Aiming at the redundancy problem in wireless sensor network, a step-by-step compression-aware codec algorithm decomposes the original signal according to the sparsity of the same or different dictionaries, and compresses and encodes the decomposed signal by using the Bernoulli observation matrix to generate a dictionary mask during the compression process. The terminal restores the data step by step according to the coding information, the sparse dictionary and the dictionary mask. This method is more robust than traditional coding compression technology and is not sensitive to packet loss. Compared with the original compressed sensing algorithm, it saves the transmission bandwidth and improves the real-time performance of data information acquisition.