Design of Machine learning based Decoding Algorithms for Codes on Graph

In a communication system, it is more challenging to receive a signal at the receiver than it is to transmit one. The receiver’s task and complexity are larger than the transmitter’s because the received signal must travel over a channel where it will be attenuated and distorted. Communication over unstable noisy channels is made possible by channel coding. Channel encoding is done at the transmitter in the baseband domain and at the receiver, it can be effectively retrieved by using a variety of techniques. This study discusses Belief Propagation (BP) and the Min sum technique for decoding the LDPC encoded codewords using Tanner graphs. We also examine a method to decode the encoded data in a communication system, where the decoding algorithm at the receiver is recast as a machine learning process. So a receiver designed using deep learning techniques can always adapt to the changes in the channel optimization techniques and thus reduce the overall computational complexity.