The Signal Detection In The Background Of The Bimodal Noise Based On Wavelet Packet Transform

On the basis of time-frequency analysis capabilities of multi-scale wavelet packet transform, this paper studys statistics of one special of non-Gaussian noise -- Bimodal Noise. Through the actual calculation and simulation, with the increase of its length of the sequence and decomposition level,its subspace wavelet packet transform coefficients are all similar to obey Gaussian distribution for the three Bimodal Noise model. Thus, this paper studies wavelet packet detection system mainly.Shortly afterward, the simulation confirms that its detection performance has nothing to do with the Bimodal Noise model.Taking into account the practical application of the feasibility, this paper compare the wavelet packet transform detection performance with the traditional detection performance in the context of the Bimodal Noise. The comparison indicates that wavelet packet transform detection system not only has signal detection in a relatively small signal to noise ratio, but also its detection performance is better than the traditional Higher-order statistics detection system and the classical detection system, and also to avoid the the shortcomings of Higher-order statistics methods such as complex calculation of variance, complex algorithms and poor real-time. Clearly, wavelet packet transform detection system is feasible to detect signal in the context of the Bimodal Noise.