Compressed sensing techniques deal with unique sparse solutions to an under-determined system of linear equations. In this presentation, we explore the applications of compressed sensing techniques for designing and decoding error control codes of very small to small lengths. Encoding is done by sparse linear combinations of columns of a dictionary matrix with low mutual coherence. We introduce structured sparsity (subblock sparse coding) to remove lookup tables at the transmitter and the receiver. We use an iterative decoding algorithm called MAD decoder based on Matching Pursuit for decoding. We further improve the MAD decoder by introducing a parallel MAD decoder, where the The MAD algorithm runs over multiple parallel paths. With the help of mutually unbiased bases (MUB) and gold codes, we show that our scheme performs similar to the existing error control codes. We also show that our scheme generalizes to multi-user codes in Multiple Access channels, Broadcast Channels, and Interference Channels under certain conditions.