Compressed sensing is recently developed theory which focuses on reconstructing a signal by finding solutions to set of linear equations which is an under determined system. Applications which involve huge data to be transmitted over wireless radio is often limited by the power consumed by radio. It is highly recommended to send lesser number of samples and recover back the original signal with minimal error. Compressed sensing (CS) is one such theory which recovers back the original signal with fewer non adaptive extracted measurements. CS provides a novel approach to sample signals below Nyquist rate with limited information loss. The main focus of this thesis is to develop an architecture which compresses the data at encoder using Discrete Cosine Transform (DCT), sample the data using +-1 random Bernoulli matrix and recover back the original signal at the decoder using a greedy algorithm called Orthogonal Matching Pursuit (OMP). We modeled our architectures both on Floating and Fixed Point representations using custom developed library of numerical operations in Matlab. The entire chain of CS architecture is tested using the neural data and several graphs are plotted and analysed. Finally The CS Encoder and Decoder hardware structures are implemented on a field-programmable gate array (FPGA) and performance of the same is evaluated in terms of mean squared error (MSE) and signal to noise distortion ratio (SNDR) metrics.