Variance fractal dimension based wireless atm lan traffic estimation for network management

摘要

Wireless ATM LAN provides convenient and high bandwidth services for its users. Transmitting ATM packets over wireless networks presents unique challenges due to the inherent nature of radio communications. In order to prevent bit error and cell loss in this network, it is necessary to analyze and evaluate traffic parameters. Fractal analysis is one of the suitable tools to characterize the wireless ATM traffic and assist in network management. One traffic parameter of interest here, bit errors per packet (bers_per_pk), has been obtained from network simulation and characterized by a variance fractal dimension technique. By analyzing the results, the traffic can be estimated, and the network management can act accordingly, such as by sending handoff or retransmission requests to the transmitter. The variance fractal dimension is based on the properties of fractional Brownian motion (fBm). A time series representing a chaotic or nonchaotic process could be analyzed directly in real time by calculating the spred of the increments in the signal amplitude. Nine fBm fractal waveforms with expected dimension values are generated for the variance fractal dimension data testing. The window size and space size between the windows in the measurement are very important.