System Concept and Simulation Model for a ProposedClass of Delay-Tolerant Variably-Priced NetworkServices

Abstract

We describe a system concept for the revenue-producing disposition of surplus capacity at off-peak times in real trunking networks. The idea is to approximate a competitive market for distribution of the networks' time-varying excess capacity with a pricing strategy controlled by the network. The scheme is intended to allow network operators to stimulate background traffic loads to gain new revenues from otherwise idle time on existing installed resources. The concept is suitable for low priority delay-tolerant or opportunistic applications such as remote backups, software distribution, dispatching batched faxes, disseminating newsgroup updates, updating web page caches or routing tables. Background service subscribers are notified of price reductions at off-peak times to elicit additional traffic for the network. Traffic aggregators act on behalf of subscribing organizations or groups of users. The background service is completely subordinate to the conventional tariff-priced on-demand calling services and the variable background pricing merges with the foreground under suitable total load. This paper focuses on the network problem of price setting to continually maximize the price-volume product in a time-varying price-sensitive traffic environment such as this concept implies. A price-stimulated offered traffic environment is simulated in which time of day, price, and hidden demand latency and demand curve characteristics all affect the offered traffic. An analytically optimum strategy is available for the particular traffic model used and the performance of a fuzzy logic price controller is tested against the revenue-optimal strategy. Depending on econometric assumptions for latent traffic demand and price-volume curves, increases in revenue from 4%-20% are obtained in simulation of a 30-trunk group having a typical daily load pattern.