PERFORMANCE ANALYSIS OF WDM-PON ARCHITECTURE FOR PERFORMANCE ANALYSIS OF WDM-PON ARCHITECTURE FOR WIRELESS SERVICES DISTRIBUTION IN FUTURE AIRCRAFT NETWORKS D. Coelho1,2, J.M.B Oliveira1, L.M.Pessoa1,2, H.M. Salgado1 and J.C.S. Castro1 1INESC TEC(formerly INESC Porto), Porto, Portugal 2Faculdade de Engenharia, Universidade do Porto, Porto, Portugal (dcoelho,joao.b.oliveira,lpessoa,hsalgado,jcastro)@inescporto.pt Keywords: WDM, Wi-Fi, signal-to-noise ratio, error vector magnitude and intermodulation distortion. Abstract: In this work, an in-depth analysis concerning the transmission performance of IEEE802.11g/n (Wi-Fi) sig- nals in a WDM-PON system is presented. It is considered that the optical/electrical transceivers are based on low-cost 850 nm VCSELs and PIN photodiodes. System modelling includes the impact of noise generated in the optical path, such as relative intensity noise (RIN), shot noise, photodetector thermal noise, clipping and intermodulation distortion. An analytic analysis based on Volterra series is conducted and mathematical ex- pressions for both the EVM and SNR are derived. The theoretical analysis is also compared with experimental results. Among several conclusions, it is observed that the laser intermodulation distortion, clipping and RIN are the most relevant factors. 1 INTRODUCTION In the last decade, wireless communications have experienced a great expansion and the demand for higher data traffic to accommodate the new services, like VoIP (Voice over IP), IPTV (Internet Protocol Television) or Video on Demand (VoD) and peer-to- peer (P2P) have increased quickly. One application domain in which wireless networks can obtain ad- ditional use is in aviation, since there is the need for connectivity and network services “every time” and “everywhere”. Another important factor is that commercial aircraft operators are currently looking for ways to attract more customers by increasing the value of their service offerings to passengers [1]. In the future, in-flight entertainment (IFE) ser- vices should offer to passengers high speed wireless internet connectivity, using their own personal com- puters as if they were earthbound [1]. Entertainment services can comprise digital video and audio ser- vices, such as high definition (HD) video on demand, music, and satellite HDTV. Figure 1, depicts a scheme of a possible future optical fiber based network suitable for aircraft cab- ins, where the communication signals between the aircraft, ground and satellite stations are represented. Head End Access Point Ground Station Satellite Figure 1: Scheme of the aircraft communication system. PONs are quickly becoming one of the most pop- ular access systems in telecommunications. Wireless and fixed access convergence over PONs, are also top- ics of much debate [2]. They can be used to provide wireless communication services in these networks due to its high bandwidth and without electromag- netic interference. Moreover, using WDM there is a simplification, compared to TDM PONs, of the net- work topology by allocating different wavelengths to individual optical network terminator (ONT). i-ETC: ISEL Academic Journal of Electronics, Telecommunications and Computers CETC2011 Issue, Vol. 2, n. 1 (2013) ID-16 http://journals.isel.pt/index.php/IAJETC PONs can transparently delivery multiple services such as IEEE 802.11(Wi-Fi), global system for mo- bile communications (GSM), WiMAX or ultra-wide band (UWB). IEEE 802.11 or Wireless-Fidelity (Wi-Fi) is a set of standards for implementing Wireless Local Area Network (WLAN) data communication in the 2.4 and 5 GHz frequency range. IEEE 802.11g works in the 2.4 GHz frequency band and uses OFDM sig- nals with 20MHz of bandwidth and power signal of 0dBm/MHz. The subcarriers can be QPSK, 16-QAM or 64-QAM. The IEEE 802.11n standard adds the use of the Multiple-Input Multiple-Output (MIMO) tech- nology and operates in the 2.4 GHz and 5.0 GHz bands with 40-MHz bandwidth. Orthogonal Fre- quency Division Multiplexing (OFDM) is a multicar- rier modulation scheme that is well known due to its robustness in multipath fading channels. Yet, due to its high peak-to-average power ratio, OFDM is sus- ceptible to nonlinear distortion from components such as optical modulators and directly modulated diode lasers [5]. A low cost solution for these systems involves the use of directly modulated Vertical Cavity Sur- face Emitted Lasers (VCSELs). These lasers are characterized by a vertical low divergence, circular beam patterns, low threshold currents (a few mA) and high bandwidths (several GHz). Their vertical wafer growth process enables in-wafer testing, and is well suited for large scale production. These are reasons that make VCSELs desirable for low cost directly modulated systems in these types of widespread com- mercial applications [3]. This article is divided in seven sections. Section 2 describes briefly the WDM-PON topology. Sec- tion 3 addresses the interference generated by source nonlinearities. Section 4 presents the laser model. Section 5 presents discusses the system performance analysis including all noise contributions and laser distortions at the receiver. The results are presented in Section 6 with the comparison between simulation and experimental results. Finally, Section 7 highlights the main conclusions. 2 WDM-PON Wavelength-division-multiplexed passive optical net- works (WDM-PON) offer many advantages such as large capacity, easy management, network security, and upgradability. The usage of an array waveg- uide grating (AWG) to multiplex/de-multiplex the upstream and downstream wavelengths, respectively, provides a dedicated point-to-point optical channel between each ONT and the optical line terminator (OLT), although this concept involves the sharing of a common point-to-multipoint physical architecture [4]. Since a wavelength mux/demux is used instead of an optical-power splitter, the insertion loss is con- siderably smaller and effectively independent of the splitting ratio. In addition, since the receiver band- width for each ONT is matched to its dedicated band- width, there is no additional penalty related to the number of users on the PON [4]. Consequently, the signal-to-noise ratio (SNR) is essentially independent of the number of ONTs, allowing efficient scaling and flexibility for a WDM-PON architecture, which is suited to transport multiple wireless standards in- cluding Wi-Fi. A WDM-PON scheme representation is depicted in Figure 2. In this architecture, each ONT-OLT pair is assigned a set of downstream and upstream wave- lengths. Figure 2: WDM-PON Scheme. An additional feature of a WDM architecture is its ability to localize any fault or optical loss in the fiber plant by using a single wavelength-tunable OTDR (optical time-domain reflectometer) located at the OLT. AWGs require temperature control to keep their optical channels locked to a wavelength grid. Tech- nology advances have allowed the recent commercial- ization of athermal AWGs that can remain locked to a WDM-wavelength grid over temperature ranges ex- perienced at the passive-node location [4]. 3 INTERMODULATION DISTORTION Due to the large number of electrical subcarriers of the WiFi signal, a high nonlinear distortion may be expected from the electrical to optical conversion when using direct modulated laser diodes, such as VCSELs. The interference resulting from source nonlinear- ity depends strongly on the number of channels and the distribution of channel frequencies. Considering D. Coelho et al. | i-ETC - CETC2011 Issue, Vol. 2, n. 1 (2013) ID-16 i-ETC: ISEL Academic Journal of Electronics, Telecommunications and Computers http://journals.isel.pt/index.php/IAJETC the transmission of three channels ( f1, f2 and f3), Figure 3 shows the harmonics generated by a nonlin- ear device. Amplitude Frequencyf 2 f 1 f 3 f 1 2f 1 f 3 f 1 + f 2 f 3 f 2 + f 3 f 1 2f 3 f 1 f 1 f 3 f 2 2f 1 3f 2 2f 3 f 1 + f 3 f 1 + f 2 f 2 + f 3 f 1 + f 2 + f 3 2f 1 + f 2 2f 1 + f 3 2f 3 + f 1 2f 3 + f 2 3f 1 Transmission Band Figure 3: Illustration of harmonics generated by a nonlinear optical modulator [5]. The most troublesome third-order intermodula- tion distortion products (IMPs) are those that origi- nate from frequencies fi + f j − fk and 2 fi − f j , and since they lie in within the transmission band, they lead to interchannel interference. The interference, thus, depends strongly on the number of channels and on the allocation of channel frequencies with respect to the resonance frequency of the laser. For a n chan- nels system with uniform frequency spacing the num- ber of IMPs, r IMn21 and r IM n 111 of type 2 fi − f j and fi + f j − fk, respectively, coincident with channel r are given by [6]. r IM n 21 = 1 2 { n − 2 − 1 2 [1 −(−1)n](−1)r } (1) r IMn111 = r 2 (n − r + 1)+ 1 4 [ (n − 3)2 − 5 ] − − 1 8 [1 −(−1)n](−1)n+r (2) Considering a WiFi system, the total number of channels is equals n = 64. Figure 4 shows the total number of third-order IMPs as a function of channel number. The channel in the middle of the band is the one with the large number of intermodulation prod- ucts. 4 LASER MODEL 4.1 Intrinsic Model The laser operation is described by the relationship between the carrier density N and the photon density 0 10 20 30 40 50 60 1000 1100 1200 1300 1400 1500 Channel Number N u m b e r o f I M P s Figure 4: Total number of third-order intermodulation prod- ucts for n=64. S under the presence of the injected current I. This is accomplished through a set of rate equations that explain all the mechanisms by which the carriers are generate or lost inside the active region. this set of equations is defined by [7]: dN dt = ηiI qV − N τn − g0 (N − N0m)(1 − εS)S (3) dS dt = Γg0 (N − N0m)(1 − εS)S − 1 τp S + βΓ N τn (4) The first term in (3) is the rate at which the carri- ers, electrons or holes are injected into the active layer due to current I. The second term in the equation is the loss due to various recombination process (spon- taneous and nonraditive emission) and the last term is due to the stimulated emission recombination that leads to the emission of light. The equation (4) states that the rate of increase in photon density is equal to the photon generation by stimulated emission less the loss rate of photons (as characterized by the photon lifetime, τp), plus the rate of spontaneous emission. The parameter V is the active region volume, g0 is the gain slope constant, ε is the normalized gain com- pression factor, N0m is the electron density at trans- parency, β is the fraction of the total spontaneous emission coupled at the laser mode, Γ is the optical confinement factor, ηi is the injection efficiency, τp is the photon lifetime and τn is the carrier lifetime. The optical output power can be expressed as P=ηhνS, where η = (ηdV )/(2Γηiτp), h is the Planck’s constant, ν is the emission frequency and ηd is the differential quantum efficiency. In this work, the VCSEL model FINISAIR HFE- 4192-582, operating at 850 nm, was used. The intrin- sic parameters extracted, by the frequency subtraction method [8], were then used in the simulation model (see Table 1). D. Coelho et al. | i-ETC - CETC2011 Issue, Vol. 2, n. 1 (2013) ID-16 i-ETC: ISEL Academic Journal of Electronics, Telecommunications and Computers http://journals.isel.pt/index.php/IAJETC Parameter Value Unit V 2.4x10−18 m3 g0 4.2x10−12 m3s−1 ε 2.0x10−23 m3 N0m 1.9x1024 m−3 β 1.7x10−4 − Γ 4.5x10−2 − τp 1.8 ns τn 2.6 ps ηi 0.8 − Table 1: Intrinsic parameters of FINISAIR HFE-4192-582. Since the laser dynamics described by the rate equations are intrinsically nonlinear, harmonic and in- termodulation distortion occurs during direct modula- tion which limits system performance. Modelling of the semiconductor laser diode in such a way as to ren- der tractable the accurate computation of intermodu- lation products (IMPs) is thus of importance for the design and dimensioning of such systems. 4.2 Package and Chip Parasitics Model In the laser model we will consider an intrinsic laser diode (ILD), whose dynamic behavior is described by the previous rate equations (equation 3 and 4) and a parasitic interconnection circuit due to the laser as- sembly in a package. The corresponding equivalent electrical circuit of the parasitics elements of the FIN- ISAIR HFE-4192-582 is shown in Figure 5. Figure 5: Parasitic model. Table 2 presents the intrinsic laser parameters of the VCSEL operating at 850 nm. Combining the pa- rameters obtained for the parasitic circuit and the in- trinsic laser model, the global transfer function is ob- tained as can be seen in Figure 6. The relative noise intensity characteristic of this VCSEL, was obtained from the rate equation with Langevin noise sources been represented in Figure 7. In the range of 3 to 9 mA RIN varies between -152 to -133 dB/Hz. Parameter Value Unit Rin 50 Ω RS 42.6279 Ω CS 0 ps Cp 1.8068 pF Lp1 7.6925 pF Table 2: Parasitic parameters of FINISAIR HFE-4192-582. Figure 6: Frequency response of FINISAIR HFE-4192-582. 10 0 10 1 �155 �150 �145 �140 �135 �130 �125 �120 �115 3 mA 4.5 mA Frequency (GHz) R IN (d B / H z ) 6 mA 9 mA VCSEL laser Figure 7: Relative Intensity Noise of the VCSEL. 5 ARCHITECTURE ANALYSIS The maximum fiber length of an optical network to be deployed inside an aircraft is considered to be 100 meters. Thus, it is reasonable to neglect both the attenuation and dispersion of RF signals with frequen- cies up to 10 GHz [9]. As aforementioned, WDM architecture provides a dedicated point-to-point optical channel between each ONTs and the OLT. Taking this into account let us consider the point-to-point architecture shown in the Figure 8. Since the RF signal which arrives at the an- tenna is a weak signal due to the wireless attenuation, D. Coelho et al. | i-ETC - CETC2011 Issue, Vol. 2, n. 1 (2013) ID-16 i-ETC: ISEL Academic Journal of Electronics, Telecommunications and Computers http://journals.isel.pt/index.php/IAJETC the SNR in the uplink is considerably lower than in the downlink. The RF uplink signal is generated by the Mo- bile Station and reaches the base station attenuated by the wireless channel. The weak RF uplink signal is then electrically amplified (G) before being con- verted from the electrical to the optical domain by the VCSEL. In the central station, the optical signal is de- tected by a PIN photodetector which converts the op- tical signal to the electrical before reaching the Wi-Fi receiver module (WiFi Rx). The RF signal detected will suffer the impact of the RIN noise, shot noise, photodetector thermal noise, clipping and intermodu- lation distortion. VCSEL WiFi TX PD Wireless Link Fiber P O I 0 Mobile Station G I s I Rx L WiFi RX Figure 8: Point-to-point transmission scheme. The SNR for the uplink path, referred at the output of the photodiode optical receiver, can be written as [10], SNR = ⟨I2Rx⟩ ⟨I2RIN⟩+⟨I 2 SN⟩+⟨I 2 th⟩+⟨I 2 IMD⟩+⟨I 2 cl p⟩ (5) where the five current noise terms are: the RIN noise current, the shot noise current, the thermal noise current from the equivalent resistance of the pho- todetector (PD) load and amplifier (Req), the cur- rent due to the third order intermodulation distortions and the current due clipping distortions, respectively. The source thermal noise can be neglected and ⟨I2Rx⟩ is the signal power at the receiver as described in [9][10][11]. ⟨I2Rx⟩ = 1 2 ( rd µ √ 2 N ⟨P0⟩ )2 (6) ⟨I2RIN⟩ = r 2 d ⟨P 2 0 ⟩10 RIN 10 ∆ f (7) ⟨I2SN⟩ = 2qrd ⟨P0⟩∆ f (8) ⟨I2th⟩ = 4kT F ∆ f Req (9) ⟨I2IMD⟩= (rd⟨P0⟩) 2 2 ( µ √ 2 N )6( D111N 2 + D21N ) (10) ⟨I2cl p⟩ = 1 √ 2π Λrd⟨P0⟩ µ5 1 + 6µ2 e −1 2 µ 2 (11) The rd parameter is the photodetector responsiv- ity, P0 is the average optical power detected by the PD, ∆ f is the electrical bandwidth of the receiver, q is the electronic charge (1.6 × 10−19 Coulomb), k is the Boltzmann’s constant, T = 290K, F is the noise factor of the amplifier following the PD and D111 and D21 are the third-order distortion coefficients (IMDs fi + f j − fk and 2 fi − f j ), which depend on the laser characteristics and operation point. The µ parame- ter is the total rms modulation index and is equal to µ = m √ N/2, where m is the optical modulation index per subcarrier [11]. The Λ parameter represents the fraction of the clipping distortion power which falls in the transmission band which is also dependent on the optical modulation index [12]. For the specific channel allocation, Λ = 1.1 × 10−3 for µ = 2%. 6 SIMULATIONS AND RESULTS The presented analysis considers the usage of Wi- Fi signals in the 2.4 GHz frequency range, with 20 MHz of bandwidth and the use of OFDM with 64 or- thogonal subcarriers (N = 64). Here we have assumed that the signal directly modulates a VCSEL. Volterra functional series, described as a “power series with memory”, has been applied previously to assess accurately the laser distortion of the semicon- ductor laser [13]. The latter analysis enables one to determine adequately the third-order intermodulation coefficients of the semiconductor laser, considering the allocation of subcarriers for Wi-Fi. 0 20 40 60 80 2 4 6 8 10 −35 −30 −25 −20 −15 −10 Subcarrier NumberIbias (mA) D 1 1 1 (d B ) Figure 9: D111 for several bias current. D. Coelho et al. | i-ETC - CETC2011 Issue, Vol. 2, n. 1 (2013) ID-16 i-ETC: ISEL Academic Journal of Electronics, Telecommunications and Computers http://journals.isel.pt/index.php/IAJETC 0 20 40 60 80 2 4 6 8 10 −40 −35 −30 −25 −20 −15 −10 Subcarrier NumberIbias (mA) D 2 1 (d B ) Figure 10: D21 for several bias current. The D111 coefficient has the major impact in the IMD limitation since it increases with N2, while the D21 contribution increases with N (eq. 10). Also from Figures 9 and 10, it is seen that a better performance is expected when the VCSEL is operated at 9 mA of bias current and a worst performance at 3 mA, when con- sidering the IMD impact on the system performance. For a bias current of 3 and 5 mA, D111 is maximum for the subcarrier number 34 and 35, and equals 0.0522 and 0.0057, respectively. The corresponding maxi- mum values for D21 occur for subcarrier number 63 and 64, and are 0.0081 and 0.0034, respectively. The resonance of the laser may actually change the loca- tion within the band (subcarrier) where we would ex- pect the maximum distortion to occur (middle channel for D111 and last channel for D21). The previous theoretical analysis, based on SNR, is compared with experimental results. Experimen- tally the performance of the system is assessed in terms of error vector magnitude (EVM), which relates to the SNR by [14]. EV Mrms = 1 √ SNR (12) The experimental setup used is depicted in Figure 11. It is composed of a vector signal generator (RODHE&SCHWARZ SMJ 100A) to generate the Wi-Fi signal, an electrical to optical converter (VC- SEL model FINISAIR HFE-4192-582), an optical to electrical converter (81495A) and a Digital Serial An- alyzer (Tektronics DSA 71254C) for the signal analy- sis and EVM measurements. The VCSEL laser has a slope efficiency of 0.075 W/A and a threshold current of 0.8 mA and the PIN photodetector is considered to have a responsivity of 50 A/W. Figures 12 and 13 show the results of both analyt- ical and experimental SNR as a function of the total rms modulation index, for the uplink point-to-point PIN VCSEL 850nm 0.075 W/A slope efficiency and 0.8 mA threshold current RODHE&SCHWARZ SMJ 100A Tektronics DSA 71254C 50 A/W responsivity Figure 11: Diagram of experimental setup. 10 −1 10 0 10 1 10 2 0 5 10 15 20 25 30 35 40 45 50 S N R (d B ) Total rms modulation index (%) Curves - Analytical Circles - Experimental Shot RIN 3mA Clipping IMD 3.0mA Figure 12: Analytical and experimental SNR IBias = 3 mA. 10 −1 10 0 10 1 10 2 0 5 10 15 20 25 30 35 40 45 50 55 S N R (d B ) Total rms modulation index (%) Curves - Analytical Circles - Experimental RIN 5mA Shot Clipping IMD 5.0mA Figure 13: Analytical and experimental SNR IBias = 5 mA. transmission scheme. The noise limiting contribu- tions are plotted in the graph as well. A minimum SNR of 20 dB can be specified considering a typical sensitivity from a commercial IEEE 802.11n of -74 dBm, in the 2.4 GHz band [15]. From the results we can see that the best performance in terms of SNR is achieved at high bias currents, when the intermodu- lation distortion is lower and the performance is lim- D. Coelho et al. | i-ETC - CETC2011 Issue, Vol. 2, n. 1 (2013) ID-16 i-ETC: ISEL Academic Journal of Electronics, Telecommunications and Computers http://journals.isel.pt/index.php/IAJETC ited by both clipping and IMD. The maximum SNR value for a bias current of 3 mA is 37.12 dB for a total rms modulation index of 13%, while for a bias current of 5 mA, the maximum SNR is 45.78 dB for a total rms modulation index of 11%. Measurements of SNR above 30 dB were not obtained, since the results show a tendency to reach a plateau, which indicate that as the RF input power increases, the photoreceiver’s am- plifier limits the power of the signal, which on the other hand limits the SNR. Let us consider now an analysis for a network with higher fiber length, which can be applied for other WDM-PON network architectures. Note that these new results are obtained taking only into considera- tion the optical loss due to fiber attenuation. From equation 5 and for IBias = 5 mA the SNR versus the total rms optical modulation index for several values of attenuation (α = 3, 6, 9,··· , 24 dB) is plotted in Figure 14. We can see that for low modulation in- dexes values the SNR performance is limited mainly by thermal noise except for the α = 0 dB case, where the RIN is the leading noise source. 10 −1 10 0 10 1 10 2 0 5 10 15 20 25 30 35 40 45 50 55 Total rms modulation index (%) S N R ( dB ) Optical Attenuation Increase α = 0 dB α = 24 dB Figure 14: SNR as a function of the total rms modulation index for different optical attenuation values (IBias = 5 mA). In Figure 15 it is depicted the maximum SNR values and the corresponding modulation indexes as a function of the optical attenuation. These modu- lation indexes are considered to be optimum in the SNR sense. The results indicates that the maximum optical attenuation that can be considered for an ac- ceptable minimum SNR of 20 dB is α = 23 dB (for IBias = 5 mA). By considering a multimode fiber with an attenuation of 3 dB/Km at 850 nm, it corresponds to 7.7 Km. By considering a minimum SNR of 20 dB for a reliable transmission, it is possible to determine from Figure 14 the corresponding minimum modulation in- dex (and minimum electrical power) of the RF signal that drives the VCSEL. This result gives us the mini- 0 5 10 15 20 25 15 20 25 30 35 40 45 Optical Attenuation (dB) S N R M A X ( dB ) 0 5 10 15 20 25 10 20 30 40 50 60 70 M od ul at io n in de x (% ) fo r S N R M A X SNR limit Figure 15: Maximum SNR and optimum modulation in- dex values as a function of the optical attenuation (IBias = 5 mA). mum RF power that can be used to directly modulate the VCSEL for an SNR of 20 dB. Figure 16 shows this result as a function of the optical attenuation. 0 5 10 15 20 25 0 5 10 15 20 25 30 m in im um m od ul at io n in de x (% ) fo r S N R = 2 0 dB Optical attenuation (dB) Figure 16: Minimum modulation index value to achieve an acceptable SNR of 20 dB as a function of the optical atten- uation (IBias = 5 mA). Assuming an RF signal with power given by PRF = 1 2 ⟨I2⟩Ri (13) where Ri is the input resistance of the VCSEL consid- ered to be 50 Ω, the relation between RF power and modulation index is plotted in Figure 17. Considering a maximum amplifier gain (G) of 30 dB and assume that the antenna noise from the ONT does not severally affect the SNR of the signal, the RF power at the input of the VCSEL is given by: Pin = PT X,max − L + G + 4 (14) where PT X,max = 13 dBm is the maximum transmitter power defined by the standard, L is the wireless FSPL D. Coelho et al. | i-ETC - CETC2011 Issue, Vol. 2, n. 1 (2013) ID-16 i-ETC: ISEL Academic Journal of Electronics, Telecommunications and Computers http://journals.isel.pt/index.php/IAJETC 0 20 40 60 80 100 −45 −40 −35 −30 −25 −20 −15 −10 −5 0 total rms modulation index (%) P ow er ( dB m ) Figure 17: RF power as a function of the modulation index (IBias = 5 mA). (Free Space Path Loss) defined by: F SPL(dB)=20log10(d)+20log10( f )−147.55 (15) where d is the wireless link length in meters and f is the RF signal frequency in Hertz. Additionally, the maximum wireless link length as a function of the optical link attenuation and optical link length for an SNR = 20 dB necessary for a re- liable transmission are plotted in Figures 18 and 19, respectively. 0 5 10 15 20 25 0 50 100 150 200 250 300 350 400 450 500 Optical Attenuation (dB) M ax im um W ir el es s ch an ne l le ng th ( m ) Figure 18: Maximum wireless link length as a function of the optical attenuation for a reliable transmission. As aforementioned, the maximum fiber length of an optical network to be deployed inside an aircraft is considered to be 100 meters (α = 0 dB case). Looking at the results, for a reliable Wi-Fi trans- mission, the system could be deployed with just one access point (or ONT), however if another services (WiMAX, UWB, etc) were integrated there would be necessity to increase the number of ONTs to achieve its SNR requirements. 0 1 2 3 4 5 6 7 0 50 100 150 200 250 300 350 400 450 500 Optical fiber length in km (considering only α = 3 dB/Km) M ax im um W ir el es s ch an ne l le ng th ( m ) Figure 19: Maximum wireless link length as a function of the optical fiber length for a reliable transmission. 7 CONCLUSION In this article, we have considered the transmission of WiFi signals through an optical channel. In particular, we analyze the uplink performance in a point-to-point transmission system for short range networks.A theo- retical analysis was performed and a good agreement with the experimental results was obtained. Both analytical and experimental results show that, for low bias currents, the intermodulation distor- tion is the main limiting performance factor at high modulation indexes, whereas the RIN is the dominant factor for low modulation indexes. For increasing bias current, the IMD distortion decreases and clip- ping distortion starts to dominate over intermodula- tion distortion, at high modulation indexes. For a maximum fiber length of 100 meters (air- craft network case), the best performance in terms of SNR is achieved at high bias currents. The maximum SNR value for a bias current of 3 mA is 37.12 dB for a total rms modulation index of 13%, while for a bias current of 5 mA, the maximum SNR is 45.78 dB for a total rms modulation index of 11%. For higher fiber length WDM-PON networks, the maximum optical attenuation that can be considered for an acceptable minimum SNR of 20 dB is α = 23 dB (for IBias = 5 mA), which correspond a fiber length of approximately 7.7 km. The detailed analysis and system performance presented is adequate for the assessment and design of radio-over-fiber systems in aircraft networks. D. Coelho et al. | i-ETC - CETC2011 Issue, Vol. 2, n. 1 (2013) ID-16 i-ETC: ISEL Academic Journal of Electronics, Telecommunications and Computers http://journals.isel.pt/index.php/IAJETC REFERENCES [1] INESCPorto, “Report on terrestrial networks and ar- chitectures,” tech. rep., Daphne project, 2009. [2] D. Qian, J. Hu, P. Ji, T. Wang, and M. Cvijetic, “10-gb/s ofdma-pon for delivery of heterogeneous services,” in Optical Fiber Communication Con- ference/National Fiber Optic Engineers Conference, p. OWH4, Optical Society of America, 2008. [3] J. M. B. OLiveira, S. Silva, L. Pessoa, D. Coelho, H. Salgado, and J. C. S. Castro, “Uwb radio over per- fluorinated gi-pof for low-cost in-building networks,” in Microwave Photonics (MWP), 2010 IEEE Topical Meeting on, pp. 317–320, 2010. [4] C.-H. Lee, W. Sorin, and B.-Y. Kim, “Fiber to the home using a pon infrastructure,” Lightwave Technol- ogy, Journal of, vol. 24, no. 12, pp. 4568–4583, 2006. [5] J. M. B. de Oliveira, Receiver Design for Nonlinearly Distorted OFDM Signals: Applications in Radio- over-Fiber Systems. PhD thesis, Faculdade de Engen- haria da Universidade do Porto, 2012. [6] R. Westcott, “Investigation of multiple f.m./f.d.m. car- riers through a satellite t.w.t. operating near to satura- tion,” Electrical Engineers, Proceedings of the Insti- tution of, vol. 114, no. 6, pp. 726–740, 1967. [7] H. Salgado, Performance Assessment of Subcarrier Multiplexed Optical Systems: Implication of Laser Nonlinearities. PhD thesis, University of Wales, 1993. [8] P. Morton, T. Tanbun-Ek, R. Logan, A. Sergent, P. Sciortino, and D. Coblentz, “Frequency response subtraction for simple measurement of intrinsic laser dynamic properties,” Photonics Technology Letters, IEEE, vol. 4, no. 2, pp. 133–136, 1992. [9] I. Cox, C.H., E. Ackerman, G. Betts, and J. Prince, “Limits on the performance of rf-over-fiber links and their impact on device design,” Microwave Theory and Techniques, IEEE Transactions on, vol. 54, no. 2, pp. 906–920, 2006. [10] C. H. CoxIII, Analog Optical Links: Theory and Prac- tice. Cambridge University Press, 2004. [11] A. Saleh, “Fundamental limit on number of chan- nels in subcarrier-multiplexed lightwave catv system,” Electronics Letters, vol. 25, no. 12, pp. 776–777, 1989. [12] J. Mazo, “Asymptotic distortion spectrum of clipped, dc-biased, gaussian noise [optical communication],” Communications, IEEE Transactions on, vol. 40, no. 8, pp. 1339–1344, 1992. [13] H. Salgado, “Modelling optical/rf impairments in ultra-wideband radio over fiber,” tech. rep., INESC Porto, 2007. [14] R. Shafik, S. Rahman, and R. Islam, “On the ex- tended relationships among evm, ber and snr as per- formance metrics,” in Electrical and Computer Engi- neering, 2006. ICECE ’06. International Conference on, pp. 408–411, 2006. [15] W. N. Corporation: Product Specifications of DNMA- 92, An IEEE 802.11n a/b/g/n Mini-PCI module, ver- sion 1.6, Apr. 2009. D. Coelho et al. | i-ETC - CETC2011 Issue, Vol. 2, n. 1 (2013) ID-16 i-ETC: ISEL Academic Journal of Electronics, Telecommunications and Computers http://journals.isel.pt/index.php/IAJETC