Artificial Neural Network-Based Transmission Power Control for Underwater Wireless Optical Communication System

Abstract

Several scientists have proposed using an underwater wireless optical communication system (UWOC) to deliver high-speed data services using the abundant optical spectrum. However, wireless optical signal propagation faces an antagonistic environment when using undersea channels due to various factors like scattering, absorption, turbulence, and optical link misalignment between the transmitter and the receiver. These factors will attenuate the optical signal and lead to degrading system performance. To reduce these factors impact on the communication system's performance, transmitted optical power (OTP) should be increased. Since the UWOC system is battery-powered, increasing OTP will consume more electrical power. Therefore, it is necessary to adjust OTP to a value commensurate with the underwater channel changes. So, an ANN model is proposed in this article for link adaptation, which can adjust the OTP level in tandem with the underwater channel conditions. Data for training, testing, and validation of the proposed system reliability was collected experimentally, and tap water was used as a transmission medium. Evaluation of the proposed model outcome demonstrates that reliable performance is achieved in predicting OTP needed in multiple scenarios. The MSE of the predicted OTP is(9.5×10-3,1.5×10-2, and 1.7×10-2) dBm in the training, testing, and validation stages, respectively. The regression values of the training, testing and validation sets are (0.9997,0.9990, and 0.9996). The results achieved by the proposed model prove it is reliable to be applied in UWOC systems.

Keywords

• uwoc
• ann
• transmitted power control
• neural network
• back-propagation algorithm