SCIENCE CHINA Information Sciences, Volume 63 , Issue 6 : 160402(2020) https://doi.org/10.1007/s11432-019-2837-0

A brief review of integrated and passive photonic reservoir computing systems and an approach for achieving extra non-linearity in passive devices

More info
  • ReceivedDec 3, 2019
  • AcceptedMar 12, 2020
  • PublishedMay 13, 2020



This work was supported by National Science Foundation (Grant No. NSF-1710885).


[1] Jaeger H. Harnessing Nonlinearity: Predicting Chaotic Systems and Saving Energy in Wireless Communication. Science, 2004, 304: 78-80 CrossRef PubMed ADS Google Scholar

[2] Maass W, Natschl?ger T, Markram H. Real-time computing without stable states: a new framework for neural computation based on perturbations.. Neural Computation, 2002, 14: 2531-2560 CrossRef PubMed Google Scholar

[3] Tanaka G, Yamane T, Héroux J B. Recent advances in physical reservoir computing: A review.. Neural Networks, 2019, 115: 100-123 CrossRef PubMed Google Scholar

[4] Vandoorne K, Mechet P, Van Vaerenbergh T. Experimental demonstration of reservoir computing on a silicon photonics chip. Nat Commun, 2014, 5: 3541 CrossRef PubMed ADS Google Scholar

[5] Katumba A, Freiberger M, Bienstman P. A Multiple-Input Strategy to Efficient Integrated Photonic Reservoir Computing. Cogn Comput, 2017, 9: 307-314 CrossRef Google Scholar

[6] Freiberger M, Katumba A, Bienstman P, et al. On-chip passive photonic reservoir computing with integrated optical readout. In: Proceedings of 2017 IEEE International Conference on Rebooting Computing (ICRC), 2017. 1--4. Google Scholar

[7] Katumba A, Heyvaert J, Schneider B. Low-Loss Photonic Reservoir Computing with Multimode Photonic Integrated Circuits. Sci Rep, 2018, 8: 2653 CrossRef PubMed ADS Google Scholar

[8] Fiers M A A, Van Vaerenbergh T, Wyffels F. Nanophotonic reservoir computing with photonic crystal cavities to generate periodic patterns.. IEEE Trans Neural Netw Learning Syst, 2014, 25: 344-355 CrossRef PubMed Google Scholar

[9] Laporte F, Katumba A, Dambre J. Numerical demonstration of neuromorphic computing with photonic crystal cavities. Opt Express, 2018, 26: 7955 CrossRef PubMed ADS Google Scholar

[10] Van Vaerenbergh T, Fiers M, Dambre J. Efficient simulation of optical nonlinear cavity circuits. Opt Quant Electron, 2015, 47: 1471-1476 CrossRef Google Scholar

[11] Schneider B, Dambre J, Bienstman P. Using Digital Masks to Enhance the Bandwidth Tolerance and Improve the Performance of On-Chip Reservoir Computing Systems.. IEEE Trans Neural Netw Learning Syst, 2016, 27: 2748-2753 CrossRef PubMed Google Scholar

[12] Katumba A, Freiberger M, Laporte F. Neuromorphic Computing Based on Silicon Photonics and Reservoir Computing. IEEE J Sel Top Quantum Electron, 2018, 24: 1-10 CrossRef ADS Google Scholar

[13] Ma C, Sackesyn S, Dambre J, et al. All-optical readout for integrated photonic reservoir computing. In: Proceedings of 2019 21st International Conference on Transparent Optical Networks (ICTON), 2019. 1--4. Google Scholar

[14] Zhang H, Feng X, Li B. Integrated photonic reservoir computing based on hierarchical time-multiplexing structure. Opt Express, 2014, 22: 31356 CrossRef PubMed ADS Google Scholar

  • Figure 1

    (Color online) Illustration showing different structure types of integrated and passive RC systems. (a) 16 nodes, spatial node system (from [4]); (b) photonic crystal cavity with seven defects, spatial node system (from [9]); (c) time-delay effect induced by cascaded ring resonators, temporal node system (from [14]).

  • Figure 2

    (Color online) Structural configuration of the device.

  • Figure 3

    (Color online) Output spectrum of one ring resonator: red curve is the TM 0th mode, and blue is the TM 1st mode.

  • Figure 4

    (Color online) The mode combiner. (a) Structure configuration, both the red and blue waveguides are Si. protectłinebreak (b) Field distribution of the entire structure. (c) Field distribution of the red arm: TM 0th mode converted to TM 1st mode. (d) Field distribution of the blue arm: TM 0th mode is maintained.

  • Figure 5

    (Color online) Time-delay function of a single sinusoidal pulse. Delay function for the (a) TM 0th mode only, (b) TM 1st mode only, and (c) whole device with the mode combiner. Black part in each figure represents the input pulse.

  • Figure 6

    (Color online) Time-delay function for the 100 digit coded input. (a) Input intensity, the code is randomly selected; (b) the output response of the input.

  • Figure 7

    (Color online) Training and test results. (a) Trained data is shown by the red curve, and the blue curve shows the training target, MSE = 0.0080. (b) Test data is shown by the pink curve, and the light blue curve shows the test target, MSE = 0.1080.