Charging ATM Internet access, an experiment of usage based ATM charging

The authors investigate user aspects of an ATM charging scheme based on measurements of time and volume. Experiments were made on the charging of an ATM-based Internet access connection of a medium sized company division (/spl sim/250 employees). The object of investigation was to observe whether experimenters equipped with a certain advice of charge interface, and given opportunities to renegotiate choice of tariff during the lifetime of the connection, were able to sensibly adjust to usage incentives of the charging scheme. Three such experiments are reported, each with duration in the range 3-5 days. Experimenter performance was compared to optimal usage strategies by post-analysis of experimental data. A simple heuristic algorithm for automatic adjustment to charging incentives was also investigated. It was found that manually performing tariff renegotiations at the theoretically optimal instants is very difficult. For the tariffs used in the experiments, the ability to hit these instants was however not critical for obtaining quite reasonable charges. Manual operation produced charges 2-13% higher than a minimal charge. The strategies produced by the heuristic algorithm gave charges 0.7-3% over the optimal.