Publication

Abstract : Nowadays, vast amounts of data transmission and data retrieval are crucial and are done using many ways. Orthogonal Frequency-Division Multiplexing (OFDM) is one of the efficient ways to transmit data with the help of orthogonal sub-carriers, which is used in applications such as WiFi, WiMax, and cellular communication. In this paper, instead of conventional detection techniques, machine learning (ML)-based methods are adopted to detect the symbols after data is being received through the Additive White Gaussian channel (AWGN). Detection is one of the areas in which the bit error rate (BER) performance of the OFDM system can be improved. Machine learning algorithms only depend on the training data to predict the outputs; hence, we can detect the symbol even without the use of cyclic prefix or channel estimation which can save a lot of time and data if the input data is large. A comparative study on the performance of receivers based on K-means clustering, k-nearest neighbors classifier, support vector machine, linear discriminant analysis, and quadratic discriminant analysis is done. The modulation techniques such as BPSK and QAM with various modulation orders ranging from 4 to 64 are used in this analysis. Performance comparison of aforementioned detection techniques using employing machine learning is done using BER vs signal-to-noise ratio (SNR) in the range of 0–30 dB.