This PDF file contains the front matter associated with SPIE Proceedings Volume 8284, including the Title Page, Copyright information, Table of Contents, and the Conference Committee listing.

The capacity of optical transmission systems has increased dramatically since their first deployments in the mid 1970s . However, studies show that the theoretical capacity limit of single-mode fiber is about to be reached, and space-division multiplexing has been proposed to overcome this limit. With the high levels of integration needed for economic deployment, space-division multiplexing may exhibit large crosstalk between the supported fiber modes. We propose to use coherent multiple-input multiple-output (MIMO) digital signal processing (DSP), a technique widely used in wireless communication, to compensate crosstalk present in spatial multiplexing over fibers. According to MIMO theory, crosstalk in multi-mode transmission systems can be completely reversed if the crosstalk is described by a unitary transformation. For optical fibers this is fulfilled if all available fiber modes can be selectively excited and if all the modes are coherently detected at the end of the fiber, provided that mode-dependent loss is negligible. We successfully applied the technique to demonstrate the transmission of six independent mode-multiplexed 20-Gbaud QPSK signals over a single, optically amplified span of 137-km few-mode fiber (FMF). Further, in a multi-span experiment, we reach 1200 km by transmitting over a 3-core coupled-core fiber (CCF). Details for both experiments will be presented, including the description of the supported polarization- and spatial modes of the fiber, the mode multiplexers used to launch and detect the modes, and the MIMO DSP algorithm used to recover the channels.

Steadily increasing data traffic gives rise to increasing capacity requirements in optical communication networks. It is well understood that systems with higher symbol rates and/or multi-level modulation formats generally demand higher optical signal-to-noise ratio (OSNR) at the receiver to achieve acceptable system performance. In terms of the optical fiber medium, higher OSNR can be attained by lowering fiber loss and reducing fiber nonlinearity. We review several recent experimental investigations of 112 Gb/s PM-QPSK transmission with reach-length results enabled by the use of optical fibers with ultra-low loss and very large effective area.

We present a novel idea of multi-input injection locking phenomenon in single mode Fabry-Pérot Laser diode (SMFP-LD). The key principle of multi-input injection locking is the proper power management of input beams for injection locking one of the side modes of Fabry-Pérot laser diode (FP-LD) and the suppression of the dominant mode of FP-LD. The multi-input injection locking (MIL) principle is extended to the combinational multi-input injection locking (CMIL) by which the dominant mode of SMFP-LD will be suppressed with certain combination of input beams. The proposed idea can be implemented for realizing logic gates, decision making circuit and others. MIL and CMIL are explained theoretically and verified by realizing some of the logic gates. Experimental results for logic gates are presented with the input data rates of 10 Gbps.

Recently efforts have been focused on ultra-high speed optical communication systems which can support 1Tb/s per channel transmission. However, 1Tb/s over a single carrier requires either or both very high-level modulation format (i.e. PDM-1024QAM) and high baud rate. Therefore, grouping a number of tightly spaced "sub-carriers", to form a terabit channel has been considered and this has been refered to as a superchannel. Nyquist-WDM and Coherent Optical- OFDM (CO-OFDM) are the two approaches to achieve ultra-high spectral efficiency in superchannel coherent optical systems. In Nyquist-WDM systems, optical subcarriers are tightly packed at channel spacing near or equal to the baud rate, potentially inducing strong inter-channel interference (ICI). The traditional way to mitigate the impact of ICI is by applying aggressive optical filters to each channel; however this typically induces severe inter-symbol interference (ISI). In this paper, we investigate receiver architectures for Nyquist-WDM superchannel coherent systems, and propose a new "super receiver" architecture, which jointly detects and demodulates multiple channels simultaneously. Several joint DSP algorithms are developed and tested through experimental and simulated data. The simulation results showed that more than 5 dB ONSR gain was achieved comparing to conventional method at narrow channel spacing conditions.

In the near-field regime, the number of spatial modes that a free-space communication system can efficiently use is given by the product of the Fresnel numbers of the transmit and receive apertures. It can be advantageous to decompose the field into modes that have rotational symmetry or definite orbital angular momentum (OAM modes). A key challenge to using OAM modes as parallel channels in a practical communication system is efficient multiplexing of single-spatial-mode transmitters to the orthogonal OAM modes, and demultiplexing the combined beam into single-spatial-mode receiver arrays. Previous approaches have utilized modes of different OAM, but ignored the radial coordinate, leading to inefficient use of the Fresnel number. We identify a method, using lenses and holographic phase plates, to efficiently and reversibly convert concentric Laguerre-Gauss OAM beams into an array of separated Gaussian beams.

The constant envelope characteristic of CPM signal is particularly interesting for use in fiber optic links since it can be leveraged to avoid nonlinear phase modulation. Implementation complexity of CPM systems is generally higher than their QPSK counterpart, partly due to the nontrivial task of generating the signal as well as the need to observe the received signal over multiple symbol periods to make an optimal decision. Because of this complexity, the use of full-response CPM systems is favorable for complexity since optimal reception is achieved with lower order, however, partial response systems can achieve higher minimum distance. We analyze parameter selection for CPM transmission, optimizing error performance and spectral efficiency in a tightly filtered reconfigurable optical add-drop multiplexor (ROADM) application. We illustrate their impact on the normalized minimum Euclidean distance (as a proxy for error performance). The impact of parameters on spectral efficiency is implicit on the choice of ROADM filters. The results provide guidance for a suitable choice of CPM scheme for consideration in DWDM systems. Optimal parameters are given for full and partial response systems for a variety of filter scenarios.

We describe a innovative OFDM scheme based on orthogonal chirped subcarriers, that corresponds to the fractional Fourier transform (FrFT) of the input signal. The FrFT can be electronically implemented with a complexity equivalent to the conventional fast Fourier transform (FFT); on the other hand, the planar device that implements the FrFT in the optical domain is similar to the passive arrayed waveguide grating (AWG) device that performs the FFT. We analyze the spectral efficiency, the peak-to-average power ratio (PAPR) and the frequency offset sensitivity of a FrFT-based optical OFDM system, and make an accurate comparison with the standard FFT-based implementation.

Recent advances in electronic data processing allow constructing reconfigurable optical transmitters and receivers, where modulation formats and symbol rates are set by software-controlled field programmable gate arrays (FPGA). We report on such a real-time optical transmitter for 8 modulation formats, which can be swapped in 5 ns without data loss. With single-polarization 64QAM symbols generated at 28 GBd, we transmit data at 168 Gbit/s in real time. A similar arrangement defines a single-polarization orthogonal frequency division multiplexing (OFDM) transmitter for a data rate of 101.5 Gbit/s, where 58 subcarriers are encoded with 16QAM data. With a different software setup, the FPGA realizes an optical 56 Gbit/s transmitter for sinc-shaped so-called Nyquist pulses, the spectrum of which is rectangular having the minimum theoretically achievable bandwidth (suitable for Nyquist wavelength division multiplexing, N-WDM). For terabit OFDM reception, optical pre-processing is required to demultiplex high-bitrate signals down to lower-bitrate tributaries, which then can be processed electronically. We discuss a 10.8 Tbit/s (26 Tbit/s) receiver employing an all-optical fast Fourier transform to demultiplex 75 (325) optical subcarriers modulated with 16QAM-formated symbols at a rate of 18 GBd (10 GBd). Groups with any number of subcarriers can be selected with a simple hardware reconfiguration step.

We experimentally investigate the performance of WDM coherent polarization-division multiplexed-return to zero-quadrature phase shift keying (PDM-RZ-QPSK) network in nonlinear transport regimes. Seven, 28 or 32-Gbaud PDM-RZ- QPSK channels are employed on a 50-GHz grid and transmitted over 1600-km fiber on an all-EDFA recirculating loop without any dispersion compensation module (DCM). The transmission link is configured entirely of either standard single-mode fiber (AllWave), medium dispersion fiber (TrueWave REACH), or ultra-large area fiber (ULAF). We sweep the launch power of the center channel and side channels together to measure the nonlinear effects of self-phase modulation (SPM), cross-phase modulation (XPM), and cross-polarization modulation (XPolM) on the center channel's BER performance. Furthermore, for all link configurations, we employ three different carrier phase recovery methods in the demodulation routine - Viterbi-Viterbi, Viterbi-Viterbi with a minimum mean-squared error (MMSE) filter, and the Optametra/Tektronix Weiner filter - to ascertain their relative performance in the presence of nonlinear effects.

The performance of polarization multiplexed, quadrature phase shift keying (PM QPSK) and polarization multiplexed 16-ary quadrature amplitude modulation (PM 16-QAM) is considered with an emphasis on the signal processing algorithms that compensate transmission impairments and implement key receiver functions.

We realized high-speed and low driving voltage InP-based Mach Zehnder modulators with an npin high-mesa waveguide structure and traveling-wave electrodes. A full C-band 40-Gbit/s DPSK signal was successfully generated using a compact tunable wavelength transmitter module, which incorporated a tunable DFB laser array as a wavelength tunable laser and an InP MZM in one package. The MZM had a low driving voltage of 3 V_pp in a push-pull driving configuration. We have demonstrated 112 Gbit/s RZ-DP-QPSK modulation using two InP MZM modules. One modulator was used as a pulse carver in a push-pull configuration, and the other was used as an IQ modulator in a dual-drive configuration. Since we used an RZ pulse carver to remove the transient region, we obtained clear eye openings without any amplitude ripple. We also demonstrated 50 Gbit/s (12.5 GSymbol/s x 4) 16QAM signal generation employing a novel dual-drive modulation method consisting of a single MZM. We utilized the electro-absorption characteristics of an InP semiconductor to adjust the QPSK amplitude. We confirmed that a single MZM operated as a DPSK, a QPSK, and a 16QAM modulator with the same device. We believe these modulators to be suitable for next generation optical transmission systems.

A study is presented of the fiber properties needed to achieve 10-mode multiplexing transmission. A combination of MIMO processing with optical LP mode separation is proposed to prevent the need for massive MIMO computation. The impact of mode crosstalk, differential mode delay, and the mode dependent loss of the few-mode fibers on mode multiplexing are discussed.

Spatial division multiplexing has been proposed for fiber capacity increase as an alternative to enhanced bandwidth efficiency by higher order modulation or wider wavelength bands. The commercial viability of this approach will depend strongly on the feasibility of cost and energy efficient optical amplification. Lumped erbium doped fiber and distributed Raman amplifiers are the most promising candidates. Different approaches have been analyzed for the realization of preamplifier and booster stages with different active fiber types. The multi mode fiber type was found to possess more potential for cost and energy efficiency than the multi core fiber type.

In response to ever growing data traffic, significant efforts are being made to increase optical network capacity. One promising candidate is mode-division multiplexing (MDM) in few-mode fibers, which uses space as a new information-bearing dimension. A fundamental element for MDM is a modal transformer. Modal transformation can be implemented in a free-space basis by using multi-region phase plates. In this work we present the design, fabrication and characterization of monolithic binary phase plates by highly-uniform Ag⁺/Na⁺ ion-exchange in glass. Diffracted optical field intensities have been measured and high quality mode transformation has been confirmed.

The limitations of crosstalk and core-to-core distance in step-index multi-core fibers (SI-MCFs) are clarified for long-haul transmission, and the low-crosstalk MCF structures of trench-assisted MCFs (TA-MCFs) are investigated for realizing large effective area (Aeff) and high core density, simultaneously, with a limited cladding diameter. It is shown that the crosstalk between neighboring cores in TA-MCFs can be greatly suppressed even if the Aeff and the cutoff wavelength are fixed compared with SI-MCFs. In addition, the possibility of MCFs with heterogeneous core arrangement is considered for transmission fibers and low-crosstalk heterogeneous MCFs with bending radius insensitive characteristics are investigated.

A simple optical technique is described for the measurement of complex mode amplitudes in multimode fibers. The technique used in this experiment involves launching a laser beam into the core of low moded optical fiber, exciting a mixture of normal modes. The emerging mode patterns are stored in microcomputer. The complex modes are then extracted from these recoded intensities with help of modified version of Gerchberg-Saxton algorithm. Many experimental results are recorded in photographic form and comparison with computer analyzed pattern is made. The evaluation of complex mode amplitudes in multimode fiber can yield useful information about the mode filling of such fiber.

We present techniques for modeling the physics and systems-level characteristics of integrated IQ-transmitters for 100G+ applications and emphasize important design aspects. Using time-and-frequency-domain modeling (TFDM) of Photonic Integrated Circuits (PIC), we present a detailed IQ-transmitter model based on the physics and setup of active and passive subcomponents. With this, we link characteristics of subcomponents (bending loss of waveguides, phase changes in MMI couplers, sweep-out time of EAMs) to systems-level characteristics of the integrated IQ-transmitter (extinction ratio, modulation bandwidth, chirp). Further, a behavioral transmitter model is introduced and utilized to assess electrical driving requirements (allowed jitter, noise, synchronization offset).

Novel closed form formulae are derived to study crosstalk degradation due to stimulated Raman scattering (SRS) in WDM systems employing multi-span bi-directional pumped distributed Raman amplifier (DRA). The formulae are used to evaluate the crosstalk performance of differential phase-shift keying (DPSK) and ON-OFF keying (OOK) modulation format which are widely used in optical communication. SRS crosstalk is further evaluated for different data rates and pulse shapes prevalent in optical data transmission. Next, crosstalk is calculated for different pumping schemes by taking into account the launched power of bi-directional pumped DRA. The study shows that minimum SRS crosstalk can be achieved for 40 Gb/s RZ-DPSK signal with 33.3% duty cycle in WDM system employing backward pumped DRA.

Recent numerical and experimental studies have shown that coherent transmission with advanced modulation formats i.e. DP-QPSK and QAM are the promising candidates for next-generation systems with data rates of 100Gbit/s and above. Coherent detection is considered efficient along with digital signal processing (DSP) to compensate many linear effects in fiber propagation i.e. chromatic dispersion (CD) and polarization-mode dispersion (PMD). Despite of fiber non-linearities (NL), which are the major limiting factors, next-generation optical systems are employing higher order modulation formats in order to fulfil the ever increasing demand of capacity requirements. However, the channel capacity is limited at higher signal input powers because the system is operating in the non-linear regime. Due to this phenomenon the compensation of non-linearities is a topic of great interest and research these days, especially for long-haul fiber transmission. Digital backward propagation (DBP) algorithm has emerged as a promising and potentially capable candidate, which can jointly compensate fiber dispersion and non-linearities along with the coherent receiver. In this paper we give a detailed overview on the advancements in DBP algorithm based on different types of mathematical models i.e. Wiener (Asymmetric Method) and Wiener Hammerstein models (Symmetric Method). We also discuss the importance of optimized step-size selection, i.e. constant step-size and logarithmic step-size based split step Fourier methods, for simplified and computationally efficient implementation of DBP algorithm. Moreover, by means of numerical investigations we refer to recent system investigations to further improve the performance of DBP algorithm.

Nonlinear distortion is one of the major obstacles in DWDM systems with enhanced spectral efficiencies. In this paper several approaches to address the issue of nonlinear impairments by means of digital signal processing are discussed. Firstly, implementation-efficient and novel intra-channel nonlinear compensation schemes are proposed; one is based on digital pre-distortion at the transmitter end and the other is based on digital back-propagation at the receiver end. The virtues of the two approaches and implications to various applications are discussed; the pre-distortion technique is in particular advantageous with QPSK modulation format; on the other hand, the improved version of digital back-propagation is attractive in transceivers with adaptive or variable modulation/demodulation. Second, digital signal processing algorithms to counteract inter-channel nonlinearities, namely cross-phase modulation, are discussed; nonlinear polarization crosstalk canceller (NPCC) is proposed for mitigating the impact of nonlinear-induced fast polarization crosstalk in dual-polarization systems (in the speed beyond MHz), which is too fast to be tracked by ordinary polarization demultiplexing algorithms; improvement to the carrier phase recovery circuit and its combination with NPCC are even more useful for further performance improvement. Numerical and experimental data are introduced to support the above discussions.

We have investigated a new method to reduce the complexity of the digital backward propagation algorithm (DBP). A logarithmic step-size based split-step Fourier method (SSFM) is investigated in this paper to compensate fiber transmission impairments i.e. chromatic dispersion (CD) and non-linearities (NL) in dual-polarization quadrature phase shift keying (DP-QPSK) system. The algorithm is numerically investigated for coherently-detected multiple channel DP-QPSK system over 2000km (25 spans) standard single mode fiber (SMF-28) with un-compensated transmission link. The algorithm is numerically evaluated for: (a) 20 channel 56Gbit/s (14GBaud) with 25GHz channel spacing; (b) 10 channel 112Gbit/s (28GBaud) with 50GHz channel spacing and (c) 5 channel 224Gbit/s (56GBaud) with 100GHz channel spacing. Each simulation configuration has the bandwidth occupancy of 500GHz and a total transmission capacity of 1.12Tbit/s. The logarithmic DBP algorithm (L-DBP) shows efficient results as compared to the conventional DBP method based on modified SSFM (M-DBP). The results depict efficient mitigation of CD and NL, therefore improving the non-linear threshold point (NLT) upto 4dB. Furthermore by implementing a low-pass-filter (LPF) in each SSFM step, the required number of DBP stages to compensate fiber transmission impairments can be significantly reduced (multi-span DBP) by 75% as compared to L-DBP and by 50% as compared to M-DBP. The results delineate improved system performance of logarithmic step size based filtered DBP (FL-DBP) both in terms of efficiency and complexity which will be helpful in future deployment of DBP algorithm with real-time signal processing modules for non-linear compensation.

We have numerically investigated the impact of non-linear impairments on the performance of 400Gbit/s DP-RZ- QPSK transmission system over 1200km standard single mode fiber (SMF-28) having an average span loss of 16dB and with no in-line optical dispersion compensation in the transmission link. Digital backward propagation (DBP) algorithm based on split-step Fourier method (SSFM) is employed along with the coherent receiver to compensate the fiber transmission impairments i.e. chromatic dispersion (CD) and non-linear (NL) impairments. The system performance is monitored in terms of Q-value (calculated form BER) for various signal input launch powers. We further quantify the impact of inter-channel non-linear impairments such as cross-phase-modulation (XPM) and four-wave-mixing (FWM) on the performance of DBP algorithm by investigating the multiple-channel transmission, i.e. 8x400Gbit/s DP-RZ-QPSK system. The results depict efficient performance of DBP algorithm as compared to the system where only linear dispersion compensation is implemented. This shows the promising impact of digital backward propagation algorithm on the high data-rate transmission systems such as 400Gbit/s per single channel which is expected to be a possible data rate for long-haul optical communication systems after 100Gb Ethernet in near future.

To optimise the design of a wavelength selective switch based on a phase-only liquid crystal on silicon spatial light modulator and mitigate crosstalk, we propose using a technique referred to as wavefront encoding that involves purposefully building a wavefront error into the optical system. Experimental results taken at 674nm are presented that show wavefront encoding based on defocus can reduce the worst case crosstalk by >10dB compared to a standard Fourier transform set-up. In the case of the WSS we propose using wavefront encoding based on astigmatism.

In the last decade, there has been increased interest in photonic technology for new satellite applications. One critical issue is the high sensitivity to radiative environments of the Erbium Doped Fiber (EDF). It leads to a radiation-induced absorption (RIA) that is not due to erbium content but mainly to the aluminium that ensures the erbium inclusion in glass. As the radiation induced losses grow as an exponential function of fiber length, the principal way so far to reduce EDFA degradation has consisted in increasing erbium concentration using conventional doping techniques. However, this is limited by the quenching effect, which impacts the fiber length needed to reach high gain, but also by the Aluminium-induced RIA. It has been recently proposed an original nanoparticle (NP) doping approach, which allows codopant content decrease with reduced quenching impact, while keeping EDF amplifying performances. A radiation-resistant amplifier can thus be designed as a "quenching-free", heavily-erbium-doped amplifier with low RIA. We demonstrate for the first time an aluminium-free EDF, exhibiting low quenching and low RIA. Despite the lack of aluminium, using silica NPs allows an erbium concentration close to the one of standard EDFs (200 ppm). This fiber is compared to a 1400 ppm Erbium-doped optical fiber with a strong aluminium concentration. Whereas the two fibers exhibit similar initial optical gain (15 dB under saturation conditions), the NP doped Al-free EDF shows only 2 dB gain reduction after a 600 Gy gamma deposit, while the Al/Er EDF incurs more than 10 dB gain degradation.