The algorithms presented in the Doctoral Thesis have been designed for microprocessor/microcontroller systems, which can perform discrete wavelet transform in real-time with filter banks and estimate from transform coefficients the magnitude of process self-similarity - the Hurst parameter. Such algorithms can be implemented in routers in order to evaluate the parameters of incoming network traffic and classify different types of traffic with further separate QoS service of these traffic types. Discrete wavelet transform algorithm can be applied in other tasks as well, which are not related to traffic analysis and require evenly time-distributed processing time of minimal value.