perfect spatial coherence over a sufficiently small photodetector area. Practical issues such as beam combining and polarization are discussed in [11]. The photodetector current is [9]

i_{l} (t) = i_{dc, l} + i_{ac, l} (t) + n_{l} (t)

where i_dc,l = R(P_s,l + P_LO) and $i_{ac, l} (t) = 2 R \sqrt{P_{s, l} P_{LO}} \cos (ω_{IF} t + ϕ + ϕ_{s, l})$ are the DC and AC terms, respectively, R is the detector responsivity, and n_l(t) is a shot-noise-limited additive white Gaussian noise (AWGN) process. The power PLO is sufficiently large so that shot noise is dominant and can be modeled as AWGN [12]. Here, P_s,l is the received lth optical signal power incident on the beamsplitter, P_LO is the local oscillator power assumed to be the same for all branches, and ω_IF = ω₀ − ω_LO is the intermediate frequency, with ω₀ and ω_LO denoting the carrier frequency and local oscillator frequency, respectively. The phase information is ϕ ∈ {0, π}, and ϕ_s,l is the lth branch phase noise. It is assumed that each diversity branch turbulence is independent. For EGC cases, we assume the signal phase is well tracked at the coherent FSO detector and the phase noise can be fully compensated [3], as the optical phase varies slowly over 1 or 2 bits at a high data rate.

2.2. SNR Analysis

The SNR of an optical receiver is defined as the ratio of the time-averaged AC photocurrent power 〈i²_ac(t)〉 to the total noise variance [12]. For EGC with L diversity branches, the received signals at different branches are co-phased and added with equal weight. In such a case, as shown in the Appendix, the instantaneous SNR at the output of the combiner becomes

γ_{EGC} = \frac{R {(Σ_{l = 1}^{L} \sqrt{P_{s, l}})}^{2}}{Lq Δ f} = \frac{{2 RP}_{LO} {(Σ_{l = 1}^{L} \sqrt{I_{N, l}})}^{2}}{L} \frac{{2 E}_{b}}{N_{0}}

where q denotes the electronic charge, Δf is the noise equivalent bandwidth of the photodetector, I_N,l is the normalized K turbulent coefficient (E[I_N,l] = 1) of the lth branch, N₀/2 = qRP_LO is the noise power spectral density, and E_b = η²AT_b denotes the energy per bit. Here, E[·] denotes the expectation operation, η² is the mean irradiance, A denotes the detector area, and T_b is the bit duration. With the relationship P_s,l = AI_s,l where I_s,l is the instantaneous received optical irradiance at the lth branch, we can rewrite the instantaneous SNR as

γ_{EGC} = \frac{RA {(Σ_{l = 1}^{L} \sqrt{I_{s, l}})}^{2}}{Lq Δ f} = g^{2} {(Σ_{l = 1}^{L} \sqrt{I_{s, l}})}^{2}

where $g ≜ \sqrt{\frac{RA}{(Lq Δ f)}}$ is a scalar factor. For L = 1, the instantaneous SNR in Eq. (3) specializes to γ_EGC = RAI_s,1/(qΔf), which agrees with the single-branch SNR obtained in [4,9]. Note that the SNR expression shown in Eq. (3) is fundamentally different from the SNR of EGC in the IM/DD applications, which is related to the sum of the irradiance [1,2]. Indeed, the EGC SNR for coherent FSO systems is related to the sum of the square root of the irradiance in each branch. This distinction can greatly impact the system performance.

Given the lth branch instantaneous SNR as γ_l = RP_s,l/(qΔf) [9], the instantaneous SNR for SD is

γ_{SD} = max {γ_{1}, γ_{2}, \dots, γ_{L}} = \frac{R}{q Δ f} max {P_{s, l}; l = 1, \dots, L} .

We can observe that the instantaneous combiner SNRs are independent of the local oscillator power for large P_LO. This is an important property of coherent FSO communication (in contrast to coherent radio frequency literature), as the local oscillator power does not affect coherent FSO system performance.

2.3. Analysis for the Square Root of the Irradiance

To facilitate the performance analysis for the coherent EGC FSO systems, we can make use of the characteristic function (CF) for the square root of the irradiance I_s. In this work, we assume a signal irradiance I_s with a K-distributed model having a probability density function (PDF) given by [4]

f (I_{s}) = \frac{2}{Γ (α) η^{α + 1}} α^{\frac{α + 1}{2}} {I_{s}}^{\frac{α - 1}{2}} K_{α - 1} (\frac{2}{η} \sqrt{α I_{s}}), I_{s} > 0

where Γ(·) denotes the gamma function, K_α(·) is the modified Bessel function of the second kind of order α, and α is a channel parameter related to the scintillation index. According to the definition of the scintillation index [13]

σ_{si}^{2} ≜ \frac{E [I_{s}^{2}]}{{(E [I_{s}])}^{2}} - 1

one can show

α = \frac{2}{σ_{si}^{2} - 1} .

Scintillation indices below unity correspond to weak turbulence conditions, which is commonly modeled with a log-normal distribution. For FSO applications, the scintillation index σ²_si in the K-distributed model is (2,3) [4], and α is in (1,2). In the strong turbulence regime, the scintillation index can be calculated from the Rytov variance [13]

σ_{R}^{2} = 1.23 C_{n}^{2} k^{\frac{7}{6}} d^{\frac{11}{6}}

through the relation [13, Eq. (53)], where C²_n is the index-of-refraction structure parameter, k is the optical wave number, and d is the propagation distance. For this coherent FSO system in the strong turbulence regime, one would expect C²_n = 10⁻¹³(m^−2/3) for a link length beyond 1km and aperture sizes below 1mm². This avoids aperture averaging (and a loss of coherence) and provides sufficient distance between individual photodetectors [10]. We denote the square root of the irradiance by $z ≜ \sqrt{I_{s}}$ . Then the PDF of z can be shown as [14]

f (z) = \frac{4}{Γ (α) η^{α + 1}} α^{\frac{α + 1}{2}} z^{α} K_{α - 1} (\frac{2 \sqrt{α}}{η} z), z > 0

which finds applications in modeling the displacement of a two-dimensional unbiased random walk. To obtain the CF of z, we let z = xy, where $x ≜ \sqrt{I_{x}}$ and $y ≜ \sqrt{I_{y}}$ are found to have Rayleigh and Nakagami (when α ≥ 1/2) PDFs, respectively, as

f (x) = \frac{2}{η^{2}} {xe}^{- \frac{x^{2}}{η^{2}}}, x > 0

and

f (y) = \frac{2 α^{α} y^{2 α - 1} e^{- α y^{2}}}{Γ (α)}, y > 0 .

Averaging the CF of z conditioned on $\sqrt{I_{x}}$ , one obtains the CF of z as

ϕ_{z} (ω) = ℜ {ϕ_{z} (ω)} + j ℑ {ϕ_{z} (ω)}

where the real and imaginary parts are respectively

ℜ {ϕ_{z} (ω)} =_{2} F_{1} (α, 1, \frac{1}{2}; - \frac{ω^{2} η^{2}}{4 α})

and

ℑ {ϕ_{z} (ω)} = \frac{ω Γ (α + \frac{1}{2}) Γ (\frac{3}{2})}{Γ (α)} {(\frac{η^{2}}{α})}^{\frac{1}{2}}_{2} F_{1} (α + \frac{1}{2}, \frac{3}{2}, \frac{3}{2}; - \frac{ω^{2} η^{2}}{4 α}) .

Here, ₂F₁(·, ·, ·; ·) is the Gaussian hypergeometric function, ℜ{·} and ℑ{·} denote the real and imaginary operators, respectively. The CF of z can also be obtained from the moment generating function (MGF) of a product of two independent Nakagami RVs [15,Eq. (3)].

3. Performance Analysis of EGC FSO Systems

3.1. Error Rate Analysis

The average BER of BPSK over a turbulence channel can be derived by averaging the conditional bit-error probability as

P_{e} = \int_{0}^{\infty} P_{e} (z_{s}) f (z_{s}) d z_{s}

where z_s is the sum of L square roots of the irradiance, i.e., $z_{s} = Σ_{l = 1}^{L} \sqrt{I_{s, l}}$ . Here, f(z_s) is the PDF of z_s and P_e(z_s) denotes the conditional bit-error probability given by

P_{e} (z_{s}) = \frac{1}{2} erfc (\sqrt{\frac{γ_{EGC}}{2}}) = \frac{1}{2} erfc (\frac{g z_{s}}{\sqrt{2}})

where erfc(·) is the complementary error function. In general, it is difficult to obtain a closed-form expression for the PDF of z_s. Hence, we use the Fourier inversion theorem to obtain the PDF of z_s by

f (z_{s}) = \frac{1}{2 π} \int_{- \infty}^{\infty} ϕ_{z_{s}} (ω) e^{- j ω z_{s}} d ω = \frac{1}{2 π} \int_{- \infty}^{\infty} \prod_{l = 1}^{L} ϕ_{z_{l}} (ω) e^{- j ω z_{s}} d ω

where j² = −1, $ϕ_{z_{s}} (ω)$ is the CF of z_s and $ϕ_{z_{l}} (ω)$ is the CF of z_l (with $z_{l} = \sqrt{I_{s, l}}$ ). With Eq. (15)–Eq. (17), one can obtain the average BER as

P_{e} = \frac{1}{2 π} \int_{- \infty}^{\infty} ϕ_{z_{s}}^{*} (ω) \int_{0}^{\infty} P_{e} (z_{s}) e^{j ω z_{s}} d z_{s} d ω

where $ϕ_{z_{s}}^{*} (\cdot)$ denotes the conjugate of $ϕ_{z_{s}} (\cdot)$ . Since the Fourier transform of the complementary error function is known as [16]

𝘍 (ω) = \int_{0}^{\infty} erfc (z_{s}) e^{j ω z_{s}} d z_{s} = \frac{1}{\sqrt{π}}_{1} F_{1} (1, \frac{3}{2}, - \frac{ω^{2}}{4}) + \frac{j}{ω} (1 - e^{- \frac{ω^{2}}{4}}),

Equation (18) can be rewritten as

P_{e} = \frac{1}{2 π} \int_{- \infty}^{\infty} \prod_{l = 1}^{L} ϕ_{z_{l}}^{*} (\frac{g ω}{\sqrt{2}}) \frac{𝘍 (ω)}{2} d ω .

With a change of variable ω = tanθ, it is possible to rewrite Eq. (20) in terms of a definite integration for easier numerical evaluation

P_{e} = \frac{1}{2 π} \int_{- \frac{π}{2}}^{\frac{π}{2}} \sec^{2} θ \prod_{l = 1}^{L} ϕ_{z_{l}}^{*} (\frac{g \tan θ}{\sqrt{2}}) \frac{𝘍 (\tan θ)}{2} d θ .

Though BPSK is only considered here, it is straightforward to extend our analysis to other coherent modulations such as quadrature phase-shift keying and M-ary quadrature amplitude modulation.

3.2. Outage Probability Analysis

To find the outage probability, we make use of the Gil-Pelaez formula to express the CDF of z_s as [18]

F_{z_{s}} (z_{o}) = \frac{1}{2} - \frac{1}{π} \int_{0}^{\infty} \frac{ℑ {ϕ_{z_{s}} (ω) \exp [- j ω z_{o}]}}{ω} d ω .

Since the turbulence in each branch is independent, $ϕ_{z_{s}} (ω)$ can be written directly as $ϕ_{z_{s}} (ω) = \prod_{l = 1}^{L} ϕ_{z_{l}} (ω)$ . To continue, we express the CF of z_l in the polar form as

ϕ_{z_{l}} (ω) = \sqrt{ℜ^{2} {ϕ_{z_{l}} (ω)} + ℑ^{2} {ϕ_{z_{l}} (ω)}} \exp [j \tan^{- 1} (\frac{ℑ {ϕ_{z_{l}} (ω)}}{ℜ {ϕ_{z_{l}} (ω)}})]

where the real and imaginary parts of the CF of z_l are obtained from Eq. (13) and Eq. (14). The numerator of the integrand in Eq. (22) can now be written as

ℑ {ϕ_{z_{s}} (ω) \exp [- j ω z_{o}]} = \prod_{l = 1}^{L} \sqrt{ℜ^{2} {ϕ_{z_{l}} (ω)} + ℑ^{2} {ϕ_{z_{l}} (ω)}} \sin [Σ_{l = 1}^{L} \tan^{- 1} (\frac{ℑ {ϕ_{z_{l}} (ω)}}{ℜ {ϕ_{z_{l}} (ω)}}) - ω z_{o}] .

With a change of variable, substitution of Eq. (24) into Eq. (22) gives the outage probability as

P_{o, EGC} (γ^{*}) = \frac{1}{2} - \frac{2}{π} \int_{0}^{\frac{π}{2}} \frac{\prod_{l = 1}^{L} \sqrt{ℜ^{2} {ϕ_{z_{l}} (\tan θ)} + ℑ^{2} {ϕ_{z_{l}} (\tan θ)}}}{\sin 2 θ \csc [Σ_{l = 1}^{L} \tan^{- 1} (\frac{ℑ {ϕ_{z_{l}} (\tan θ)}}{ℜ {ϕ_{z_{l}} (\tan θ)}}) - \tan θ \frac{ω \sqrt{γ^{*}}}{g}]} d θ

where γ^* is the specified outage threshold. Equation (25) can provide accurate outage calculation with a simple numerical integration algorithm.

4. Performance Analysis of SD FSO Systems

SD has the least complexity when compared to MRC and EGC, as it simply selects the branch with the strongest instantaneous received SNR. In the following, we present exact outage probability and exact error rates expressions for coherent FSO systems with SD reception.

4.1. Outage Probability Analysis

After a change of variable and help from [17,Eq. 6.561(8)], we can obtain the CDF of the optical signal irradiance, I_s,l, at the lth branch from Eq. (5) as

F_{I_{s, l}} (I) = 1 - \frac{2 {(α_{l} I)}^{\frac{α}{2}}}{Γ (α_{l}) {η_{l}}^{α_{l}}} K_{α_{l}} (\frac{2}{η_{l}} \sqrt{α_{l} I}), I > 0 .

It is straightforward to show that the outage probability with SD becomes

P_{o, SD} (γ^{*}) = \prod_{l = 1}^{L} F_{I_{s, l}} (\frac{γ^{*}}{g^{2}}) .

4.2. Error Rate Analysis for Differential and Asynchronous Detection

Unlike the outage probability analysis, the desired error rate analysis depends upon the modulation scheme and requires the PDF of I_s,SD ≜ max{I_s,l, l = 1, …, L}. For identically distributed turbulence, we can show the PDF of I_s,SD as

f_{I_{s, SD}} (I_{s}) = L {[F_{I_{s, l}} (I_{s})]}^{L - 1} f (I_{s}) .

Differential and asynchronous detection are two alternatives for synchronous detection to minimize phase noise effects. In this work, coherent FSO with asynchronous detection refers to coherent photodetection at the receiver front-end, followed by envelope detection for signal recovery [12, P. 490]. The conditional BER with an average SNR of $\bar{γ}$ is known to be

P_{e} (I_{s}) = \frac{1}{2} \exp [- β \bar{γ} I_{s}]

where β = 1/2 for DPSK (assuming moderate frequency offset) and β = 1/4 for asynchronous FSK [12]. Hence, the average BER for DPSK or FSK with SD is

P_{e} = \frac{1}{2} \int_{0}^{\infty} \exp [- β \bar{γ} I_{s}] f_{I_{s, SD}} (I_{s}) {dI}_{s} = \frac{1}{2} \int_{0}^{\frac{π}{2}} \sec^{2} θ \exp [- β \bar{γ} \tan θ] f_{I_{s, SD}} (\tan θ) d θ .

5. Analysis of Diversity Order and Coding Gain

The analysis in this section allows us to gain valuable insight into BER behavior in the large SNR region for coherent FSO with both EGC and SD reception. The asymptotic BER can be expressed as $P_{e} \dot{=} {(G_{c} \bar{γ})}^{- G_{d}}$ where G_d and G_c respectively denote diversity order and coding gain [9]. Similarly we denote G_dl and G_cl as the diversity order and coding gain of the lth branch.

Using the MGF of a K-distributed RV and following [9], one can show that the diversity order and coding gain offered by the lth branch are G_dl = 1 and G_cl = 2βη²(α−1)/α respectively. Consequently, the diversity order of SD becomes G_d,SD = ∑^L_l=1 G_dl = L and the coding gain of SD can be derived as

G_{c, SD} = {[\frac{2^{L - 1} Γ (Σ_{l = 1}^{L} G_{dl}) Σ_{l = 1}^{L} (G_{dl})}{\prod_{l = 1}^{L} G_{cl}^{G_{dl}} Γ (G_{dl}) G_{dl}}]}^{- \frac{1}{Σ_{l = 1}^{L} G_{dl}}} .

For BPSK EGC, the diversity order and coding gain offered by the lth branch can be found to be G_dl = 1 and G_cl = 2η²(α − 1)/α respectively. The overall EGC diversity order is G_d,EGC = L, and the expression for the overall EGC coding gain, G_c,EGC can be found in [19,Eq. (12)]. The diversity order here is similar to the diversity order observed in an RF system (with the exception of the Nakagami-m fading model).

6. Numerical Results

Coherent FSO numerical case studies are presented. The average SNR per branch is $\bar{γ}$ = RAη²/(qΔf). BER curves for BPSK-modulated signals with EGC diversity reception (α = 1.8) are presented in Fig. 1 for average SNRs from 0 to 30 dB. Asymptotic BERs are also shown and agree with our analytical results when the SNR is large. When the number of diversity branches increases, the BER performance improves. At an average SNR=30 dB, a three-branch EGC reception achieves an error rate of 2.5 × 10⁻⁸. Though it is not shown in Fig. 1, all analytical results have been verified with computer simulations and they have excellent agreement.

Fig. 1. The exact BERs and asymptotic BERs of BPSK for coherent EGC diversity operating on L-branch K-distributed turbulence channel with α = 1.8.

Download Full Size | PDF

Figure 2 compares the BER performance of MRC and EGC for BPSK-modulation and shows that coherent FSO systems using EGC offer a close error rate performance to that of MRC. At a BER of 10⁻⁷, 1 dB SNR (or less) is required for EGC to achieve the MRC performance. Hence, EGC offers a good balance between system performance and complexity.

To further reduce the system complexity, SD becomes a viable alternative. The received optical signals need to be co-phased for EGC, but this procedure is not required for SD reception. Therefore, SD can reduce the receiver complexity. The SD outage probability is shown in Fig. 3 along with EGC and MRC. It is apparent that the outage probability can be improved by using systems with increased complexity (e.g. MRC) and higher diversity orders. With little knowledge of the turbulence phase noise, asynchronous DPSK and FSK with SD become the preferred choices. BERs for these two modulation schemes are presented in Fig. 4 and Fig. 5. It is apparent that SD reception mitigates turbulence effects for differential and asynchronous modulations. Moreover, the performance improves as the effective aperture size increases. The asymptotic solutions are also shown in both Fig. 4 and Fig. 5. It is found that they agree with the analytical results when the SNR becomes large. Turbulence effects are shown in Fig. 6 for σ²_si ranging from 2.1 to 2.9 for both MRC and EGC. As expected, BER performance improves for coherent FSO communication as the value of σ²_si decreases.

It is interesting to observe from Fig. 6 that EGC with σ²_si = 2.1 outperforms MRC with σ²_si = 2.5 for the same number of diversity branches. For practical applications, a modest parametrization error in estimating the scintillation indices or channel parameters will overshadow the benefits provided by MRC. In such a case, EGC may be a better choice due to its reduced complexity.

Fig. 2. Exact BERs of BPSK for coherent MRC and EGC diversity operating on L-branch K-distributed turbulence channel with α = 1.8.

Download Full Size | PDF

Fig. 3. Outage probabilities of coherent MRC, EGC and SD operating on L-branch K-distributed turbulence channel with α = 1.8.

Download Full Size | PDF

Fig. 4. The exact BERs and asymptotic BERs of asynchronous DPSK for SD operating on L-branch K-distributed turbulence channel with α = 1.8.

Download Full Size | PDF

Fig. 5. The exact BERs and asymptotic BERs of asynchronous FSK for SD operating on L-branch K-distributed turbulence channel with α = 1.8.

Download Full Size | PDF

Fig. 6. The impact of scintillation index σ²_si on BPSK BERs for MRC and EGC diversity operating with three diversity branches on K-distributed turbulence channels.

Download Full Size | PDF

7. Conclusion

In this work, we have studied the performance of diversity reception for coherent FSO communication systems in strong atmospheric turbulence. The exact BER expressions and outage probability with EGC and SD reception have been obtained. Our numerical results show that coherent FSO transmission with EGC outperforms that with SD and gives comparable performance to that of MRC with reduced complexity, while the numerical results have demonstrated that the new analytical expressions are useful in providing highly accurate error rate estimation. In addition, we studied the diversity order and coding gain for coherent FSO combining showing the performance impact of optical systems and channel parameters. Our analytical study can be easily extended to coherent FSO communications under different turbulence conditions.

Appendix

At the output of the lth photodetector, the received photocurrent is given in Eq. (1). Let $c_{l} = 2 R \sqrt{P_{LO} P_{s, l}}$ , and the AC current can be expressed as follows

i_{ac, l} (t) = c_{l} [\cos ϕ \cos (ω_{IF} t + ϕ_{s, l}) - \sin ϕ \sin (ω_{IF} t + ϕ_{s, l})] .

Then, two real filters are used to implement the complex filtering in the downconverter process. The real and imaginary parts of the baseband signal are, respectively, obtained as

y_{c, l} (t) = \sqrt{2} {c_{l} [\cos ϕ \cos (ω_{IF} t + ϕ_{s, l}) - \sin ϕ \sin (ω_{IF} t + ϕ_{s, l})]} \cos (ω_{IF} t)

and

y_{s, l} (t) = - \sqrt{2} {c_{l} [\cos ϕ \cos (ω_{IF} t + ϕ_{s, l}) - \sin ϕ \sin (ω_{IF} t + ϕ_{s, l})]} \sin (ω_{IF} t) .

After a lowpass filter, the inphase and quadrature components become

{\tilde{y}}_{c, l} (t) = \frac{c_{l}}{\sqrt{2}} [\cos ϕ \cos ϕ_{s, l} - \sin ϕ \sin ϕ_{s, l}]

and

{\tilde{y}}_{s, l} (t) = \frac{c_{l}}{\sqrt{2}} [\cos ϕ \cos ϕ_{s, l} + \sin ϕ \sin ϕ_{s, l}] .

Now we combine (35) and (36) to obtain the equivalent baseband signal as

{\tilde{y}}_{l} (t) = {\tilde{y}}_{c, l} (t) + j {\tilde{y}}_{s, l} (t) = \frac{c_{l}}{\sqrt{2}} [\cos ϕ + j \sin ϕ] e^{j ϕ_{s, l}} .

The last step is to compensate the phase noise, giving the branch signal before combining as

{\tilde{y}}_{l} (t) e^{- j ϕ_{s, l}} = \sqrt{2} R \sqrt{P_{LO} P_{s, l}} e^{j ϕ} .

Consequently, with the assumption of equal noise variance in each diversity branch, the instantaneous SNR at the output of the equal gain combiner becomes

γ_{EGC} = \frac{{(Σ_{l = 1}^{L} \sqrt{2} R \sqrt{P_{LO} P_{s, l}})}^{2}}{Σ_{l = 1}^{L} σ_{n, l}^{2}} = \frac{{(Σ_{l = 1}^{L} \sqrt{2} R \sqrt{P_{LO} P_{s, l}})}^{2}}{L (2 q R P_{LO} Δ f)} = \frac{R {(Σ_{l = 1}^{L} \sqrt{P_{s, l}})}^{2}}{Lq Δ f}

where σ²_n,l = 2qRP_LOΔf,l = 1, …, L, is the noise variance in the lth diversity branch at the receiver side, and it is assumed to be the same for all diversity branches.

Acknowledgments

The research was supported in part by the Natural Sciences and Engineering Research Council of Canada (NSERC).

References and links

1. M. M. Ibrahim and A. M. Ibrahim, “Performance analysis of optical receivers with space diversity reception,” IEE Proc. Commun. 143, 369–372 (1996). [CrossRef]

2. T. A. Tsiftsis, H. G. Sandalidis, G. K. Karagiannidis, and M. Uysal, “Optical wireless links with spatial diversity over strong atmospheric turbulence channels,” IEEE Trans. Wirel. Commun. 8, 951–957 (2009). [CrossRef]

3. E. Lee and V. Chan, “Diversity coherent receivers for optical communication over the clear turbulent atmosphere,” in Proceedings of IEEE International Conference on Communications (Institute of Electrical and Electronics Engineers, New York, 2007).

4. K. Kiasaleh, “Performance of coherent DPSK free-space optical communication systems in K-distributed turbulence,” IEEE Trans. Commun. 54, 604–607 (2006). [CrossRef]

5. T. A. Tsiftsis, “Performance of heterodyne wireless optical communication systems over Gamma-Gamma atmospheric turbulence channels,” Electron. Lett. 44, 373–375 (2008). [CrossRef]

6. H. G. Sandalidis, T. A. Tsiftsis, and G. K. Karagiannidis, “Optical wireless communications with heterodyne detection over turbulence channels with pointing errors,” J. Lightwave Technol. 27, 4440–4445 (2009). [CrossRef]

7. A. Belmonte and J. M. Kahn, “Performance of synchronous optical receivers using atmospheric compensation techniques,” Opt. Express 16, 14151–14162 (2008). [CrossRef] [PubMed]

8. A. Belmonte and J. M. Kahn, “Capacity of coherent free-space optical links using diversity combining techniques,” Opt. Express 17, 12601–12611 (2009). [CrossRef] [PubMed]

9. M. Niu, J. Cheng, J. F. Holzman, and L. McPhail, “Performance analysis of coherent free-space optical communication systems with K-distributed turbulence,” in Proceedings of IEEE International Conference on Communications (Institute of Electrical and Electronics Engineers, New York, 2009).

10. R. L. Phillips and L. C. Andrews, “Measured statistics for laser light scattering in atmospheric turbulence,” J. Opt. Soc. Am. 71, 1440–1445 (1981). [CrossRef]

11. M. Jafar, D. C. O’Brien, C. J. Stevens, and D. J. Edwards, “Evaluation of coverage area for a wide line-of-sight indoor optical free-space communication system empolying coherent detection,” IET Commun. 2, 18–26 (2008). [CrossRef]

12. G. P. Agrawal, Fiber-Optical Communication Systems, (New York: Wiley, 2002). [CrossRef]

13. L. C. Andrews, R. L. Phillips, C. Y. Hopen, and M. A. Al-Habash, “Theory of optical scintillation,” J. Opt. Soc. Am. 16, 1417–1429 (1999). [CrossRef]

14. E. Jakeman and R. J. A. Tough, “Generalized K distribution: a statistical model for weak scattering,” J. Opt. Soc. Am. A 4, 1764–1772 (1987). [CrossRef]

15. G. K. Karagiannidis, N. C. Sagias, and P. T. Mathiopoulos, “N*Nakagami: A Novel Stochastic Model for Cascaded Fading Channels,” IEEE Trans. Commun. 55, 1453–1458 (2007). [CrossRef]

16. A. Annamalai, C. Tellambura, and V. K. Bhargava, “Exact evaluation of maximal-ratio and equal-gain diversity receivers for M-ary QAM on Nakagami fading channels,” IEEE Trans. Commun. 47, 1335–1344 (1999). [CrossRef]

17. I. S. Gradshteyn and I. M. Ryzhik, Table of Integrals, Series, and Products, (San Diego: Academic Press, 2000).

18. Q. T. Zhang, “Outage probability in cellular mobile ratio due to nakagami signal and interferers with arbitrary parameters,” IEEE Trans. Veh. Technol. 45, 364–372 (1996). [CrossRef]

19. Z. Wang and G. B. Giannakis, “A simple and general parameterization quantifying performance in fading channels,” IEEE Trans. Commun. 51, 1389–1398 (2003). [CrossRef]

Exact error rate analysis of equal gain and selection diversity for coherent free-space optical systems on strong turbulence channels

Abstract

1. Introduction

2. The Coherent FSO System Model and Statistics

2.1. Channel Model

2.2. SNR Analysis

2.3. Analysis for the Square Root of the Irradiance

3. Performance Analysis of EGC FSO Systems

3.1. Error Rate Analysis

3.2. Outage Probability Analysis

4. Performance Analysis of SD FSO Systems

4.1. Outage Probability Analysis

4.2. Error Rate Analysis for Differential and Asynchronous Detection

5. Analysis of Diversity Order and Coding Gain

6. Numerical Results

7. Conclusion

Appendix

Acknowledgments

References and links

Cited By

Figures (6)

Equations (39)

Optics Express