Search

We propose a novel optimization method that combines two design criteria to reduce the differential modal gain (DMG) in few-mode cladding-pumped erbium-doped fiber amplifiers (FM-EDFAs). In addition to the standard criterion that considers the mode intensity and dopant profile overlap, we introduce a second criterion that ensures that all doped regions have the same saturation behavior. With these two criteria, we define a figure-of-merit (FOM) that allows the design of MM-EDFAs with low DMG without high computational cost. We illustrate this method with the design of six-mode erbium-doped fibers (EDFs) for amplification over the C-Band targeting designs that are compatible with standard fabrication processes. The fibers have either a step-index or a staircase refractive index profile (RIP), with two ring-shaped erbium-doped regions in the core. With a staircase RIP, a fiber length of 29 m and 20 W of pump power injected in the cladding, our best design leads to a minimum gain of 22.6 dB while maintaining a DMGmax under 0.18 dB. We further show that the FOM optimization achieves a robust design with low DMG over a wide range of variations in signal power, pump power and fiber length.

The use of multiple segments on a silicon photonic modulator can increase bandwidth at the cost of greater complexity in the driving signals. We propose and demonstrate a simple driving scheme for use with dual segment modulators that involves very little additional complexity when equal length segments are used. The increased bandwidth with segmentation scales linearly with the number of segments, but the net bit rate does not; net rate depends on many factors. With an IQ modulator with two 2 mm segments, we demonstrate an improvement of 14% in net bit rate as compared to a single 4 mm segment. We examine the trade-offs in moving to three-segment modulation. We explore implementation penalties and use probabilistic shaping and optical pre-emphasis to achieve a net rate of 1.07 Tb/s with dual polarization transmission over 80 km of fiber at 116 Gbaud using 64QAM modulation.

Photonics offers a transformative approach to artificial intelligence (AI) and neuromorphic computing by providing low latency, high bandwidth, and energy-efficient computations. Here, we introduce a photonic tensor core processor enabled by time-multiplexed inputs and charge-integrated outputs. This fully integrated processor, comprising only two thin-film lithium niobate (TFLN) modulators, a III-V laser, and a charge-integration photoreceiver, can implement an entire layer of a neural network. It can execute 65 billion operations per second (GOPS) per neuron, including simultaneous weight updates-a hitherto unachieved speed. Our processor stands out from conventional photonic processors, which have static weights set during training, as it supports fast "hardware-in-the-loop" training, and can dynamically adjust the inputs (fan-in) and outputs (fan-out) within a layer, thereby enhancing its versatility. Our processor can perform large-scale dot-product operations with vector dimensions up to 131,072. Furthermore, it successfully classifies (supervised learning) and clusters (unsupervised learning) 112*112-pixel images after "hardware-in-the-loop" training. To handle "hardware-in-the-loop" training for clustering AI tasks, we provide a solution for multiplications involving two negative numbers based on our processor., Comment: 19 pages, 6 figures

Orbital angular momentum (OAM) modes have low model interactions during fiber propagation at data center distances, and thus are suitable for ultra-high capacity systems at low digital signal processing. Generating OAM modes using free-space setups is useful for proof-of-concept experiments, but is not a scalable solution. We use an optical phased array (OPA) with two-dimensional antennas for on-chip circularly polarized OAM beam generation. Our previous work demonstrated an OAM multiplexer for lower-order modes. In this work, we demonstrate an OAM multiplexer that supports a record of 46 (23 per polarization) simultaneous spatial modes up to OAM order 11. We also improve the crosstalk performance of our multiplexer. We incorporate an intensity tuning capability that substantially improves the OAM quality by enabling a uniform power distribution across the antennas. The worst-case crosstalk for the supported OAM5 to OAM11 are found experimentally to be better than -12 dB, with OAM10 achieving -17.2 dB.

Occupying more channels increases transmission rate, however, crosstalk increases to an unacceptable level long before all 24 channels can be exploited. The crosstalk is not uniform between modes, hence occupying different subsets of channels leads to vastly different achievable transmission capacities. In addition to optimizing occupied subsets of channels, we adopt a bit loading approach. We examine several resource allocation strategies in a coherent detection system, starting with typical OAM mode group granularity (all channels in a group occupied) and ending with single channel granularity. By exhaustive search at mode group granularity, we find a bit load increasing the total capacity by ∼ 15% compared to a minimax solution for mode group allocation. Single channel granularity imposes great computational effort to optimize bit loading. We propose search algorithms that are computationally tractable and improve capacity by ∼ 30% vis-à-vis the minimax solution for mode groups. Finally, we examine the impact of signal-to-noise ratio (SNR) and receiver digital signal processing (DSP) complexity on the overall capacity. We include a discussion of DSP with limited or with no multi-input, multi-output (MIMO) processing.Our algorithms could be applied to any mode multiplexing fiber, as it only relies on knowledge of the crosstalk matrix across modes.

Segmenting a silicon modulator can substantially increase its electro-optic bandwidth without sacrificing modulation efficiency. We demonstrate a segmented silicon IQ modulator and experimentally explore both modulator design and operating point to optimize systems trade-offs in coherent detection. An electro–optic bandwidth of greater than 40 GHz is measured for a 4-mm-long segment, and greater than 60 GHz for a 2-mmlong segment. We evaluate optical transmission experimentally at 120 Gbaud for 16-ary quadrature amplitude modulation (QAM) and 32QAM. The segments are operated in tandem with identical data at each segment. We present an experimental method to align data timing between the segments. Through the optimization of segment biasing and linear compensation, we have achieved a bit error rate (BER) of 16QAM well below the 20% forward error correction (FEC) threshold (2 × 10−2 ). Adding nonlinear pre-compensation allows for 32QAM with a BER below the 24% FEC threshold (4.5 × 10−2 ), enabling a net rate of 483 Gbs per polarization. The modulator can also be operated as an optical digital analogy converter for complex optical signal generation, for which 100 Gbs is achieved for a proof of concept.