I. INTRODUCTION
The unlicensed millimeter band at 60 GHz is nowadays a solution to offer multi-gigabits/s wireless data [Reference Park and Rappaport1]. However, communication distances are limited to few meters due to walls, obstacles, and oxygen absorption [Reference Collonge, Zaharia and El Zein2]. These short coverage ranges can be extended by means of radio-over-fiber (RoF) technology [Reference Guillory3, Reference Gomes4]. This solution offers the advantage of the high bandwidth of optical fibers and removes signal-processing functions in transceivers. Various RoF systems were proposed to cancel local oscillators in receivers by optical generation of millimeter-wave signal making thus an expensive solution [Reference Lecoche5]. Another alternative is direct detection followed by complementary metal oxide semiconductor (CMOS) circuits which can be more competitive solution for domestic applications [Reference Galal and Razavi6]. The data transmission through the optical fiber can be realized by direct modulation (DM) at an intermediate frequency (IF) of lasers, electro-absorption modulated lasers (EMLs) or Mach–Zehnder modulator (MZM). DM of distributed feedback (DFB) lasers is a simple technique but it has a bandwidth limitation. The EML is attractive in short-reach networks because it offers an easy integration, low-driving voltage, and power dissipation [Reference Ido, Sano, Moss, Tanaka and Takai7]. The MZM offers low chirp and higher performances that is why it is widely used in the long-haul transmission system [Reference Kamisaka, Kuri and Kitayama8]. However, the nonlinear effects induced by optical modulators cause intermodulation distortion and result in signal degradation.
In this paper, we present performance analysis of orthogonal frequency-division multiplexing (OFDM) signal transmission over two RoF links: DM of a DFB laser and external modulation (EM) with an electro absorption modulator. The OFDM signal matches with the specification given by the IEEE802.15.3c standard for the ultra wideband (UWB) millimeter-band. System simulation is performed with the developed electrical circuit models of photonic and optical components [Reference Billabert9, Reference Deshours, Billabert, Algani and Blache10]. These models have realistic characteristics and include noise sources and nonlinearities. To validate the simulation, we used a simple measurement setup having a direct solution to perform the error vector magnitude (EVM) of the first band (3.1–4.6 GHz) of UWB centimeter-band standard (ECMA-368). The EVM is a good performance metric to analyze degradation of such complex signal [Reference Schmogrow11]. Then, signal broadband and data rate effects on the EVM are also discussed.
II. ROF SYSTEMS MODELING
The photonic components of the RoF systems are modeled with equivalent electrical circuits by using an analogy between the optical power and a current or voltage. As the optical phase noise is eliminated in the direct detection, the power consideration of these models is appropriate to represent the optical signal. The fiber is a standard single-mode fiber with a length lower than 100 m which is typical for home area network applications. Chromatic dispersion effect and attenuation are incorporated in the model but the influence in the system performance is neglected. For the optical/electrical (O/E) part, the photodiode model includes noise and is linear due to the case here where the received optical power is much lower than its saturation level. With the realistic characteristics of E/O and O/E modules, gain compression, intermodulation distortions, and noise power of the RoF links can then be investigated by simulation.
A) E/O and O/E transducers models
The DFB laser is from 3S Photonics Company having a wavelength of a 1.56 μm, a modulation bandwidth of 16.8 GHz and an efficiency of 0.087 W/A. The EML is from III–V Laboratory and integrates a 1.53 μm DFB laser with a threshold current of 10 mA and a slope efficiency of 0.16 W/A, and an electro-absorption modulator. The electrical model of DFB lasers is based on two rate equations for carriers and photons including Langevin noise sources to consider the random nature of carrier recombination and photon generation [Reference Billabert9]. The physical parameters of the laser equivalent circuit are obtained from the known data of its heterostructure layers and some of system parameters extracted from measurements of static curve and relative intensity noise (RIN). A matching circuit and parasitic elements due to a laser packaging are also added to the model. The EML circuit is formed by the DFB laser model and Electro-Absorption Modulator (EAM) model constituting of a single RC low-pass filter and nonlinearities behavior of the electro-optical effect describing the dependence of the optical output power with the reverse bias voltage [Reference Deshours, Billabert, Algani and Blache10].
The measured and simulated characteristics of the static curves and RIN of these devices are presented in Fig. 1. The RIN values of both DFB lasers are almost equal in the working frequency range from 2 to 5 GHz. The same photodetector is used for both RoF systems. It includes a PIN photodiode with a 3 dB bandwidth of 20 GHz and a responsivity of 0.83 A/W followed by a low-noise amplifier (LNA). The photodiode model includes a current source for optical to electrical conversion, a resistor–capacitor (RC) circuit for the frequency response and packaging parasitic elements. This model considers the thermal noise and the shot noise linked to the photodetected. The LNA with a gain of 33 dB and a noise figure of 3.5 dB has a flat response which does not introduce amplitude and phase distortions in the working frequency bandwidth.
Fig. 1. Measured and simulated static response and laser RIN with DM (left) and EM (right).
B) Noise power and compression point
The thermal noise, laser RIN and shot noise are the different noise sources in the RoF link. At the output of the LNA, the total noise power NP out of the RoF systems is given by (1).
$$\eqalign{& N{P_{out}} = {({(kT\,BG_{RoF}} + kT\,B)_{th}} \cr & \quad\quad\quad + {(RIN\,I_{ph}^2 {R_l}B)_{RIN}} + {(2q\,{I_{ph}}{R_l}B)_{shot}})_{RoF} {G_{LNA}}\,\cr & \quad\quad\quad + {(kT\,B(NF_{LNA}} - 1){G_{LNA}})_{LNA},}$$where indices (th, RIN, and shot) refer to thermal, laser, and shot noises, k the Boltzmann's constant, T the standard temperature in Kelvin, B the noise bandwidth, G RoF the link gain, I ph the photodetected DC current, q the electron charge and R l the load impedance, and G LNA and NF LNA are the LNA available gain and noise figures. For B equals to 1 Hz, the noise calculations at the LNA output at a frequency of 4 GHz are represented in Fig. 2. For DM, noise powers estimated with a laser bias current from 30 to 80 mA show a dominant laser noise for low bias current and both laser and shot noises when bias current increase. For EM, noise powers are calculated for EAM bias voltage from −3.2 to −1 V and for a laser bias current of 50 mA. Laser and shot noises decrease when the EAM absorbs strongly the light due to the low photodetected current levels.
Fig. 2. Calculated noise powers with DM (left) and EM (right). G RoF, DM = [−24, −30, −31] dB for [30, 50, 80] mA, G RoF, EM = [−48, −49.5, −59.5] dB for [−2, −2.4, −2.8] V, G LNA = 33 dB and NF LNA = 3.5 dB.
The 1 dB compression point (P 1 dB) depends on the bias conditions as shown in Fig. 3. Measured and simulated input power compression point (P in,1 dB) at a carrier frequency of 4 GHz without the LNA for DM increases with bias current and has a value of −13, 14.5, and 21 dBm for 30, 50, and 80 mA, respectively. For the external modulation, P in,1 dB increases with the EAM bias voltage and is equal to 13 and 18.9 dBm for −2 and −2.6 V, respectively.
Fig. 3. Compression points of the RoF links with DM (left) and EM (right).
III. ROF SYSTEM PERFORMANCES
The transmission efficiency of an OFDM signal is evaluated by the EVM from the received constellation and the ideal one. The EVM of UWB centimeter-band standard is measured and simulated for the first band group with a Time Frequency Codes number 1 following the ECMA-368 standard. This standard exhibits a signal bandwidth of 528 MHz for each sub-band with total subcarriers of 128 and a data rate up to 480 Mb/s. For the UWB millimeter-band, the IEEE 802.15.3c standard is used which deals with a 512 total sub-carriers, a bandwidth of 1815 MHz and a data rate up to 6.16 Gb/s. The schematic blocks of RoF systems are described in Fig. 4. The measurement was performed only for the centimeter-band due to the setup simplicity. The OFDM signal is generated directly at an IF by an arbitrary wave generator (AWG). A variable attenuator and a broadband Radio Frequency amplifier were used to control the OFDM signal power. The EVM was calculated with a digital storage oscilloscope having a vector signal analysis solution. The simulation is done within Advanced Design Systems (ADS) with the developed photonics models and with the existing modules of generation and processing of OFDM signal for the UWB centimeter-band. For the millimeter-band, ADS interface with Matlab is used for these functionalities.
Fig. 4. Schematic blocks for OFDM signal transmission over fiber systems.
The simulation is performed by envelope simulation for analog signal and data flow controller for digital part. Amplifiers were also characterized in frequency, noise and nonlinearities, and implemented with ADS modules for simulations.
A) UWB centimeter-band standard
The band group 1 of the UWB OFDM spectrum is used with a data rate of 480 Mb/s and a 16QAM constellation. The spectrum is thus centered at a frequency of 3.96 GHz and constituted of three bands of 528 MHz. In this case, only one sub-band modulates the laser or EML. Before evaluation of the EVM with the photonic links, the reference EVM was determined for a back-to-back connection (B-to-B) consisting of the attenuator and input amplifier. This EVM is reported in Fig. 5 for measured OFDM power (P OFDM) from −20 to 16 dBm (at the amplifier output). For power levels higher than 10 dBm, the EVM increases due to the amplifier gain compression. The realistic parameters of these attenuator and amplifier are used for simulation giving thus the same measured EVM values.
Fig. 5. Measured and simulated EVM versus P OFDM for the B-to-B connection, DM (left) and EM (right) RoF systems – UWB OFDM signal of the centimeter band according to ECMA-368 standard.
The EVM evaluations with the photonic links are represented in Fig. 5 and show a good agreement between measurement and simulation. These EVM are the average values of the three sub-bands. For the direct modulation, the results indicate a strong dependence of the EVM with laser bias current. A minimum EVM value of 17% is obtained for a bias current of 30 mA at P OFDM of −14 dBm. This EVM value increases for lower P OFDM due to the contribution of the different noise sources. As the DFB laser is biased near the threshold value, the optical signal is clipped causing distortion and EVM increase for higher P OFDM. For bias currents of 50 and 80 mA, EVMs around 3% are obtained over 10 dB range power. The EVM increasing for P OFDM below −10 dBm is related to the noise and for P OFDM above 5 dBm is caused by the system compression. However, the input amplifier has an output P 1 dB of 17 dBm but its contribution on EVM appears rather linked to a peak-to-average power ratio (PAPR) of the OFDM signal. As the amplifier P 1 dB value is close to the RoF compression point for a bias current of 50 mA, the EVM degradation is more important due to the common contribution of the amplifier and RoF link nonlinearities.
For the external modulation, the EVM is estimated as a function of P OFDM for a bias current of 50 mA and bias voltages of −2 and −2.4 V. For P OFDM higher than 5 dBm, the EVM increases due to the RF amplifier P 1 dB. For lowers P OFDM, the EVM increases as much the modulator absorbs the light. Furthermore, increasing the bias current of the EML like in DM can enhance the dynamic range.
B) UWB millimeter-band standard
For the UWB millimeter band, the OFDM signal with the same constellation as before is used but with an analog bandwidth of 1.815 GHz and a data rate of 6.16 Gb/s. The central frequency of the OFDM signal is chosen at 4 GHz. Figure 6 reports the EVM values obtained by simulation for direct and external modulations for different bias conditions. It is shown a similar results than those obtained for a centimeter band standard. The difference is due to the AWG noise, which is not considered here, giving thus a lower value of the minimum EVM.
Fig. 6. Simulated EVM and constellations versus P OFDM for DM (left) and EM (right) RoF systems with UWB millimeter-band following IEEE 802.15.3c standard.
C) Discussions
The RoF links with DFB laser and EML can present closest performances depending on the bias conditions of the transducers and noise contribution. For direct modulation, the laser RIN and shot noise are the main noise contributions when the laser is biased far from its threshold. As the link gain for EM is low due to the EML attenuation and the insertion losses between the laser and electro-absorption modulator, the principal noise degradation is due to the receiver electronic circuits before −2 V. The use of electronic circuits with better performances permits to enhance the performances of RoF system with EML [Reference Kangbaek, Dong-Soo, Yong-Duck, Kwang-Seong and Jae-Sik12].
The increase of the signal bandwidth and data rate does not affect the EVM of the OFDM signal. This is due to the closed signal PAPR value which is equal to 9 for the centimeter-band standard and 11 for the millimeter-band standard. Then, the nonlinearities appear at the same order of power levels.
IV. WIRELESS CHANNEL EFFECT
The simulation setup is completed by the 60 GHz radio channel for indoor applications. It is characterized by high free-space attenuation and high-energy absorption by walls and obstacles. Thus, the number of significant reflection paths is reduced and the structure of such channel is composed principally by the direct path line-of-sight (LOS) and some low-order reflection paths where first and second orders are generally considered [Reference Maltsev, Maslennikov, Sevastyanov, Lomayev and Khoryaev13]. This clustering approach is based on Saleh–Valenzuela model with coefficients given in [Reference Emami14]. Two scenarios of the office LOS channel configuration are considered here. The difference between these scenarios is the Tx/Rx antenna orientations, which are 30°/30° for the channel 1 and 60°/60° for channel 2.
After photodetection, the OFDM signal is upconverted to 60 GHz band with ideal circuits. A distance of 5 m is used between antennas and only thermal noise is considered in the front-end receiver. Simulation results for the two scenarios are presented in Fig. 7 and show an increase of the EVM minimum. For channel 1, most of the transmission power is spread between the transmitter and the receiver through the LOS path. While for channel 2, the transmitted power is scattered on higher number of secondary Non-Line-Of-Sight paths, and hence more interferences are presents.
Fig. 7. Simulated EVM with two wireless paths for DM with a bias current of 50 mA (left) and EM with a bias voltage of −2 V (right).
V. CONCLUSION
The developed photonic components coupling to ADS analog/digital co-simulation method are used to analyze performances of RoF to transmit OFDM signals. Modulation of DFB laser and EML with UWB centimeter and millimeter standards is compared by EVM evaluation as function of the signal power. It is shown that both modulation schemes can give the same EVM values depending on the biasing conditions regardless on the dominant noise sources. RIN and shot noise are the contributor sources for DM and the noise receiver front-end electronic circuits for EM. The increase of OFDM signal bandwidth from 528 to 1815 MHz has no effect on the EVM which is due to the closest PAPR value of these signals.
ACKNOWLEDGEMENT
The authors gratefully acknowledge Y. Chatelon from L2E, J-L. Polleux and C. Viana from ESYCOM ESIEE Paris and F. Van Dijk from III-V Lab for their measurement helps.
Ali Kabalan received his engineering degree in Telecommunications Engineering from the University of Aleppo, Aleppo, Syria, in 2009, and the Research Master (master Phot-IN) from the Ecole Nationale d'Ingénieurs de Brest, Brest, France, in 2012. He is presently working toward his Ph.D. degree at Conservatoire National des Arts et Métiers, Laboratoire d'Electronique, Systèmes De Communication Et Microsystèmes, Paris, France. His research interests include radio-over-fiber with analog and digital optical transmission systems, photonic processing, and optical fiber modeling.
Salim Faci received his Dipl. Ing. in Electronics from University of Mouloud Mammeri of Tizi-Ouzou in 2001 and the M.D. in Electronics from University of Pierre et Marie Curie (UPMC) in 2003. He received his Ph.D. degree in Electronics, Optronics, and Systems from UPMC in 2007. His doctoral research is focused on photoconductive switches modeling and optical control of microwave oscillators. He is presently in the Electronic, Communication Systems and Microsystems Laboratory (Esycom) and his research has focused on microwave photonics, which includes RoF systems, photonic generation of mm-wave signals, optoelectronic devices modeling.
Anne-Laure Billabert received her Engineering degree in Electronics from IRESTE, University of Nantes, France, and the DEA degree in electronics and radar from the University of Nantes, both in 1995. She received her Ph.D. degree in Radar Polarimetry in 1999 from the University of Nantes. Since 1999, she has joined the ESYCOM laboratory and Cnam, Paris, France, as a Lecturer. Her present researches are centered in the topics of opto-microwave, mainly the simulation of opto-microwave link. She is responsible for the French microwave photonic club of the Société Française d'optique.
Frédérique Deshours received her Ph.D. degree in Electronic and Telecommunications from both the Université Pierre et Marie Curie (UPMC), Paris, France, and Telecom ParisTech, Paris, in 1996. Since 1997, she has been an Assistant Professor at the Electronic and Electromagnetic Laboratory (L2E), UPMC. She has been involved in the simulation, design, and measurement of microwave and millimeter-wave integrated circuits. Her present research interests include the modeling of high-speed electro absorption modulators for radio-over-fiber communications and the development of millimeter-wave architectures of ultrawideband transmission systems with optical link.
Catherine Algani received her DEA degree in Electronics, and her Ph.D. degree from the University of Paris 6, France, respectively, in 1987 and 1990. Her dissertation concerns the area of active MMIC design using GaAs HBTs Technology in CNET-Bagneux. In 1991, she joined the Electronics Engineering Department and the LISIF Laboratory, at the University of Paris 6, as a Lecturer. From 1991 to 2005, she worked on the design of microwave and millimeter-wave integrated circuits on different GaAs technologies. In 1997, she began to work in the area of microwave photonics (optically controlled microwave switches on GaAs and electro-optic organic modulator). In 2005, she joined ESYCOM at CNAM-Paris, where she is presently a full Professor. Her present research interests are the development of devices, circuits, and sub-systems for ultrahigh speed digital and analog communications for ROF and wireless applications. These researches include the modeling, the design, and the characterization of such structures.
