Underwater optical wireless communication system performance improvement using convolutional neural networks

Many applications that could benefit from the underwater optical wireless communication technique face challenges in using this technology due to substantial, varying attenuation affects signal transmission through waterbodies. This research demonstrated convolutional neural networks (CNNs) readily address these problems. A modified CNN model was proposed recover original data of a non-return zero on–off keying modulated transmitted optically tank full Gulf seawater. comparison between and conventional fixed-threshold decoder (FTD) demonstrates excellent performance model, which improved bit error ratio (BER), signal-to-noise (SNR), effective channel length. The BER signals are at powers 24, 26, 27 dBm rate 10 Mbit/s distance 3 m transmitter when FTD is used 7.826 × ?7 , 5.049 ?8 8.38 ?10 respectively. When same powers, 6.23 ?14 1.44 ?16 2.69 ?18 In conclusion, decreased by about seven orders magnitude, length increased four times, SNR 20 dB. simplicity independent prior knowledge conditions. Furthermore, magnificent obtained results make an ideal substitute for ordinary decoders.