Boris Karanov

University College London | UCL · Department of Electronic and Electrical Engineering

Short reach optical fiber communications rely on the intensity modulation/direct detection (IM/DD) technology to enable simple and inexpensive solutions in many data center, metro and access network scenarios. The IM/DD links are severely impaired by fiber dispersion as well as non-linearity and noise stemming from the low-cost transmitter and rece...

We have optimized 4D modulation formats up to 12 bits/symbol with hybrid probabilistic and geometric shaping. For high-order cardinalities, the results suggest that hybrid shaping is necessary.

In this paper we carry out a joint optimization of probabilistic (PS) and geometric shaping (GS) for four-dimensional (4D) modulation formats in long-haul coherent wavelength division multiplexed (WDM) optical fiber communications using an auto-encoder framework. We propose a 4D 10 bits/symbol constellation which we obtained via end-to-end deep lea...

Strategies for optimization of 100 Gb/s neural network-based transceivers are investigated for ensuring tolerance to transmission link variations. Exploiting the robustness/error rate trade-off, operation below HD-FEC is achieved for a wide range of distances without transceiver reconfiguration.

Conventional communication systems consist of several signal processing blocks, each performing an individual task at the transmitter or receiver, e.g. coding, modulation, or equalisation. However, there is a lack of optimal, computationally feasible algorithms for nonlinear fibre communications as most techniques are based upon classical communica...

An advanced digital backward-propagation (DBP) method using a separate-channels approach (SCA) is investigated for the compensation of inter-channel nonlinearities in spatial- and wavelength-multiplexed systems. Compared to the conventional DBP, intra- and inter-mode cross-phase modulation can be efficiently compensated by including the effect of t...

Fiber-optic auto-encoders are demonstrated on an intensity modulation/direct detection testbed, outperforming state-of-the-art signal processing. Algorithms for end-to-end optimization using experimentally collected data are discussed. The end-to-end learning framework is extended for performing optimization of the symbol distribution in probabilis...

Optical fiber networks form the major part of the modern telecommunication infrastructure and are carrying over 95% of the Internet digital data. The increasing demand for higher data rates has promoted the development of higher-order modulation formats, denser wavelength multiplexing and more powerful signal processing. With the entire suppression...

We investigate methods for experimental performance enhancement of auto-encoders based on a recurrent neural network (RNN) for communication over dispersive nonlinear channels. In particular, our focus is on the recently proposed sliding window bidirectional RNN (SBRNN) optical fiber auto-encoder. We show that adjusting the processing window in the...

Performance for optical fibre transmissions can be improved by digitally reversing the channel environment. When this is achieved by simulating short segment by separating the chromatic dispersion and Kerr nonlinearity, this is known as digital back-propagation (DBP). Time-domain DBP has the potential to decrease the complexity with respect to freq...

We investigate end-to-end optimized optical transmission systems based on feedforward or bidirectional recurrent neural networks (BRNN) and deep learning. In particular, we report the first experimental demonstration of a BRNN auto-encoder, highlighting the performance improvement achieved with recurrent processing for communication over dispersive...

We investigate methods for experimental performance enhancement of auto-encoders based on a recurrent neural network (RNN) for communication over dispersive nonlinear channels. In particular, our focus is on the recently proposed sliding window bidirectional RNN (SBRNN) optical fiber autoencoder. We show that adjusting the processing window in the...

We investigate end-to-end optimized optical transmission systems based on feedforward or bidirectional recurrent neural networks (BRNN) and deep learning. In particular, we report the first experimental demonstration of a BRNN auto-encoder, highlighting the performance improvement achieved with recurrent processing for communication over dispersive...

We perform an experimental end-to-end transceiver optimization via deep learning using a generative adversarial network to approximate the test-bed channel. Previously, optimization was only possible through a prior assumption of an explicit simplified channel model.

We present the first experimental demonstration of learned time-domain digital back-propagation (DBP), in 64-GBd dual-polarization 64-QAM signal transmission over 1014 km. Performance gains were comparable to those obtained with conventional, higher complexity, frequency-domain DBP.

We perform an experimental end-to-end transceiver optimization via deep learning using a generative adversarial network to approximate the test-bed channel. Previously, optimization was only possible through a prior assumption of an explicit simplified channel model.

In this paper, we apply deep learning for communication over dispersive channels with power detection, as encountered in low-cost optical intensity modulation/direct detection (IM/DD) links. We consider an autoencoder based on the recently proposed sliding window bidirectional recurrent neural network (SBRNN) design to realize the transceiver for o...

We propose an autoencoding sequence-based transceiver for communication over dispersive channels with intensity modulation and direct detection (IM/DD), designed as a bidirectional deep recurrent neural network (BRNN). The receiver uses a sliding window technique to allow for efficient data stream estimation. We find that this sliding window BRNN (...

We propose an autoencoding sequence-based transceiver for communication over dispersive channels with intensity modulation and direct detection (IM/DD), designed as a bidirectional deep recurrent neural network (BRNN). The receiver uses a sliding window technique to allow for efficient data stream estimation. We find that this sliding window BRNN (...

We demonstrate the first multi-terabit/s WDM data transmission through hollow-core antiresonant fiber (HC-ARF). 16 channels of 32-GBd dual-polarization (DP) Nyquist-shaped 256QAM signal channels were transmitted through a 270-m long fiber without observing any power penalty. In a single-channel high power transmission experiment, no nonlinearity pe...

We experimentally demonstrate an IM-DD system relying on deep neural networks from transmitter to receiver delivering 42 Gb/s over 20 and 40 km at 1550 nm below 3.8×10−3 . The ANN is trained to tolerate deviations in dispersion by as much as ±170ps/nm

The impact of spectral broadening effects on the performance of nonlinear compensation applied to both modern dispersion-unmanaged and legacy dispersion-managed optical communication systems has been analysed and quantified. It is found that including the full broadened spectrum at the receiver enables a substantial improvement in the performance o...

The impact of spectral broadening on the performance of nonlinear compensation applied to legacy submarine dispersion-managed links is studied. An additional 2.2 dB SNR improvement at optimum launch power is achieved by optimizing the compensated bandwidth.

In this paper, we implement an optical fiber communication system as an end-to-end deep neural network, including the complete chain of transmitter, channel model, and receiver. This approach enables the optimization of the transceiver in a single end-to-end process. We illustrate the benefits of this method by applying it to intensity modulation/d...

Meeting the ever-growing information rate demands has become of utmost importance for optical communication systems. However, it has proven to be a challenging task due to the presence of Kerr effects, which have largely been regarded as a major bottleneck for enhancing the achievable information rates in modern optical communications. In this work...

Meeting the ever-growing information rate demands has become of utmost importance for optical communication systems. However, it has proven to be a challenging task due to the presence of Kerr effects, which have largely been regarded as a major bottleneck for enhancing the achievable information rates in modern optical communications. In this work...

The optimisation of span length when designing optical communication systems is important from both performance and cost perspectives. In this paper, the optimisation of inter-amplifier spacing and the potential increase of span length at fixed information rates in optical communication systems with practically feasible nonlinearity compensation sc...

Nyquist-spaced transmission and digital signal processing have proved effective in maximising the spectral efficiency and reach of optical communication systems. In these systems, Kerr nonlinearity determines the performance limits, and leads to spectral broadening of the signals propagating in the fibre. Although digital nonlinearity compensation...

Nonlinearity compensation and achievable information rates were investigated in fully-loaded C-band communication systems considering transceiver limitations. It is found that the efficacy of nonlinearity compensation in enhancing the achievable information rates depends on the modulation formats and transmission distances.

The first multi-terabit/s WDM data transmission through anti-resonant hollow-core fiber is demonstrated