3.1 Publishable summary

Transcription

1 3.1 Publishable summary Project description and objectives The project goal is to develop the disruptive technology and concepts needed to enhance our communications infrastructure 1-fold to meet future needs, avert network gridlock and reduce energy consumption. Driven by the exponentially growing demand for capacity, it is apparent that the next generation of telecommunication networks will be radically different from previous implementations, coherent detection and multi-carrier techniques, along with powerful digital signal processing will be deployed to maximise the available capacity of each fibre strand within the network. However, whilst welcome, such developments will only delay the inevitable single mode fibre capacity crunch. Put simply, once these developments are deployed, it will only be possible to increase network capacity by lighting additional fibres, with a cost linearly increasing with capacity. With a per-annum growth rate of over 4%, this only delays a total capacity exhaust by a few years before new cable deployments are required. The imminence of new and extensive deployment of new fibre cables in the next decade (22-3) provides a unique opportunity to re-examine our choice of transmission medium in order to have a dramatic impact on the exploitable network capacity increase which would accompany such a deployment. MODE-GAP is exploiting this opportunity by developing two complementary technologies, both of which would individually offer sufficient benefit, but which combined offer extraordinary potential increases in available network capacity. The concept of the project is to address the future needs by optical Multiple Input Multiple Output (MIMO) through longhaul Mode Division Multiplexing (MDM) over multimode fibre. The Project objective has been to demonstrate increased transmission capacity potential by applying MIMO techniques to few mode optical fibre and gaining the benefits of utilising Photonic Band Gap Fibre (PBGF) in the 2 m region. A three strand approach has been adopted i. MDM techniques in solid core few mode fibre transmission in the C-band ii. Transmission in PBG fibre in the C-band iii. Transmission in PBG fibre at 2 m To meet the overall objective there are component and system challenges to be met: Fibres Develop few mode solid core fibres as well as ultra-low loss multi-mode photonic band gap transmission fibre (MM-PBGF). Rare earth doped optical amplifiers - Develop novel few mode rare earth doped optical amplifiers as well as amplifiers for the new transmission windows necessary for the achievement of the lowest loss. Sources and detectors - Develop sources and detectors operable in to the 1.8 to 2.1 um region. Multimode fibre SDM coupling - Develop multiplexing and demultiplexing components for operation in the C-band and 2 m window MIMO Processing - Develop MIMO and dispersion compensation signal processing algorithms applicable to both conventional solid core fibres and MM-PBGF. Long Haul WDM transmission- Demonstrate the concept on a long haul WDM transmission testbed Publishable summary Page 1 of 6

2 3.1.2 Description of work and key results System trials The last reporting year has seen huge progress in data transmission over novel transmission fibres, targeting important demonstrations of the promise of SDM which allows up to two orders of magnitude capacity increase with respect to SMFs. SDM is achieved through multiple-input multiple output (MIMO) transmission employing spatial modes of a multi-mode fibre (MMF), or multiple single-mode cores as channels. Recently, a distinct type of MMF, the few-mode fibre (FMF), has been developed to co-propagate 3 or 6 linear polarised (LP) modes Driven by rapid enhancements in high-speed electronics, digital signal processing (DSP) MIMO techniques can faithfully recover mixed transmission channels, allowing spectral efficiency increases as spatial channels occupy the same wavelength. State-of-the-art single-carrier FMF transmission experiments have demonstrated capacity increases in a single fibre by exploiting 6 spatial modes, achieving 32 bit s -1 Hz -1 spectral efficiency. By employing multicore transmission, a spectral efficiency of 19 bit s -1 Hz -1 has been demonstrated using 12 single-mode cores. In year 4, the project has demonstrated ultra-high capacity transmission over a 1 km hole-assisted few-mode multi-core fibre (FM-MCF), employing 7 few-mode cores, each allowing the LP 1 and two degenerate LP 11 modes to co-propagate in both polarisations. A custom designed butt-coupled integrated 3D waveguide multiplexed all 21 spatial LP modes per linear polarisation being simultaneously transmitted. The fibre design minimizes inter-core crosstalk and reduces the required MIMO equalizer complexity from to 7 (6 6), and hence reduced energy consumption. In addition, an energy efficient MIMO frequency domain equalizer (FDE) was employed per core. A single-carrier spectral efficiency of 12 bit s -1 Hz -1 (if conventional Dual Pol OOK SMF is about 2bit s -1 Hz -1 this represents a 5 fold increase) was achieved by encoding 24.3 GBaud 32 quadrature amplitude modulation (QAM), allowing for next generation 5.13 Tbit s -1 carrier -1 gross (4 Tbit s -1 carrier -1 net) data rate spatial super channels. Combining the spatial dimension with 5 wavelength channels on a 5 GHz International Telecommunication Union (ITU) grid, a gross total capacity of 255 Tbit s -1 (2 Tbit s -1 net) was demonstrated, further indicating the viability of combining few-mode and multi-core transmission techniques in a single fibre for achieving ultra-high capacity data transmission This work was widely cited in the press in 214 and was published in Nature Photonics. (a) ChUT Laser Loading Channels Laser 1... Laser 5 Multiplexer EDFA 1 km 24.3Gbaud Transmitter EDFA 44n s Polarization Multiplexer Odd/Even Channel Decorrelation 293ns EDFA Odd 1GHz Interleaver Even Loading Cores Local Oscillator 75ns 185ns 551ns EDFAs CoUT 1:18 Core Switch = 3D waveguide Multiplexer 21 1km FM-MCF 3D waveguide Demultiplexer 21 = Core Switch CoUT TDM-SDM Receiver (b) (c) (d) Even Odd Wavelength [nm] Relative Trans mitted Power [db] µm Core 1 Core 2 Core 3 Core 4 Core 5 Core 6 Core 7 Core 4 Publishable summary Page 2 of 6

3 Fig 1. FM-MCF PDM/WDM/SDM experimental transmission setup. a, the loading channels and one channel under test are simultaneously modulated by a 24.3 GBaud 16 or 32 QAM constellation sequence. Consecutively, polarisation, carriers, cores and modes are decorrelated. The 3D multiplexer guide the transmission channels into and out of the FM-MCF through butt-coupling, where the CoUT is varied through all cores consecutively. b, the decorrelated wavelength spectrum after being interleaved by a wavelength selective switch. c, saturated camera image taken at the receiver side, where all cores are simultaneously lit. d, independent cores are excited, indicating low crosstalk per core. Right bottom, selective launching of the LP 1 and LP 11 modes in the centre core, respectively, where the modal energy is confined to the centre of the hexagonal core structure In other work at 155nm, we demonstrated the transmission of 12 spatial and polarization modes multiplexed on 255 optical channels in ultra-dense WDM using OFDM across the full C-band over km 6-mode fibre at 1.55µm. An aggregate bit rate of 41.6 Tbit/s was achieved. We have also demonstrated real-time transmission over the longest hollow core fibre to date (2.75km) using a commercial 1Gb/s interface. Given its excellent optical properties with low distortion and high optical performance, this contribution marks an important stepping stone towards a large scale manufacturability of longer-length hollow core fibre targeting intra data centre connectivity with target reaches of 5m-2km. In the final year, the 2 m devices were incorporated within the systems test bed, in conjunction with amplifiers and low loss fibre developed within the project. We demonstrated the first coarse WDM (CWDM) and the 1 st dense WDM (DWDM) system at this new waveband. Fibres Work on development of solid silica few mode with a low differential modal group delay (DGD) has continued in the last period of MODE-GAP. For the four LP mode fibre the focus has been on further improving the DGD control. 5 km of 4 LP mode fibre was fabricated where DGD for 4 % of the fibres is controlled to within ±.1 ps/m and for 6 % of the fibres this is controlled to within ±.2 ps/m. A low DGD fibre supporting nine LP modes equivalent to 15 spatial modes was developed. The fibre showed low attenuation between.2 and.22 db/km for all modes. As for the HC-PBGF development, Year 4 has seen very substantial activity aimed at the main project goal of demonstrating very long lengths of low loss HC-PBGF. Work focussed on further understanding the loss mechanisms and refining the models and lowest loss predictions and related optimum fibre structures. A direct measurement of the wavelength scaling of the loss in a 19cell HC-PBGF was achieved, confirming the region around 2um as the region affording the lowest transmission loss and scattering from surface roughness as the main source of loss. Various tools to investigate the longitudinal uniformity of the HC-PBGFs and to investigate possible defects (their onset, evolution and decay) were demonstrated, and some of these tools (e.g. X-ray computational tomography) can be applied not just to the fibres but also to the first and second stage preforms, providing a very powerful way to investigate the origin of such defects. As a result of such improvements, we were able to fabricate record lengths of low loss HC-PBGFs, including in particular an 11km long fibre operating at 1.55µm with ~5dB/km loss and numerous km long lengths at 2µm with very wide BW and minimum loss consistently in the region of 2dB/km (record length ~3dB/km). In our effort to further reduce the loss of measured fibres, we have implemented procedures for accurate control of the core ring surround which is critical to realise both broad bandwidth and low loss. Fibre Amplifiers Having demonstrated a fully functional erbium doped fibre amplifier (EDFA) supporting 4 mode groups (12 spatial and polarization modes) in year 3, the primary focus for year 4 was to demonstrate a few-mode thulium doped fibre amplifier (TDFA) at 2µm. A cladding pumped TDFA supporting 2 mode groups was demonstrated with over 15dB gain and NF as low as 6dB with low differential modal gains (<2 db) over 195nm-25nm as shown in Fig. 2(a). Gain at Publishable summary Page 3 of 6

4 Gain [db] NF [db] Gain (db) NF (db) the longer wavelength can be increased by optimizing the amplifier design whilst the NF could be reduced by implementing a core pump approach LP1 LP S- (FL) S (FL) S (LD) S+(LD) C (LD) L (LD) Wavelength [nm] Wavelength (nm) Fig. 2. (a) Measured modal gain and noise figure (NF) of the cladding pumped TDFA. (b) Small-signal (-2 dbm) gain and noise figure (NF) of the TDFA incorporating different lengths of fibres and different cavity architectures. We have further extended the gain of the thulium doped fibre amplifier (TDFA) at the shorter wavelength edge of the thulium emission band in an attempt to bridge the gap between the EDFA and TDFA. In fact, this is the first demonstration of a silica-based TDFA operating in the µm waveband and which was previously inaccessible by any kind of silica-based rare-earth doped fibre amplifier. A 15dB gain bandwidth over a bandwidth of 5.3THz was achieved which is comparable with the 4.4 THz bandwidth of the C-band. Overall, we have now demonstrated TDFAs with spanning a gain bandwidth extending from 165nm-25nm or 4nm in total by using different lengths of fibres and different cavity architectures. Using the 2dB small signal gain as a benchmark, we have successfully demonstrated silica-based TDFAs operating from 1675nm-225nm or 35nm, which is a factor of 3 wider (in wavelength terms) than the combined S, C and L band EDFAs. Just recently we have extended the gain to longer wavelength using a holmium doped fibre, achieving high gain and low noise figure out to wavelength as 215nm. Multiplexing/Demultiplexing To support the transmission studies shown above, a customised and compact 3D waveguide multiplexer was designed to simultaneously spot-launch all spatial channels into the FM-MCF. Accordingly, the waveguides in the mode multiplexers were formed in a 5.3 mm x 1 mm borosilicate glass substrate by direct laser writing using focused ultrafast femtosecond laser pulses. The inscription technique produces controllable sub-surface refractive index modification and allows the required 3D pattern of transparent waveguides to be carefully controlled to ±5 nm. The 21 single mode fibres inputs with a 127 m pitch v-groove were attached to waveguides assigned in 7 sets of 3 waveguides and inscribed in a hexagonal arrangement of 8 µm diameter to match the core arrangement and structure of the FM-MCF. The individual square waveguides have a cross-sectional effective area of 36 µm 2, as depicted in Fig. 3b, and each set of 3 waveguides was placed in a triangular arrangement. This arrangement minimizes insertion losses, whilst equally exciting the LP 1 and LP 11 degenerate modes in each core to minimize mode dependent loss (MDL). The MDL is approximated at 1.5 to 2 db, and the insertion loss on average is 1.1 db across all 21 waveguides (excluding fibre) at 155 nm. The waveguides are designed to minimize polarisation dependent loss, and were measured to be <.2 db, which is incorporated in the MDL approximation. The compact nature of the 3D waveguide allows a highly stable butt-coupled interface to the FM-MCF end-facets, without requiring additional bulky imaging optics. A 3D waveguide was used as both the spatial multiplexer and demultiplexer in this experimental setup. The total end-to-end loss measured after transmission is 12 db (including multiplexer and demultiplexer 3D waveguide), which is in-line with single core few-mode results. Publishable summary Page 4 of 6

5 Fig: 3 (a) FM-MC fibre cross section, (b) schematic diagram of the 3D waveguide, where sets of 3 transparent waveguides are placed in a triangular arrangement to address respective few mode cores. c, 3D waveguide FM-MCF facet microscope image All-fibre Photonic Lanterns for 3 and 6 mode multiplexing were further developed following year 3 activity. The primary aim has been to optimise the mode matching between lantern and output FMF. Several mode adapter techniques were investigated and provided good loss and MDL performance for both lantern types. These solid state fibre lanterns have been fully packaged and production processes and prototypes developed. MIMO Processing In the digital domain, substantial progress was made this year in coding and equalization. Novel optical transmission coding schemes have been experimentally demonstrated, including 8- dimensional constellation transmission. This work shows that the proposed E8 constellation outperforms the conventional 8QAM constellation by.8 db at the HD-FEC limit. The performance of the proposed E8 constellation is similar to the performance of TCM. However, both the proposed E8 constellation and TCM have a more complex encoder/decoder than the conventional 8QAM constellation. At the HD-FEC limit the proposed E8 constellation outperforms the 8QAM by.8 db SNR. Novel coding schemes are envisaged to be vital for future multi-mode transmission systems particularly as more modes 4LP, 6LP or 9LP are propagated with multi-mode amplifiers. Another key issue very important as more modes are transmitted is the computational complexity scaling of the MIMO equalizer for a combined coupled and uncoupled transmission system. The work undertaken indicates that a conventional time domain equalizer is not the preferred choice of equalization for the coupled transmission case, as the computational complexity scales linearly with the number of receivers, and linearly with the impulse response length. In contrast to the time domain equalizer, the frequency domain equalizer scales linearly with the number of received channels, and logarithmically with the impulse response. As is intuitively clear, uncoupled transmission results in the same computational complexity as a single transmission system Transmitters & receivers A suite of high performance active components required for the first DWDM system demonstration at 2 m over PBGF has been developed. The components developed cover lasers, photodetectors, modulators and functional passive components such as optical hybrids and mux/demux components. The first quantum confined Stark effect (QCSE) based Mach Zehnder modulator (MZM) operating around 2 nm has been achieved. The polarization sensitive modulators consisted of 15 compressively strained quantum wells and achieved a bandwidth of 1 GHz, with an extinction ratio of 9 db, and a V L ~ 9.6 Vmm. We demonstrated 1 Gbps back-to-back communication around 2 nm using MODEGAP lasers and amplifiers. Publishable summary Page 5 of 6

6 Power (db) Normalized Transmission mm (b) Reverse Bias(V) (a) (b) Fig. 4 (a): Optical transmission characteristics of modulator as the function of DC voltage supplied to one arm of the interferometer. (b): Dual-electrode operation of modulator: Inset Measured EO response of 2 mm TWE MZM at a bias voltage of 6.74 V. Inset: Measured eye diagram at 1 Gbps for PRBS signal The design, growth and fabrication of high bandwidth surface normal photodetectors were improved. Thin buffer layers were engineered to reduce the defect density in the relaxed absorbing layer grown on an InP substrate. Non-radiative recombination was reduced by cladding the absorber with a wide bandgap material and by optimising the mesa etching and passivation processes. 1GHz bandwidth and 15Gbps performance were achieved by reducing the carrier transit time by the use of a thinner absorber and by optimising p-doping levels to promote carrier transport. Fully packaged 2 m surface normal p-i-n photodiode were realised. Passive components were developed to provide waveguiding, splitting, optical hybrid and mux/demux functionality. A large spot size waveguide facilitated easy fibre coupling. 1 and 2 channel arrayed waveguide gratings were demonstrated Wavelength (nm) Fig. 5 Optical microscope image of back-to-back AWGs (left) and output from the 2 channels normalised to a straight waveguide Final results and expected impact The project has met its original objectives and has in many cases exceeded them, including the first field trial over live networks including FMF fibre links. In addition to the extensive leading research results, commercial products have emerged from the work that will enable researchers around the world to investigate this technology as a possible solution to the potential capacity crunch. Successful achievement of the anticipated results will have major implications for future transmission systems enabling the development of future networks capable of substantially higher information carrying capacity than achievable with single mode fibre solutions. This will have obvious benefits in terms of cost and ability to supply society s constantly increasing demand for information capacity. Publishable summary Page 6 of 6