Symbol error rate minimization using deep learning approaches for short reach optical communication networks

Abstract

The rapid expansion and development of communication networks and systems in the optical domain causes conventionally used methods in the transmission of long haul networks, to enter the systems that cover shorter transmission distances underneath 100 km. They can be used in various applications where high-speed connectivity is needed like the inter and intra data center interliks, optical access networks, indoor and in-building communication. There are many approaches which can be used for short-reach communication, but they are subjected to the complexity and high margins. Machine learning (ML) approaches provide key solutions for numerous challenging situations due to their robust decision making, problem solving, and pattern recognition abilities. In the domain of signal processing, ML techniques have been studied extensively in short-reach optical communications and networks. We will apply deep learning approaches to randomly generated dataset. A deep neural network (DNN), which includes the transmitter, channel, and receiver for an optical fiber communication network is to be modeled and implemented. Transceiver optimization can be implemented in a single end-to-end operation by enabling this approach. The channel model consists of different impairments like chromatic dispersion (CD), nonlinearity, and attenuation. We expect to predict the symbol error rate (SER) of received output in the existence of numerous channel impairments with minimum error in short-reach optical fiber communication networks.