Orthogonal Frequency Division Multiplexing (OFDM) has emerged as a dominant modulation technique in modern wireless communication systems due to its high spectral efficiency and resilience to multipath fading. This project proposes an enhanced modulation classification framework utilizing Radial Basis Function Neural Networks (RBFNN) for robust channel estimation and symbol detection in OFDM systems. The proposed approach integrates data-driven learning through RBFNN with traditional channel estimation techniques to improve symbol detection accuracy and mitigate the effects of noise and channel distortions. The system architecture incorporates Quadrature Amplitude Modulation (QAM) for data transmission, followed by encoding and symbol mapping, while the receiver leverages RBFNN-assisted channel estimation to optimize symbol DE mapping and decoding. Simulation results demonstrate that the RBFNN-based classification method significantly enhances classification accuracy, reduces bit error rates (BER), and outperforms conventional estimation techniques in various signal-to-noise ratio (SNR) conditions. This project paving the way for improved wireless communication reliability and efficiency.