Open Access
Add to BibSonomyAbstract
Two optical transmitters are described, generating new modulation formats based on the simultaneous modulation of the amplitude and the phase of an optical signal. The proposed formats are compared to the traditional Non-Return-to-Zero (NRZ), showing improved tolerances to chromatic dispersion (CD), differential group delay (DGD) and to wavelength division multiplexing (WDM) channel spacing, and requiring electronics with halved bandwidth.
©2005 Optical Society of America
1. Introduction
Orthogonal ASK-DPSK signals [1–3] are characterized by the optical multiplexing of independent amplitude and phase codes into the same optical signal; both the optical power and the phase are used to carry information, doubling the transport capacity and increasing the overall spectral efficiency, or equivalently reducing the required bandwidth for the opto-electronics at the transmitter and the receiver. Key features for the success of such technologies are the tolerances to linear and to nonlinear optical effects, the receiver sensitivity, the reachable signal extinction ratio, the number of modulators required at the transmitter, the electronics bandwidth as well as the maximum spectral efficiency in WDM transmission.
Dark pulse and bright pulse Phase and Intensity Modulation (PhIM) are two new, patent pending modulation formats [4] belonging to the class of orthogonal ASK-DPSK signals. In both cases an amplitude and a phase tributary are independently modulated in a same optical signal, doubling the transport capacity with a modest increase in the optical bandwidth, and with no need to reduce the modulators extinction ratio.
For pulsed PhIM, the receiver is composed by a common intensity and a balanced DPSK detector; the generated photocurrents for the intensity and the phase are polarity-inverted both for graphical and decoding reasons, and neglecting the noise terms are evaluated by Eq. (1) in [3].
2. Dark pulse PhIM
PhIM format using dark pulses has been described in detail in Ref. [3]. A continuous wave optical beam is intensity modulated in the form of a coded dark pulse sequence with bit rate R (Gb/s), representative of a first amplitude tributary; with reference to Fig. 1, let TB=1/R be the tributary bit time slot; the bit time portion TB-Δt between one bit and the following is characterized by having nearly unperturbed optical power. A phase modulation with bit rate R, representative of a second phase tributary, is added to the dark pulse sequence using the unperturbed bit time portion, i.e., the optical phase level in the time interval TB-Δt shall be 0 or Δϕ≤π. The phase tributary is delayed by half bit time slot respect to the amplitude tributary, and is commonly differentially pre-encoded in order to be detected by a common DPSK receiver.
A simplified scheme for a dark pulse PhIM transmitter includes one intensity and one phase modulator with electrical cut-off frequency comparable to R (GHz).
In this work a new transmitter is proposed, requiring a single optical modulator with dual drive design, that generates both the orthogonal amplitude and phase modulations. If NRZA and NRZϕ are the normalized electrical amplitude and phase tributaries and Vπ is the modulator inversion voltage, both the amplitude and phase optical signals are generated applying to the modulator the driving voltages V 1(t) and V 2(t) obtained as
Dual drive modulator with a finite extinction ratio ERlin is modeled by the amplitude transfer function [5]
The transmitter scheme corresponding to Eqs. (1) is shown in Fig. 2. The electrical amplitude tributary is differentially pre-encoded in order to produce fast transitions around the modulator zero transmission point, i.e. optical dark pulses, in correspondence of a tributary mark; this method also produces π phase jumps at each generated optical dark pulse. The phase tributary performs an XOR operation with the amplitude tributary in order to subtract the π phase jumps to the phase modulation; optical signal thus results also phase modulated by a signal proportional to NRZϕ, which is differentially pre-encoded to produce a DPSK code. Driving signals V 1(t) and V 2(t) are obtained as the sum and the difference of the elaborated tributaries, producing orthogonal amplitude and phase terms into the modulator transfer function. The two buffers in Fig. 2 are needed to produce the desired voltage levels to the modulator; the upper buffer is also used to delay the amplitude tributary by an amount equivalent to one logic gate, in order to recover the input synchronization between tributaries; buffers may be replaced by clocked gates.
The generated optical signal is similar to the one shown in Fig. 1, with a NRZ optical phase. Figure 4(b) shows an example of received eyes (left: amplitude, right: phase tributary) at 2×42.7 Gb/s. Both the eyes show polarity-inverted currents; this is needed for the amplitude tributary to recover the logical conjugation produced by the dark pulses, while for the phase it is used just for graphical purposes; amplified spontaneous emission (ASE) noise is visible in the lower levels of the amplitude tributary.
3. Bright pulse PhIM
PhIM using bright pulses is implemented as the polarization interleaving of a return-to-zero (RZ) signal and a return-to-zero differential phase shift-keying (RZ-DPSK) signal. As shown in Fig. 3, a shaper modulator is needed to generate a pulse train that is split in two portions; the former is intensity modulated by the amplitude tributary NRZA, the latter is phase modulated by a differentially pre-encoded phase tributary NRZϕ and is delayed by half bit slot before being polarization multiplexed with the RZ signal. The ratio s between the RZ and the RZ-DPSK pulses peak power should preferably fall between 1 and 2 (an optimal value of 1.5 is found in section IV).
The basic advantage of bright pulse PhIM respect to other polarization-interleaved signals is that neither a polarization controller nor the use of a time window are needed at the receiver (the receiver scheme is identical to the one for dark pulse PhIM). In terms of received amplitude tributary, this feature is explained observing that the received optical power eye is formed by a central RZ-like shape, added to two smaller side lobes representing the DPSK signal pulses; once the overall signal is detected and electrically filtered, a received eye like the example shown in the left Figure 4(c) is obtained, for any received state of polarization (SOP). Electrical polarity has been inverted in the figure for graphical reasons; in this case a logically conjugated NRZA at the transmitter is needed. ASE noise is visible in the lower eye levels. In terms of received phase tributary, the RZ signal does not contribute to the received DPSK eye, because it adds a constant phase to the signal; phase eye thus shows the phase information carried only by the RZ-DPSK signal independently on the received SOP; the DPSK detector produces a received eye like the one shown in right Figure 4(c), where polarity is also inverted for graphical purposes.
4. Numerical results
Numerical simulations have been performed using a commercial software [6], making a direct comparison between pulsed PhIM and the common NRZ format. Three cases have been considered, with optimized parameters:
i) NRZ optical signal at 85.4 Gb/s bit rate, i.e., including a forward error correction (FEC) overhead; signal has 15 dB extinction ratio and 5.8 ps rise and fall times (equal to half the bit time slot), corresponding to 57 GHz cut-off frequency for the modulator and the transmitter electronics. At the receiver, a PIN photodiode and a front-end with 70 GHz cut-off frequency are used.
ii) Dark pulse PhIM at 2×R=85.4 Gb/s capacity, and R=42.7 Gb/s tributary bit rate; modulator extinction ratio is 15 dB and optimal rise and fall times for both the amplitude and phase are found to be 14 ps (60% the bit time slot), corresponding to electronics with 24 GHz cut-off frequency at the transmitter. Both the intensity receiver and the DPSK balanced detector have 35 GHz cut-off front-ends.
iii) Bright pulse PhIM at 2×R=85.4 Gb/s; the pulse train has 7.8 ps full-width at half-maximum (33% duty cycle). The choice of the duty cycle is forced by the use of a shaper modulator; it has been observed that larger pulses would distort the amplitude eye, while shorter pulses would increase the signal bandwidth. An optimal value for the split ratio is s=1.5; the intensity data modulator is characterized by a 15 dB extinction ratio; the phase modulator is with linear response and π phase modulation amplitude; the intensity and DPSK receivers have 35 GHz cut-off front-ends.
Figure 4(a) shows the transmitted spectra for cases i) (left), ii) (middle) and iii) (right). Figure 4(b) represents the received eyes (left: amplitude, right: phase tributary) for case ii) in single-channel case; ASE noise is included with 28 dB OSNR. Figure 4(c) gives the received eyes in case iii); the phase eye (right) has more complex shape, which is affected by the choice of the split ratio s; both eyes show a noteworthy time width also in this case, in spite of the used time-polarization interleaving.
Fig. 4. (a) transmitted spectra for the 3 cases described in the text. (b) received amplitude (left) and phase (right) eyes for dark pulse PhIM. (c) received amplitude (left) and phase (right) eyes for bright pulse PhIM.
A back-to-back test has been performed in the 3 cases, simulating a transmission line composed by a lossy and birefringent element between two erbium doped fiber amplifiers (EDFAs); birefringence is used to produce a random output SOP. At the receiver, a 2nd order Gaussian optical filter is used with 300 GHz bandwidth. Figure 5(a) shows the obtained Q factors vs. OSNR for the NRZ signal and for the amplitude and phase tributaries of the pulsed PhIMs; Gaussian approximation has been used to evaluate the Q factors for both the amplitude and the phase; this is the cause of a different slope for the phase curves, which are affected by the Gaussian approximation at high OSNR [7]. In the following, the worse result between amplitude and phase will be considered for the comparison with the NRZ. At Q=10 dB (corresponding to approximately a BER=10-9 when FEC is considered) dark pulse PhIM requires a OSNR increased by 2.2 dB respect to NRZ (correspondingly, sensitivity is reduced), while for bright pulse PhIM it is on the contrary reduced by 1.2 dB.
Tolerance to CD is tested using a linear dispersive element in place of the lossy element; again birefringence is used, and no ASE noise is included. Figure 5(b) shows the eye closure penalty (ECP, the eye closure is defined as the ratio between the average and the minimum eye opening) vs. CD: the 1 dB penalty for cases i), ii) and iii) is obtained at -10 to 10, -14 to 14, -20 to 20 ps/nm respectively. The phase tributary of bright pulse PhIM shows reduced tolerance due to a more complex eye shape.
Tolerance to DGD is simulated rotating the transmitted polarization by 45° and introducing a delay between the two principal states of polarization. Figure 5(c) shows 1 dB penalty for cases i), ii) and iii) at 6.5, 8 and 4.5 ps respectively, and 5 dB penalty at 10, 16.5 and 12.5 ps. Bright pulse PhIM has lower tolerance respect to dark pulse because of the polarization interleaving.
WDM behavior is simulated transmitting 3 channels for each case, with variable channel spacing. Channels are multiplexed with orthogonal polarizations, and a birefringent line with no loss is used to generate a random output SOP. De-multiplexing optical filter is a 2nd order Gaussian with bandwidth equal to 75% the channel spacing. Figure 5(d) gives the obtained ECP for the middle channel vs. channel spacing. Cases i), ii) and iii) experience a 1 dB penalty at 145, 120 and 100 GHz respectively, showing the possibility to transmit pulsed PhIM with spectral efficiencies over 0.8 bit/s/Hz.
Fig. 5. (a) Q factor vs. OSNR in the 3 cases described in the text. (b) eye closure penalty vs. cumulated dispersion. (c) eye closure penalty vs. DGD. d): eye closure penalty vs. channel spacing in WDM transmission.
In conclusion, dark pulse PhIM presents significant advantages respect to NRZ in terms of DGD tolerance and spectral efficiency; transmitter in Fig. 2 requires a single dual drive modulator and electronics with cut-off frequency around 0.55R, while the required NRZ modulator bandwidth is over 1.3R; receiver sensitivity is on the contrary reduced by 2.2 dB. Bright pulse PhIM shows improved tolerances to CD, greater spectral efficiency and increased sensitivity by 1.2 dB respect to NRZ; conversely it requires two modulators and a shaper with cut-off frequencies comparable to R.
Dark pulse PhIM is compatible with the polarization domain multiplexing (PDM) technique.
References and links
1. M. Ohm and J. Speidel, “Quaternary optical ASK-DPSK and receivers with direct detection,” IEEE Photon. Technol. Lett. 15, 159–161 (2003). [CrossRef]
2. N. Chi, J. Zhang, P. V. Holm-Nielsen, C. Peucheret, and P. Jeppesen, “Transmission and transparent wavelength conversion of an optically labeled signal using ASK/DPSK orthogonal modulation,” IEEE Photon. Technol. Lett. , 15, 760–762 (2003). [CrossRef]
3. M. Zitelli, “Optical phase and intensity modulation using dark pulses,” IEEE Photon. Technol. Lett. 16, 1972–1974 (2004). [CrossRef]
4. M. Zitelli, “Optical phase and intensity modulation with improved transmitters,” U.S. patent application 10/732,404, Dec. 11, 2003.
5. S. Walklin and J. Conradi, “Multilevel signaling for increasing the reach of 10 Gb/s lightwave systems,” IEEE J. Lightwave Technol. 17, 2235–2247 (1999). [CrossRef]
6. “Optolink Software user documentation” (TelCon, 2004), http://www.tel-con.com.
7. G. Bosco and P. Poggiolini, “On the Q factor inaccuracy in the performance analysis of optical direct-detection DPSK systems,” IEEE Photon. Technol. Lett. 16, 665–667 (2004). [CrossRef]
