Reservoir-Computing Based Equalization With Optical Pre-Processing for Short-Reach Optical Transmission

Ros, Francesco Da; Ranzini, Stenio M.; Bülow, H.; Zibar, Darko

doi:10.1109/jstqe.2020.2975607

Cited by 25 publications

(32 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…While demonstrations of full RNNs for IM/DD system equalization have been reported, 7 RNNs suffer from training challenges due to vanishing gradients which may prevent the use of standard training methods such as back-propagation. Simplified RNN models have therefore been considered, focusing on TD-FNNs, 5,10,11 and RC. 10,11,[14][15][16] Alternative lower-complexity RNNs architectures such as gated recurrent units (GRUs), 12 and long short-term memory (LSTM) 13 have been applied mainly to coherent transmission so far.…”

Section: Neural Networkmentioning

confidence: 99%

“…To decrease the reservoir complexity from the perspective of hardware implementation, only sparse connectivity with an average of 5 interconnections per node is considered in this work. 10 The characteristics of the reservoir, both in terms of number of nodes and of the spectral radius ρ of the reservoir matrix W res , contribute to defining its memory and thus its ability to equalize ISI.…”

Section: Reservoir Computingmentioning

confidence: 99%

“…All results are averaged over 100 reservoir realization and follow the reservoir properties of. 10 For the analysis, we consider the average performance of 1000 randomly initialized reservoirs with an average of 5 interconnections per node, and a single input (plus bias) and output node. The input weights are drawn from a uniform distribution between 0 and 1, and the reservoir weights are drawn from a normal distribution with zero mean and variance σ 2 res = 1.…”

Section: Reservoir Memorymentioning

confidence: 99%

“…The spectral radius is then varied by re-scaling the reservoir matrix, see. 10 For fair comparison between the two definitions, no directed input-to-output interconnections are considered for the 95%-memory capacity, as is the case for the settling time.…”

Section: Reservoir Memorymentioning

confidence: 99%

“…In this work, we review our systematic comparison between linear and nonlinear equalizers focusing on a simple OOK transmission system. [9][10][11] In particular, we analyze the performance improvement provided by two recent machine-learning tools which still allow for relatively limited training complexity, i.e. time-delay feed-forward NN (TD-FNN) and reservoir computing (RC).…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Machine-learning-based equalization for short-reach transmission: neural networks and reservoir computing

Ros¹,

Ranzini

Dischler

et al. 2021

Metro and Data Center Optical Networks and Short-Reach Links IV

Self Cite

View full text Add to dashboard Cite

The substantial increase in communication throughput driven by the ever-growing machine-to-machine communication within a data center and between data centers is straining the short-reach communication links. To satisfy such demand-while still complying with the strict requirements in terms of energy consumption and latency-several directions are being investigated with a strong focus on equalization techniques for intensitymodulation/direct-detection (IM/DD) transmission. In particular, the key challenge equalizers need to address is the inter-symbol interference introduced by the fiber dispersion when making use of the low-loss transmission window at 1550 nm. Standard digital equalizers such as feed-forward equalizers (FFEs) and decision-feedback equalizers (DFEs) can provide only limited compensation. Therefore more complex approaches either relying on maximum likelihood sequence estimation (MLSE) or using machine-learning tools, such as neural network (NN) based equalizers, are being investigated. Among the different NN architectures, the most promising approaches are based on NNs with memory such as time-delay feedforward NN (TD-FNN), recurrent NN (RNN), and reservoir computing (RC). In this work, we review our recent numerical results on comparing TD-FNN and RC equalizers, and benchmark their performance for 32-GBd on-off keying (OOK) transmission. A special focus will be dedicated to analyzing the memory properties of the reservoir and its impact on the full system performance. Experimental validation of the numerical findings is also provided together with reviewing our recent proposal for a new receiver architecture relying on hybrid optoelectronic processing. By spectrally slicing the received signal, independently detecting the slices and jointly processing them with an NN-based equalizer (wither TD-FNN or RC), significant extension reach is shown both numerically and experimentally.

show abstract