Algorithms of fractal time-series analysis of a sensor network

It is experimentally proved that the time series of the data received from sensors of the sensor network possess fractal properties (self-similarity, self-affinity, fractal dimensionality), that gives the chance to predict their dynamics, to reveal the latent correlations, cycles, etc. Algorithms of the fractal analysis, such as DFA method, correlation analysis, Fano’s factor, Hurst's index have shown a high level of statistical correlation of the traffic transiting through a router of a sensor network which accepted data from sensors with different periodicity of inquiry on long time slots. The considered example has shown a high persistent process that, in particular, testifies to the general tendency of magnification of the traffic. The analysis of self-similarity of the traffic of each sensor can be used as a technology for forecasting realization.