Parallel optimization of frequency density distribution based on non-parametric method

Hopping-frequency forecasting is one of the key problems in information countermeasure. Based on pseudo-randomness of hopping-frequency and similarity of continuous distribution between these frequency data blocks, a novel method is presented. It applies non-parametric density estimation to predict those sections of frequency with bigger probabilities, which can guide smart comb-like jam of communication. To improve the performance of the method, a scalable computation based on message passing is implemented on multi-node computer system and is verified by the test. Key words: complex electromagnetic environment; hopping-frequency forecasting; on-parametric density estimation; R; message passing; parallel optimization