Jochen Schröder
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Jochen Schröder has a total of 222 co-authors. Here are the top co-authors by number of shared publications:
Benjamin J. Eggleton
Centre for Ultrahigh Bandwidth Devices for Optical Systems (CUDOS), Institute of Photonics and Optical Science (IPOS), School of Physics, The University of Sydney, New South Wales 2006, Australia
47 shared publications- Magnus Karlsson
Chalmers University of Technology
20 shared publications - Barry Luther-Davies
The Australian National Univ. (Australia)
18 shared publications - Duk-Yong Choi
Australian National University
18 shared publications - Peter A. Andrekson
Chalmers University of Technology
17 shared publications Trung D. Vo
Centre for Ultrahigh bandwidth Devices for Optical Systems (CUDOS), Institute of Photonics and Optical Science (IPOS) School of Physics A28, University of Sydney, NSW 2006, Australia
15 shared publications
Latest Publications
- Open Access
- 13 November 2026
- Published by Wiley in Nanophotonics
- Vol. 15 (22)
High‐Frequency Stimulated Brillouin Scattering in Suspended Thin‐Film Lithium Niobate Waveguides
Stimulated Brillouin scattering (SBS) enables photonic filtering at gigahertz center frequencies with narrow linewidths at megahertz‐range, making it a key building block for integrated microwave photonics. Accessing radio‐frequency bands beyond 15 GHz with SBS requires high frequency shifts and high gain coefficients, a combination that yet remains a challenge. Here, we show that SBS in suspended Z‐cut thin‐film lithium niobate (TFLN) waveguides satisfies these demands, exhibiting Brillouin frequency shifts exceeding 15 GHz when pumped at telecom wavelengths, nearly twice the highest values previously reported in TFLN. We observe gain coefficients up to 23.8 , among the largest achieved for such high Brillouin frequency shifts in any integrated Brillouin platform to date. For the first time, we apply spontaneous Brillouin microspectroscopy to suspended TFLN waveguides and confirm the high‐frequency acoustic mode observed in our SBS experiments. Our results unlock TFLN as a compelling material platform for scalable microwave photonics, with future applications spanning communication, sensing, and information processing.
- 13 November 2026
- Published by Wiley in Nanophotonics
- Vol. 15 (22)
Open AccessHigh‐Frequency Stimulated Brillouin Scattering in Suspended Thin‐Film Lithium Niobate Waveguides
Stimulated Brillouin scattering (SBS) enables photonic filtering at gigahertz center frequencies with narrow linewidths at megahertz‐range, making it a key building block for integrated microwave photonics. Accessing radio‐frequency bands beyond 15 GHz with SBS requires high frequency shifts and high gain coefficients, a combination that yet remains a challenge. Here, we show that SBS in suspended Z‐cut thin‐film lithium niobate (TFLN) waveguides satisfies these demands, exhibiting Brillouin frequency shifts exceeding 15 GHz when pumped at telecom wavelengths, nearly twice the highest values previously reported in TFLN. We observe gain coefficients up to 23.8 , among the largest achieved for such high Brillouin frequency shifts in any integrated Brillouin platform to date. For the first time, we apply spontaneous Brillouin microspectroscopy to suspended TFLN waveguides and confirm the high‐frequency acoustic mode observed in our SBS experiments. Our results unlock TFLN as a compelling material platform for scalable microwave photonics, with future applications spanning communication, sensing, and information processing.
- Open Access
- 1 September 2025
- Published by AIP Publishing in APL Photonics
- Vol. 10 (9)
Suspended Z-cut lithium niobate waveguides for stimulated Brillouin scattering
On-chip stimulated Brillouin scattering (SBS) has recently been demonstrated in thin-film lithium niobate (TFLN), an emerging material platform for integrated photonics offering large electro-optic and nonlinear properties. While previous works on SBS in TFLN have focused on surface SBS, in this contribution we experimentally demonstrate, for the first time, backward intra-modal SBS generation in suspended Z-cut TFLN waveguides. Our results show trapping of multiple acoustic modes in this structure, featuring a multi-peak Brillouin gain spectrum due to the excitation of higher-order acoustic modes. The findings expand the TFLN waveguide platform exploration for SBS interactions and provide a crucial step toward realizing advanced optical processors for sensors or microwave signals integrated on TFLN.
- 1 September 2025
- Published by AIP Publishing in APL Photonics
- Vol. 10 (9)
Open AccessSuspended Z-cut lithium niobate waveguides for stimulated Brillouin scattering
On-chip stimulated Brillouin scattering (SBS) has recently been demonstrated in thin-film lithium niobate (TFLN), an emerging material platform for integrated photonics offering large electro-optic and nonlinear properties. While previous works on SBS in TFLN have focused on surface SBS, in this contribution we experimentally demonstrate, for the first time, backward intra-modal SBS generation in suspended Z-cut TFLN waveguides. Our results show trapping of multiple acoustic modes in this structure, featuring a multi-peak Brillouin gain spectrum due to the excitation of higher-order acoustic modes. The findings expand the TFLN waveguide platform exploration for SBS interactions and provide a crucial step toward realizing advanced optical processors for sensors or microwave signals integrated on TFLN.
- Open Access
Tightly-confined and long Z-cut lithium niobate waveguide with ultralow-loss
Lithium niobate (LN) is a promising material for future complex photonic-electronic circuits, with wide applications in fields like data communications, sensing, optical computation, and quantum optics. There was a great step toward LN photonic integrated circuits (PICs) with the development of dry etching for low-loss LN on insulator (LNOI) waveguides. However, the versatility of the LN waveguide platform for applications like $χ^3$ nonlinear devices and passive phase sensitive components, has not been fully utilized. The main challenges are the difficulty of making highly confined ultralow-loss waveguides and overcoming the strong material birefringence. Here, we developed a fabrication technology for an ultralow-loss, tightly-confined, dispersion-engineered LN waveguide. We demonstrated an ultra-low propagation loss of 5.8 dB/m in a decimeter-long LN spiral waveguide. We focused on Z-cut LN waveguides with TE mode to avoid the material birefringence. Aiming for $χ^3$ nonlinear applications, we demonstrated the first all normal-dispersion (ANDi) based coherent octave-spanning supercontinuum frequency comb in integrated LN waveguide. Our ultralow-loss Z-cut LN long waveguide might be useful in on-chip narrow linewidth lasers, optical delay lines, and parametric amplifiers.Open AccessTightly-confined and long Z-cut lithium niobate waveguide with ultralow-loss
Lithium niobate (LN) is a promising material for future complex photonic-electronic circuits, with wide applications in fields like data communications, sensing, optical computation, and quantum optics. There was a great step toward LN photonic integrated circuits (PICs) with the development of dry etching for low-loss LN on insulator (LNOI) waveguides. However, the versatility of the LN waveguide platform for applications like $χ^3$ nonlinear devices and passive phase sensitive components, has not been fully utilized. The main challenges are the difficulty of making highly confined ultralow-loss waveguides and overcoming the strong material birefringence. Here, we developed a fabrication technology for an ultralow-loss, tightly-confined, dispersion-engineered LN waveguide. We demonstrated an ultra-low propagation loss of 5.8 dB/m in a decimeter-long LN spiral waveguide. We focused on Z-cut LN waveguides with TE mode to avoid the material birefringence. Aiming for $χ^3$ nonlinear applications, we demonstrated the first all normal-dispersion (ANDi) based coherent octave-spanning supercontinuum frequency comb in integrated LN waveguide. Our ultralow-loss Z-cut LN long waveguide might be useful in on-chip narrow linewidth lasers, optical delay lines, and parametric amplifiers. - 4 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Photonics Technology Letters
- Vol. 36 (23) , 1349-1352
Perturbation-Based Joint SPM and XPM Compensation for Superchannel System
In contrast to digital backpropagation (DBP), perturbation-based nonlinear compensation (PB-NLC) is a low-complexity alternative to mitigate fiber Kerr nonlinearity. In this paper, we experimentally demonstrate a novel receiver-side perturbation approach to cancel the self-phase modulation and cross-phase modulation for superchannel systems using three independent receivers. With the inverse perturbation theory, we develop a nonlinear compensation model that does not require knowing the transmitted symbols and therefore avoids the penalty from the estimation error. We implement the PB-NLC in a 3×24.5 GBaud 64-QAM comb-based superchannel system spaced at 25 GHz. Compared to chromatic dispersion compensation, the full PB-NLC achieves a 0.2 dB Q 2 factor gain after 1200 km transmission, which is equivalent to the single-channel DBP operating at 1 step per span.
- 4 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Photonics Technology Letters
- Vol. 36 (23) , 1349-1352
Perturbation-Based Joint SPM and XPM Compensation for Superchannel System
In contrast to digital backpropagation (DBP), perturbation-based nonlinear compensation (PB-NLC) is a low-complexity alternative to mitigate fiber Kerr nonlinearity. In this paper, we experimentally demonstrate a novel receiver-side perturbation approach to cancel the self-phase modulation and cross-phase modulation for superchannel systems using three independent receivers. With the inverse perturbation theory, we develop a nonlinear compensation model that does not require knowing the transmitted symbols and therefore avoids the penalty from the estimation error. We implement the PB-NLC in a 3×24.5 GBaud 64-QAM comb-based superchannel system spaced at 25 GHz. Compared to chromatic dispersion compensation, the full PB-NLC achieves a 0.2 dB Q 2 factor gain after 1200 km transmission, which is equivalent to the single-channel DBP operating at 1 step per span.
- 18 July 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 42 (23) , 8366-8373
Dynamic Model for Coupled-Core Fibers
Capacity trends in fiber optical communication systems are closely related to space-division multiplexing (SDM) solutions. A coupled-core fiber (CCF) is one of the most promising SDM candidates for a high data rate transmission because of its tolerance to nonlinear effects and reduced accumulation of group delay spread (GDS) with distance. As the modes of the cores in CCFs are strongly coupled, a communication link requires multiple-input multiple-output (MIMO) processing to recover the input signals. MIMO algorithms, in turn, operate with the different number of finite impulse response taps that represents the memory length of an equalizing filter. The calculation of the number of taps requires the knowledge of GDS, which can be estimated from the received data if a channel model is given. Moreover, if the model accounts for dynamic (time varying) effects it can be used to test the adaptive performance of MIMO equalizer. In this work we propose a straightforward dynamic random coupling model for CCFs. We investigate the dynamic behavior of group delays and impulse responses and also derive and analyze frequency and time autocorrelation functions.
- 18 July 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 42 (23) , 8366-8373
Dynamic Model for Coupled-Core Fibers
Capacity trends in fiber optical communication systems are closely related to space-division multiplexing (SDM) solutions. A coupled-core fiber (CCF) is one of the most promising SDM candidates for a high data rate transmission because of its tolerance to nonlinear effects and reduced accumulation of group delay spread (GDS) with distance. As the modes of the cores in CCFs are strongly coupled, a communication link requires multiple-input multiple-output (MIMO) processing to recover the input signals. MIMO algorithms, in turn, operate with the different number of finite impulse response taps that represents the memory length of an equalizing filter. The calculation of the number of taps requires the knowledge of GDS, which can be estimated from the received data if a channel model is given. Moreover, if the model accounts for dynamic (time varying) effects it can be used to test the adaptive performance of MIMO equalizer. In this work we propose a straightforward dynamic random coupling model for CCFs. We investigate the dynamic behavior of group delays and impulse responses and also derive and analyze frequency and time autocorrelation functions.
- 20 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 42 (14) , 4829-4836
12.2 Bit/s/Hz C-Band Transmission With High-Gain Low-Complexity 24-Dimensional Geometric Shaping
Multidimensional (MD) modulation formats enable a larger minimum Euclidean distance than conventional two-dimensional formats. As a structured geometric shaping method, MD Voronoi constellations (VCs) avoid the use of look-up tables and can be implemented with low-complexity encoding and decoding algorithms, outperforming quadrature amplitude modulation (QAM) formats in terms of mutual information. However, it is challenging to maintain the shaping gain in a practical system requiring bit-mapping and forward error correction (FEC) coding. By using a hybrid labeling and multilevel coding scheme integrated with soft-decision FEC coding, MD VCs can achieve high shaping gain compared to QAM formats after soft-decision (SD) decoding. In practical systems, transceiver impairments significantly degrade the system performance, especially for high-cardinality constellations. To investigate of performance of high spectral-efficiency (SE) MD VCs, we employ a 24-dimensional (24D) VC with uncoded SE of 8 bit/symbol/dimension-pair to a spectral superchannel spanning the full C-band. Specifically, the 24D VC with a record constellation size of 7.9×10 28 is applied to dual-polarization coherent transmission over six time slots. The superchannel is composed of 5.0 Gbaud channels spaced at 5.2 GHz, resulting in an overall spectral efficiency of 12.2 bit/s/Hz and a net throughput of 54.2 Tb/s over the entire C-band after 40 km single-mode fiber. It is the first experimental demonstration of a high-SE MD format providing a significant shaping gain after SD decoding.
- 20 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 42 (14) , 4829-4836
12.2 Bit/s/Hz C-Band Transmission With High-Gain Low-Complexity 24-Dimensional Geometric Shaping
Multidimensional (MD) modulation formats enable a larger minimum Euclidean distance than conventional two-dimensional formats. As a structured geometric shaping method, MD Voronoi constellations (VCs) avoid the use of look-up tables and can be implemented with low-complexity encoding and decoding algorithms, outperforming quadrature amplitude modulation (QAM) formats in terms of mutual information. However, it is challenging to maintain the shaping gain in a practical system requiring bit-mapping and forward error correction (FEC) coding. By using a hybrid labeling and multilevel coding scheme integrated with soft-decision FEC coding, MD VCs can achieve high shaping gain compared to QAM formats after soft-decision (SD) decoding. In practical systems, transceiver impairments significantly degrade the system performance, especially for high-cardinality constellations. To investigate of performance of high spectral-efficiency (SE) MD VCs, we employ a 24-dimensional (24D) VC with uncoded SE of 8 bit/symbol/dimension-pair to a spectral superchannel spanning the full C-band. Specifically, the 24D VC with a record constellation size of 7.9×10 28 is applied to dual-polarization coherent transmission over six time slots. The superchannel is composed of 5.0 Gbaud channels spaced at 5.2 GHz, resulting in an overall spectral efficiency of 12.2 bit/s/Hz and a net throughput of 54.2 Tb/s over the entire C-band after 40 km single-mode fiber. It is the first experimental demonstration of a high-SE MD format providing a significant shaping gain after SD decoding.
- 6 October 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 42 (2) , 793-801
Modeling of 3-Coupled-Core Fiber: Comparison Between Scalar and Vector Random Coupling Models
Simple, fast and accurate channel models can assist not only in a process of manufacturing of novel multi-core fibers, but also in a communication link design, providing a convenient way to investigate different properties of fibers that could be appealing for transmission. Models for coupled-core fibers are in a great demand since this type of the fiber provides advantages for data transmission, e.g. by reducing the effects of modal dispersion and fiber nonlinearity. In this work we present a comparative study of key features in a three coupled-core fiber (3CCF) for scalar and vector random coupling models. Supermodes and their group delays in an unperturbed 3CCF with polarization mode mixing are comprehensively analyzed for the first time. It is shown that birefringence does not impact significantly the group delay values but it may affect the mode mixing. The intensity impulse response and its RMS width were then investigated in a 3CCF with perturbations and compared for two types of models. The shape of the averaged intensity impulse response is found to be very similar for two models and retained for various input polarization states, while in the absence of averaging, the shape of the impulse response strongly depends on the input state of polarization. We show that the calculated RMS widths of the intensity impulse response agree well with theoretical values, as well as between scalar and vector cases.
- 6 October 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 42 (2) , 793-801
Modeling of 3-Coupled-Core Fiber: Comparison Between Scalar and Vector Random Coupling Models
Simple, fast and accurate channel models can assist not only in a process of manufacturing of novel multi-core fibers, but also in a communication link design, providing a convenient way to investigate different properties of fibers that could be appealing for transmission. Models for coupled-core fibers are in a great demand since this type of the fiber provides advantages for data transmission, e.g. by reducing the effects of modal dispersion and fiber nonlinearity. In this work we present a comparative study of key features in a three coupled-core fiber (3CCF) for scalar and vector random coupling models. Supermodes and their group delays in an unperturbed 3CCF with polarization mode mixing are comprehensively analyzed for the first time. It is shown that birefringence does not impact significantly the group delay values but it may affect the mode mixing. The intensity impulse response and its RMS width were then investigated in a 3CCF with perturbations and compared for two types of models. The shape of the averaged intensity impulse response is found to be very similar for two models and retained for various input polarization states, while in the absence of averaging, the shape of the impulse response strongly depends on the input state of polarization. We show that the calculated RMS widths of the intensity impulse response agree well with theoretical values, as well as between scalar and vector cases.
- 10 August 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 42 (1) , 48-56
Inter-Channel Interference Cancellation for Long-Haul Superchannel System
Optical frequency combs enable superchannel systems with dense channel spacing by providing equally-spaced lines to overcome the relative frequency drifts of free-running lasers, achieving high spectral efficiency. However, strong inter-channel interference is induced in densely spaced systems due to imperfect pulse-shaping, preventing the further improvement of system throughput. By including the spectra information from the adjacent channels, comb-based multi-channel equalization has been previously validated in a back-to-back and 80 km transmission system for linear crosstalk mitigation. In this case, the limiting factor of the system was the effective number of bits (ENOB). Here, we investigate the influence of amplified spontaneous emission noise and the effectiveness of joint equalization in long-haul transmission systems. We show that the performance of the multi-channel equalizer (MCE) is significantly degraded when reducing the optical-signal-to-noise ratio (OSNR), especially for high-order modulation formats, in both noise-loading and long-distance transmission experiments. Moreover, the ASE noise from the neighboring channels can be coupled into the systems by the MCE, degrading the performance improvement. We find that a large-tap MCE can maximize the system performance in the high OSNRs, whereas MCE with a smaller tap should be applied in low OSNRs to alleviate the effect of ASE noise. After transmission of 800 km, an achievable information rate (AIR) increase of more than 0.1 bits/s/Hz is enabled by the optimized MCE for 16-QAM and 64-QAM. Our results confirm the validity of the joint equalization gain for short-distance transmission but also clearly demonstrate the performance penalty originating from transmission optical noise. Moreover, the equalizer tap number needs to be optimized for each OSNR to maximize the performance of joint equalization. We believe this limitation is also valid for other impairments such as optical noise in the comb and electrical noise in the transceiver.
- 10 August 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 42 (1) , 48-56
Inter-Channel Interference Cancellation for Long-Haul Superchannel System
Optical frequency combs enable superchannel systems with dense channel spacing by providing equally-spaced lines to overcome the relative frequency drifts of free-running lasers, achieving high spectral efficiency. However, strong inter-channel interference is induced in densely spaced systems due to imperfect pulse-shaping, preventing the further improvement of system throughput. By including the spectra information from the adjacent channels, comb-based multi-channel equalization has been previously validated in a back-to-back and 80 km transmission system for linear crosstalk mitigation. In this case, the limiting factor of the system was the effective number of bits (ENOB). Here, we investigate the influence of amplified spontaneous emission noise and the effectiveness of joint equalization in long-haul transmission systems. We show that the performance of the multi-channel equalizer (MCE) is significantly degraded when reducing the optical-signal-to-noise ratio (OSNR), especially for high-order modulation formats, in both noise-loading and long-distance transmission experiments. Moreover, the ASE noise from the neighboring channels can be coupled into the systems by the MCE, degrading the performance improvement. We find that a large-tap MCE can maximize the system performance in the high OSNRs, whereas MCE with a smaller tap should be applied in low OSNRs to alleviate the effect of ASE noise. After transmission of 800 km, an achievable information rate (AIR) increase of more than 0.1 bits/s/Hz is enabled by the optimized MCE for 16-QAM and 64-QAM. Our results confirm the validity of the joint equalization gain for short-distance transmission but also clearly demonstrate the performance penalty originating from transmission optical noise. Moreover, the equalizer tap number needs to be optimized for each OSNR to maximize the performance of joint equalization. We believe this limitation is also valid for other impairments such as optical noise in the comb and electrical noise in the transceiver.
- 5 April 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 41 (17) , 5557-5563
Physical Realizations of Multidimensional Voronoi Constellations in Optical Communication Systems
Multidimensional geometric shaping has been shown to outperform uniform quadrature amplitude modulation (QAM) in optical communication systems but the complexity of symbol decision and bit mapping can often be significant as dimensionality increases. In this article, a low-complexity geometric shaping method based on multidimensional lattices is investigated both in experiments and simulations. The modulation formats designed based on this method are called Voronoi constellations (VCs) and we study them in 8, 16, and 32 dimensions. We obtain transmission reach improvements of up to 22 and 70% for VCs compared to 4 QAM and 16 QAM, respectively, in nonlinear long-haul fiber transmission. Moreover, we compare different physical realizations of multidimensional VCs over wavelengths, polarizations, and time slots in both the Gaussian and nonlinear fiber channels. We demonstrate that different physical realizations perform similarly in the fiber-optic back-to-back channel. However, in long-haul transmission systems, spreading the dimensions over time slots can increase the transmission reach up to 4% compared to wavelengths and polarizations. Furthermore, the mutual information and generalized mutual information are estimated and compared to QAM formats at the same spectral efficiencies.
- 5 April 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 41 (17) , 5557-5563
Physical Realizations of Multidimensional Voronoi Constellations in Optical Communication Systems
Multidimensional geometric shaping has been shown to outperform uniform quadrature amplitude modulation (QAM) in optical communication systems but the complexity of symbol decision and bit mapping can often be significant as dimensionality increases. In this article, a low-complexity geometric shaping method based on multidimensional lattices is investigated both in experiments and simulations. The modulation formats designed based on this method are called Voronoi constellations (VCs) and we study them in 8, 16, and 32 dimensions. We obtain transmission reach improvements of up to 22 and 70% for VCs compared to 4 QAM and 16 QAM, respectively, in nonlinear long-haul fiber transmission. Moreover, we compare different physical realizations of multidimensional VCs over wavelengths, polarizations, and time slots in both the Gaussian and nonlinear fiber channels. We demonstrate that different physical realizations perform similarly in the fiber-optic back-to-back channel. However, in long-haul transmission systems, spreading the dimensions over time slots can increase the transmission reach up to 4% compared to wavelengths and polarizations. Furthermore, the mutual information and generalized mutual information are estimated and compared to QAM formats at the same spectral efficiencies.
- 6 March 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 41 (11) , 3261-3277
End-to-End Learning for VCSEL-Based Optical Interconnects: State-of-the-Art, Challenges, and Opportunities
Optical interconnects (OIs) based on vertical-cavity surface-emitting lasers (VCSELs) are the main workhorse within data centers, supercomputers, and even vehicles, providing low-cost, high-rate connectivity. VCSELs must operate under extremely harsh and time-varying conditions, thus requiring adaptive and flexible designs of the communication chain. Such designs can be built based on mathematical models (model-based design) or learned from data (machine learning (ML) based design). Various ML techniques have recently come to the forefront, replacing individual components in the transmitters and receivers with deep neural networks. Beyond such component-wise learning, end-to-end (E2E) autoencoder approaches can reach the ultimate performance through co-optimizing entire parameterized transmitters and receivers. This tutorial paper aims to provide an overview of ML for VCSEL-based OIs, with a focus on E2E approaches, dealing specifically with the unique challenges facing VCSELs, such as the wide temperature variations and complex models.
- 6 March 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 41 (11) , 3261-3277
End-to-End Learning for VCSEL-Based Optical Interconnects: State-of-the-Art, Challenges, and Opportunities
Optical interconnects (OIs) based on vertical-cavity surface-emitting lasers (VCSELs) are the main workhorse within data centers, supercomputers, and even vehicles, providing low-cost, high-rate connectivity. VCSELs must operate under extremely harsh and time-varying conditions, thus requiring adaptive and flexible designs of the communication chain. Such designs can be built based on mathematical models (model-based design) or learned from data (machine learning (ML) based design). Various ML techniques have recently come to the forefront, replacing individual components in the transmitters and receivers with deep neural networks. Beyond such component-wise learning, end-to-end (E2E) autoencoder approaches can reach the ultimate performance through co-optimizing entire parameterized transmitters and receivers. This tutorial paper aims to provide an overview of ML for VCSEL-based OIs, with a focus on E2E approaches, dealing specifically with the unique challenges facing VCSELs, such as the wide temperature variations and complex models.