With the increase use of portable devices such as Personal Digital Assistants (PDA), laptops, voice and data integrated cell phones and many more, there is a need of wireless communication method using air as the medium to transmit and receive information between terminals. Radio waves propagate from transmitting antenna and travel through free space undergoing reflections, diffractions and scattering. They are greatly affected by ground terrain, the atmosphere and the objects in their path like buildings, bridges, hills etc. Nowadays, the existence of a direct line of sight path between the transmitter and the receiver is unlikely. These multiple phenomena are responsible for most of the characteristic features like the quality of the received signal. In the above case propagation is mainly due to reflection and cattering from the buildings and by diffraction. So, in practice the transmitted signal arrives at the receiver via several paths with different time delays creating a multi path situation at the receiver, these multipath waves with randomly distributed amplitudes and phases combine to give a resultant signal that fluctuates in time and space. This phenomenon of random fluctuations in received signal level is termed as fading. The existing demodulation techniques like FM, AM will determine the signal from the received signal based on the mean distance method, which cannot provide the desired level of BER, which fails in proper estimation under high fading and high Doppler-Shift effect. SOLUTION: This project provides the implementation of an enhancement to the demodulation technique using Bayesian approach for the physical layer simulation of a General Packet Radio System (GPRS) considering variable Rician fading and variable Doppler-Shift effect for an AWGN channel. The system performance is evaluated based on Bit Error Rate (BER) and Signal to Noise Ratio (SNR) for the realized GPRS system. Matlab platform is used for the implementation, analysis of the proposed system with for functional verification in terms of BER and SNR. We have showed the comparative difference between the theoretical calculation of QPSK signal and to the values obtained by our program. The values show difference up to 0.4 db for a 1000 bit random vector. Moreover, we also compared with QAM demodulation technique in MATLAB code to show difference up to 1.4 db for a 1000 bit vector. These results signify better performance of the system as it has saved bandwidth.