Artificial Intelligence (AI) based Forward Error Correcting (FEC) codes for Next Generation of Broadband Networks

Project: HDR ProjectPhD

Project Details


The rapid evolution of coding technology is driven by the growing demand for reliable communication in real-time and non-real-time systems over AWGN channels with varying noise levels. Advanced Forward Error Correction (FEC) coding methods have raised expectations for desired Bit Error Rate (BER) performance, which includes the use of two or more codes concatenated in serial, parallel, or hybrid forms with an iterative decoding methodology. With the trend towards limited bandwidth capacity in next-generation broadband networks, the need for intelligent FEC coding models to handle unpredictable channel conditions while achieving high information gain and efficient data transfer rates has become essential. The LDPC code is a widely recognized and popular for its low-complex decoding structure. Furthermore, the cyclic-LDPC code offers higher convergence properties in the waterfall region with a larger minimum distance and higher achievable data rate, which makes it more suitable for meeting the requirements of next-generation broadband networks. The integration of Artificial Intelligence (AI) in deep machine learning techniques to produce strong codes has become increasingly popular, presenting an opportunity to improve the efficiency of the system models. Deep Learning Neural Network (DNN) models have been identified as a significant contributor to performance improvement in various applications. In this study, we seek to enhance the performance of the Parallel Concatenated Block (PCB) encoder structure using AI to optimize its performance. Our objective is to create a cost-effective AI model that integrates DNN with the PCB encoder code structure, leading to stronger and more robust codewords. The effectiveness of our model will be evaluated through a comparative analysis with existing methods for performance measurements.
StatusNot started


Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.