Abstract: In order to mitigate the interference problem by noise in wireless sensor network,a novel traffic prediction algorithm CAWP (Cellular Annealing-based Wavelet Prediction) is proposed. In this algorithm, traffic is decomposed by wavelet transform,which wavelet coefficients are seen as antibody group.Then,the crossover and mutation operation are replaced with cellular evolution rules,and the optimal wavelet transform in the next moment is got.Finally,a simulation was conducted to study the performance in those algorithms.The results show that CAWP has good suitability.