Modulation and Detection Techniques for Optical Communication Systems

Joseph M. Kahn
Stanford University, Department of Electrical Engineering, 372 Packard Building, 350 Serra Mall, Stanford, CA 94305-9515 USA [email protected]

Abstract: Performance and implementation complexity of various binary and nonbinary modulation methods with coherent, differentially coherent and noncoherent detection are compared. Nonbinary modulation with coherent detection maximizes spectral efficiency and improves tolerance to transmission impairments, while enabling effective, low-complexity electrical compensation of these impairments. 1. Introduction Currently deployed fiber and free-space optical communication systems use on-off keying (OOK) with direct detection, and some are beginning to use differential phase-shift keying (DPSK) with interferometric detection. The further evolution of modulation and detection techniques will increase spectral efficiency in wavelength-divisionmultiplexed (WDM) systems, improve robustness against transmission impairments, and facilitate electrical compensation of such impairments. As explained in this paper, nonbinary modulation with coherent detection yields the best performance in all of the above respects, but at the cost of increased implementation complexity. Throughout this paper, we consider fiber or free-space systems that use optical amplifiers and/or nonlinear optical wavelength converters, and assume that that amplified spontaneous emission (ASE) is the dominant noise source. We follow the notation used in [1], to which the reader is referred for further details and references. 2. Detection and Modulation Techniques Detection methods for digitally modulated optical signals are classified most logically using traditional distinctions [2], as given in Table 1. In noncoherent detection, only the presence or absence of energy is ascertained, and no phase information is recovered. We consider noncoherent detection of M-level intensity modulation, referred to as M-ary pulse-amplitude modulation (M-PAM). OOK is the special case of 2-PAM. In differentially coherent detection, the phase in one symbol is compared to the phase in the previous symbol. We consider differentially coherent detection of M-ary differential phase-shift keying (M-DPSK). In coherent detection, a signal is detected using a carrier phase reference generated at the receiver. We consider coherent detection of M-ary phase-shift keying (M-PSK) or M-ary quadrature-amplitude modulation (M-QAM). Any of the three detection methods can be implemented using heterodyne or homodyne downconversion by a local-oscillator (LO) laser and balanced optical receiver(s), followed by the appropriate electrical-domain detector. But noncoherent detection can be implemented at lower complexity using simple direct detection, while differentially coherent detection can be implemented at lower complexity using interferometer(s) in conjunction with balanced optical receiver(s). Under our assumption that ASE dominates, these simpler implementations achieve the same performance as heterodyne or homodyne [3].
Table 1. Classification of detection techniques.

Detection Technique Noncoherent Differentially coherent Coherent

Heterodyne or Homodyne Implementation Envelope detection Delay-and-multiply detection Coherent detection

Direct Detection Implementation Direct detection Interferometric detection

Modulation Techniques Considered PAM DPSK PSK, QAM

Class of Technique Asynchronous Synchronous

Fig. 1 compares the spectral efficiencies and receiver sensitivities of various modulation and detection methods. Achievable spectral efficiency is proportional to log2(M), the number of bits encoded per symbol. The receiver sensitivity nb/neq is the average number of photons per bit nb, divided by the equivalent ASE noise factor of the optical amplifier chain neq [1]. Binary modulation can achieve spectral efficiency up to 1/b/s/Hz (per polarization). 2-DPSK with interferometric detection is an attractive scheme for its excellent sensitivity and relatively low implementation complexity. Quaternary modulation can double spectral efficiency, while achieving higher tolerance to impairments, such as fiber chromatic dispersion (CD) and polarization-mode dispersion (PMD). 4-DPSK with interferometric detection is attractive for its reasonable sensitivity and complexity, while 4-PSK with coherent

detection offers better sensitivity. Going to still higher spectral efficiencies (M > 4), QAM with coherent detection offers the best sensitivity among the various schemes. The superior performance of QAM can be traced to the fact that QAM encodes information in two degrees of freedom (two quadrature phases), while PSK, DPSK or PAM encode information in only one degree of freedom (phase, phase or magnitude).
Receiver Sensitivity nb/neq Required for Pb = 109 (photons/bit) 20 50 100 200 500 1000 2000 16

4 Relative Spectral Efficiency log2(M) (b/symbol)

2 QAM / Coherent PSK / Coherent DPSK / Differentially Coherent PAM / Noncoherent 1 12 18 15 21 24 27 30 SNR/bit GEb/Seq Required for Pb = 109 (dB) 33

Fig. 1. Spectral efficiency vs. required receiver sensitivity (or SNR/bit) of various modulation/detection techniques.

Table 2 compares the three detection techniques, including maximum number of degrees of freedom and receiver sensitivities for binary and quaternary modulations. Noncoherent or differentially coherent detection can be implemented using direct detection, avoiding the need for a LO laser and polarization control or diversity. Any of the three techniques can be implemented using heterodyne or homodyne downconversion, yielding two potential advantages. First, by using a fast-tunable LO laser, one can construct a frequency-agile receiver to enable wavelength-routed switching or frequency-hopped transmission. Second, fiber CD appears as a linear distortion, facilitating its effective, low-complexity compensation in the electrical domain.
Table 2. Comparison of detection techniques. Shading denotes an advantage. Noncoherent and differentially coherent detection assume singlepolarization filtering at the receiver. Receiver sensitivities are the values of nb/neq required for Pb = 109.

Attribute Maximum number of degrees of freedom (per polarization) Receiver sensitivity for binary Receiver sensitivity for quaternary

Noncoherent 1 (magnitude) 38 photons/bit (2-PAM) 134 photons/bit (4-PAM) Heterodyne / Homodyne Direct Detection No No No No

Differentially Coherent 1 (phase) 20 photons/bit (2-DPSK) 31 photons/bit (4-DPSK) Heterodyne / Homodyne Yes Yes Yes Yes Direct Detection No No No No

Number of Constellation Points M

Coherent 2 (two quadrature components) 18 photons/bit (2-PSK) 18 photons/bit (4-PSK) Heterodyne / Homodyne Yes Yes Yes Yes

Electrical filtering can be used to select channel (enables frequency-agile receiver) Chromatic dispersion is linear distortion (enables effective electrical compensation) Local oscillator laser required at receiver Polarization control or diversity required at receiver

Yes Yes Yes Yes

3. Implementation Issues Table 3 compares homodyne to heterodyne downconversion. Homodyne requires the optical receiver to have an electrical bandwidth of the order of the symbol rate Rs, while heterodyne requires a bandwidth of about 2Rs, which can be prohibitive at high symbol rates. For many applications, it is necessary to downconvert two quadrature phases of an optical signal: (a) demodulation of quadrature modulation, (b) synchronous demodulation in the electrical

domain (as opposed to using an optical PLL), (c) asynchronous demodulation in the electrical domain, (d) electrical compensation of CD in fiber, or (e) electrical compensation of nonlinear phase noise (NLPN) in fiber. In such cases, homodyne requires two balanced optical receivers, while heterodyne requires only one. Heterodyne downconversion requires narrowband filtering or image rejection to avoid image-band interference (in WDM systems) and a 3-dB noise penalty from image-band ASE [4], while homodyne avoids these complications.
Table 3. Comparison of homodyne and heterodyne and downconversion. Shading denotes an advantage. Rs denotes the symbol rate.

Attribute Electrical bandwidth requirement of balanced optical receiver Number of balanced optical receivers for required (per polarization) for quadrature detection Narrowband filtering or image rejection required to avoid image-band interference and noise

Homodyne ~Rs 2 No

Heterodyne ~2Rs 1 Yes

Table 4 compares the laser linewidths required to implement various modulation and detection methods. While 2- and 4-DPSK and 2-PSK can be implemented using typical distributed-feedback lasers, other modulation/detection methods require narrower-linewidth lasers [11].
Table 4. Laser linewidth requirements for various modulation and detection techniques, assuming a 0.5 dB penalty. Differentially coherent detection assumes an interferometric implementation, so the transmitter laser has linewidth , while for coherent detection, each of the transmitter and local oscillator lasers has linewidth .

Modulation 2-DPSK 4-DPSK 2-PSK 4-PSK 8-PSK 16-PSK 8-QAM 16-QAM

Detection Differentially coherent Differentially coherent Coherent Coherent Coherent Coherent Coherent Coherent

/Rb 3.0 103 5.0 10 8.0 10

for Rb = 10 Gb/s 30 MHz 5 MHz 8 MHz 250 kHz 15 kHz 2.4 kHz 90 kHz 6.9 kHz

Reference [5] [6] [7] [8] [9] [9] [10] [10]

2.5 105 1.5 106 2.4 10 9.0 10 6.9 10

Electrical signal processing is useful for carrier phase synchronization, compensation of CD, PMD and NLPN in fiber, or compensation of atmospheric turbulence and Doppler shifts in free-space links. Some operations, such as carrier phase synchronization, can be done easily in the analog domain. But over time, we expect to see increased use of analog-to-digital conversion and digital signal processing (DSP). DSP-based processing is particularly useful to track time-varying perturbations, or when it is necessary to compensate multiple perturbations simultaneously. 4. References
J. M. Kahn and K.-P. Ho, Spectral Efficiency Limits and Modulation/Detection Techniques for DWDM Systems, IEEE. J. on Sel. Topics in Quantum Electron. 10, 259-272 (2004). 2. J. G. Proakis, Digital Communications, 4th Ed., McGraw-Hill (2000). 3. O. K. Tonguz and R. E. Wagner, Equivalence Between Preamplified Direct Detection and Heterodyne Receivers, IEEE Photon. Technol. Lett. 3, 835-837 (1991). 4. B. F. Jorgensen, B. Mikkelsen and C. J. Mahon, Analysis of Optical Amplifier Noise in Coherent Optical Communication Systems with Optical Image Rejection Receivers, J. of Lightwave Technol. 10, 660-671 (1992). 5. C. P. Kaiser, P. J. Smith and M. Shafi, An Improved Optical Heterodyne DPSK Receiver to Combat Laser Phase Noise, J. Lightwave Technol. 13, 525-533 (1995). 6. S. Savory and A. Hadjifotiou, Laser Linewidth Requirements for Optical DQPSK Systems, IEEE Photon. Technol. Lett. 16, 930-932 (2004). 7. S. Norimatsu and K. Iwashita, Linewidth Requirements for Optical Synchronous Detection Systems with Nonnegligible Loop Delay Time, J. Lightwave Technol. 10, 341-349 (1992). 8. J. R. Barry and J. M. Kahn, Carrier Synchronization for Homodyne and Heterodyne Detection of Optical Quadriphase-Shift Keying, J. Lightwave Technol. 10, 1939-1951 (1992). 9. E. Ip and J. M. Kahn, unpublished (2005). 10. E. Ip and J. M. Kahn, Carrier Synchronization for 3- and 4-bit-per-Symbol Optical Transmission, J. of Lightwave Technol. 23, 4110-4124 (2005). 11. E. Ip, J. M. Kahn, D. Anthon and J. Hutchins, Linewidth Measurements of MEMS-Based Tunable Lasers for Phase-Locking Applications, IEEE Photon. Technol. Lett. 17, 2029-2031 (2005). 1.