Experimental Investigation of Optoelectronic Receiver With Reservoir Computing in Short Reach Optical Fiber Communications

The cloud edge data center will enable reliable and low latency options for the network, interconnection among these data-centers demand a scalable low-complexity scheme. An intensity-modulated directed detected transmission system is an attractive solution, but chromatic dispersion main limitation higher symbol rate systems. To overcome this challenge, we have proposed experimentally demonstrated receiver with shared-complexity between optical digital domains that enables 80 km reach below KP4 FEC limit 32 GBd on-off keying signal. stage consists of filters slices signal into smaller sub-bands each by photodetector. A feedforward neural network reservoir computing are compared to reconstruct full from mitigate dispersion. Both equalizers shown similar performance advantage requiring fewer inputs easier training process. In work, linear nonlinear activation functions in investigate gain using equalizer. maximum reduced almost half, ?45 km, when linear. also if number used receiver, as demonstrated. case, 2 reduce complexity system, instead total 4, ?55 limit. work provided numerical comparison 4x8 subcarriers system. results 40 increase optoelectronic trade-off should be analyzed different hardware required situation.