Aiming at energy saving problem, simplified prediction coding algorithm (SPCA) is introduced to compress the source data. As time dependence exists among the source data, redundant information can be deducted if the source data is effectively estimated. Prediction orders and the bit counts of the coding can be set freely to meet the engineering requirement, and linear quantization is introduced to simplify the realization of the algorithm. Two sensor nodes are selected and download original algorithm program and SPCA program separately. Experimental results show not only the precision of source data is assured, but also the energy is saved by SPCA, although data-preprocessing and SPCA occupy a little more storage and processing time.