Network Actual Traffic Prediction Algorithm Based on alpha-stable Distribution and Wavelet Transformation

In order to improve the prediction precision of wireless sensor network traffic, a new prediction algorithm (State Prediction algorithm based on alpha-stable distribution alpha, SP-alpha) is proposed, combined with alpha-stable distribution and wavelet transformation. The algorithm proposed first defines the characteristics of alpha-stable distribution and then gives the judge basis that obeys alpha-stable distribution. At the same time, it reduces the prediction error of the actual traffic by fusion of the prediction results of alpha-stable distribution with wavelet transformation. Finally, the paper thoroughly researches the key factors impacting on the new algorithm through simulations in OPNET and MATLAB. Compared with the performance of FARIMA model, the simulation results proved that SP-alpha algorithm has better adaptability.