Impact of Feedback Dimension in Enhanced Reservoir Computing for IMDD Transmission Equalization

Transmission capacity and receiver sensitivity of an intensity-modulation direct detection (IM-DD) optical discrete multi-tone (DMT) system can be improved by using the probabilistically shaping (PS) technique. However, different probabilistic distributions will be required owing to the unbalanced signal-to-noise ratio (SNR) among data-carrying subcarriers (SCs) induced by the imperfect frequency response of optical/electrical devices, which can increase the implementation complexity of the PS-DMT transceiver. In this work, different signal pre-processing schemes including pre-equalization, Walsh–Hadamard transform (WHT)-based full data-carrying SCs precoding (FDSP) and fast WHT-based partial data-carrying SCs precoding (PDSP) are investigated for SNR equalization in a short-reach PS-DMT transmission system. After transmission over 50 km single-mode fiber, the experimental results indicated that three pre-processed signals have almost the same generalized mutual information (GMI) performance and receiver sensitivity improvements. The proposed fast WHT-based PDSP scheme may be a good option for the implementation of the PS-DMT transmission systems with a large SC SNR fluctuation regarding computational complexity.

In this paper, we propose and experimentally investigate a novel source separation chaotic optical communication architecture based on optical Reservoir Computing (RC). Benefiting from the RC's powerful high-dimensional mapping capabilities, the well-trained RC can replace the traditional chaotic receiver to achieve high-quality separation decryption without chaos synchronization, which significantly reduces costs and challenges. Specifically, the RC adopted in the proposed scheme has a deep cascade electro-optical feedback (DCEO) structure, which effectively enriches the reservoir state response. Numerical results show that the NMSE of the separated chaotic carrier is 0.0078 and the Q-factor of the separated OOK signal is 9.239, which achieves satisfactory decryption results. Simultaneously, the experimental and numerical conclusions are well verified with each other, which reflects the desirable reproducibility and reliability of the proposed scheme. Furthermore, the DCEO-RC system was compared with the other 3 RCs, and it had the lowest NMSE and the highest Q-factor in both experiments and numerical simulations, which represents the best separation performance. Subsequently, the impact of mask coefficient, SNR, chaotic system configurations, and RC parameters on separation performance was analyzed in detail. Finally, we analyze communication transmission quality. The results show that 20GBaud OOK/QPSK signals can achieve normal communication over 200 km. In summary, the proposed scheme is expected to provide strong support for high-speed chaotic secure communication.

Photonic reservoir computing (RC) enables us to introduce artificial neural networks (ANNs) as a deployable hardware-based processing unit beyond the von Neumann architecture. Its unique approach to consider only the output layer weights as adaptive, reduces training complexity and enables the use of ANNs in time-sensitive areas like optical communications. In this paper, we numerically investigate and evaluate the requirements of a photonic RC using an active silicon micro-ring resonator as a photonic integrated circuit (PIC) based reservoir system to compensate for transmission impairments such as chromatic dispersion and nonlinear effects occurring in optical transmission systems. We compare the performance of this RC with results from digital signal processing using Kramers-Kronig reception of distorted PAM-4 symbols from a 56 GBaud WDM transmission experiment over 100 km uncompensated SSMF.

Real-world applications such as first-person video activity recognition require intelligent edge devices. However, size, weight, and power constraints of the embedded platforms cannot support resource intensive state-of-the-art algorithms. Machine learning lite algorithms, such as reservoir computing, with shallow 3-layer networks are computationally frugal as only the output layer is trained. By reducing network depth and plasticity, reservoir computing minimizes computational power and complexity, making the algorithms optimal for edge devices. However, as a trade-off for their frugal nature, reservoir computing sacrifices computational power compared to state-of-the-art methods. A good compromise between reservoir computing and fully supervised networks are the proposed deep-LSM networks. The deep-LSM is a deep spiking neural network which captures dynamic information over multiple time-scales with a combination of randomly connected layers and unsupervised layers. The deep-LSM processes the captured dynamic information through an attention modulated readout layer to perform classification. We demonstrate that the deep-LSM achieves an average of 84.78% accuracy on the DogCentric video activity recognition task, beating state-of-the-art. The deep-LSM also shows up to 91.13% memory savings and up to 91.55% reduction in synaptic operations when compared to similar recurrent neural network models. Based on these results we claim that the deep-LSM is capable of overcoming limitations of traditional reservoir computing, while maintaining the low computational cost associated with reservoir computing.

This paper investigates the application of an erbium doped fibre amplifier (EDFA) in a coherent receiver using pulse position modulation (PPM) format. Such a combination can exploit the abundant monomode fibre bandwidth to offer improved sensitivity and improved multichannel selectivity through the exploitation of coherent transmission. Original theoretical results are presented for a system operating at a bit rate of 622 Mbit/s and a wavelength of 1.53 /spl mu/m. By using a suboptimal matched filter approach and assuming thermal noise dominant conditions, the results demonstrate that the system offers 7.5 dB sensitivity improvement over a homodyne PPM system and 9.4 dB improvement over PSK homodyne. The system investigated also offers a 10.4 dB sensitivity improvement over an equivalent optically preamplified direct detection (DD) PPM system. It was also demonstrated that by using an optical filter in order to reduce the ASE noise the improvement in receiver sensitivity was only 0.3 dB. This greatly reduces the requirement for optical filters in coherent systems and therefore the full bandwidth of the optical amplifiers can be used.

The main objective of this paper is to investigate the application of an erbium doped fiber amplifier (EDFA) in a coherent receiver using the pulse position modulation (PPM) format. Such a combination can exploit the abundant monomode fiber bandwidth to offer improved sensitivity and improved multichannel selectivity through the exploitation of coherent transmission. Original theoretical results are presented for a system operating in thermal noise dominant condition at a bit rate of 622 Mbit/s and a wavelength of 1.53 micrometer. By using a suboptimal matched filter the results demonstrate that the system offers 7.5 dB sensitivity improvement over a homodyne PPM system and 9.4 dB improvement over PSK homodyne. The system investigated also offers a 10.4 dB sensitivity improvement over an equivalent optically preamplified direct detection (DD) PPM system. It is also demonstrated that by using an optical filter in order to reduce the ASE noise due to EDFA the improvement in receiver sensitivity was only 0.3 dB. This greatly reduces the requirement for optical filters in coherent systems and therefore the full bandwidth of the optical amplifiers can be utilized.

A new version of Absolute Polar Duty Cycle Division Multiplexing transmission scheme over Wavelength Division Multiplexing system is proposed. We modeled and analyzed a method to improve the performance of AP-DCDM over WDM system by using Dual-Drive Mach–Zehnder-Modulator (DD-MZM). Almost 4.1 dB improvement in receiver sensitivity of 1.28 Tbit/s (32 × 40 Gbit/s) AP-DCDM-WDM over 320 km fiber is achieved by optimizing the bias voltage in DD-MZM.

We propose an enhanced IEEE 802.15.4 gateway for use in a hierarchical network architecture. We perform a system comparison between it and a gateway using a normal sensor mote. This comparison includes a PHY and MAC simulation benefiting from ray trace propagation modeling of a large office building. The proposed enhanced gateway offers a 21.2 dB improvement in receiver sensitivity and an ideal channel equaliser. We analyze these benefits before performing a full system simulation. We report improvements over a classical sensor gateway in all key areas of packet delivery, latency and mote energy efficiency. Just under a six-fold increase in indoors range is also reported. We observe a reduction in performance due to channel distortion and combat it by proposing a fractionally spaced equaliser. We show that for buildings with large RMS delay spreads - greater than 100ns (e.g. conference centres), significant benefit may be derived.

In this study, we propose a novel wavelength shift keying (WSK) technique that is combined with the conventional intensity modulation scheme for the transmission of point-to-point bidirectional data at the rate of 10 Gbps in each direction. We observe that WSK technique has not been investigated for implementation in point-to-point free space optical (FSO) links. Therefore, to the best of our knowledge, our study is the first one to perform this investigation. Our proposed link uses WSK in the downlink direction while we re-use the optical carriers for the transmission of uplink data. The use of WSK in the downlink direction enables us to perform balanced detection at the receiver, resulting in 3 dB improvement in receiver sensitivity compared to simple direct detection. We present bit error rate (BER) results for the signals transmitted in both the directions under different turbulence conditions and FSO link lengths. It was observed that the downlink signals generally perform better compared to the uplink due to the use of balanced detection and higher intensity fluctuations induced over the re-used optical carrier transmitted in the uplink direction.Please check and confirm that all the author names (given name, family name) and initials are correctly identified. The author names and initials are CORRECT.

A method is developed for calculating carrier-to-intermodulation product ratios for conventional modulators and for linearized modulators in which the third-order products are cancelled, and fifth-order products are dominant. Simple expressions are derived for the r.m.s. optical modulation index which optimizes system performance. The improvement in receiver sensitivity which can be achieved with linearized modulators is determined. For an AM-VSB system operating with carrier-to-noise ratios of 50 dB, linearized modulators are predicted to yield an improvement in receiver sensitivity of 8 dB. It is shown that for other SCM systems operating with CNRs of 25 dB or less, no performance improvement can be expected.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

The aim of this presentation is to show how various ideas coming from the nonlinear stability theory of functional differential systems, stochastic modeling, and machine learning, can be put together in order to create an approximating model that explains the working mechanisms behind a certain type of reservoir computers. Reservoir computing is a recently introduced brain-inspired machine learning paradigm capable of excellent performances in the processing of empirical data. We focus on time-delay based reservoir computers that have been physically implemented using optical and electronic systems and have shown unprecedented data processing rates. Reservoir computing is well-known for the ease of the associated training scheme but also for the problematic sensitivity of its performance to architecture parameters. The reservoir design problem is addressed, which remains the biggest challenge in the applicability of this information processing scheme. Our results use the information available regarding the optimal reservoir working regimes in order to construct a functional link between the reservoir parameters and its performance. This function is used to explore various properties of the device and to choose the optimal reservoir architecture, thus replacing the tedious and time consuming parameter scannings used so far in the literature.

Neuroendocrine system mainly consists of hypothalamus, anterior pituitary, and target organ. In this paper, a three-state-variable delayed Goodwin model with two Hill functions is considered, where the Hill functions with delays denote the hormonal feedback suppressions from target organ to hypothalamus and to anterior in the reproductive hormonal axis. The existence of Hopf bifurcation shows the circadian rhythms of neuroendocrine system. The direction and stability of Hopf bifurcation are also analyzed using the normal form theory and the center manifold theorem for functional differential equations. Furthermore, based on the sparse identification algorithm, it is verified that the transient time series generated from the delayed Goodwin model cannot be equivalently presented by ordinary differential equations from the viewpoint of data when considering that a library of candidates are at most cubic terms. The reason is because the solution space of delayed differential equations is of infinite dimensions. Finally, we report that reservoir computing can predict the periodic behaviors of the delayed Goodwin model accurately if the size of reservoir and the length of data used for training are large enough. The predicting performances are evaluated by the mean squared errors between the trajectories generated from the numerical simulations and the reservoir computing.

We report the use of a 2-m-long Bismuth Oxide fiber with an ultra-high nonlinearity of ~1100 W(-1)km(-1) in a simple 2R regeneration experiment based on self phase modulation and offset filtering. Numerical simulations and experimental results confirm the suitability of this kind of fiber for 2R regeneration. An improvement in receiver sensitivity of more than 5 dB at 10 Gb/s and 2 dB at 40 Gb/s is achieved.

Reservoir computing is a recently introduced brain-inspired machine learning paradigm capable of excellent performances in the processing of empirical data. We focus in a particular kind of time-delay based reservoir computers that have been physically implemented using optical and electronic systems and have shown unprecedented data processing rates. Reservoir computing is well-known for the ease of the associated training scheme but also for the problematic sensitivity of its performance to architecture parameters. This article addresses the reservoir design problem, which remains the biggest challenge in the applicability of this information processing scheme. More specifically, we use the information available regarding the optimal reservoir working regimes to construct a functional link between the reservoir parameters and its performance. This function is used to explore various properties of the device and to choose the optimal reservoir architecture, thus replacing the tedious and time consuming parameter scannings used so far in the literature.

Amplitude Shift Keying (ASK), Frequency Shift Keying (FSK) and Phase shift Keying (PSK) optical heterodyne detection transmission systems have been developed using 1.3 μm and 1.55 μm wavelength DFB/DBR laser diodes. Receiver sensitivity improvement as much as 7-10 dB over direct detection was realized experimentally at 100-200 Mb/s using 50-105 km single mode fibers. The spectral linewidth characteristics of the laser diodes have been studied. Intermediate frequency beat with a few 10 MHz spectral linewdith has been realized by using DFB-DC-PBH LDs. When external cavity loaded DBR-DC-PBH LDs were used, the beat spectral linewidth has been reduced to less than 1 MHz. The intermediate beat frequency fluctuation was stabilized within a few MHz by controlling the local oscillator laser diode injection current.