Integration of time-dependent pricing with transmission rate control for flattening out peak-time demand

This paper studies the impact of transmission rate control and pricing on the congestion management practices on bottleneck links of Internet service provider (ISP). QoS negotiation between ISP and users is conducted by letting users specify their minimum and maximum transmission rate required. Each link tentatively allocates the maximum transmission rates to the traffics that are going to cross it. When the available capacity becomes not enough to satisfy the maximum transmission rate requirements of all on-going users, ISP reduces the allocated transmission rates of connections one by one according to their priorities. When the transmission rate allocated to every user decreases to the minimum transmission rate required, ISP then sends peak-time notification to all on-going users and employs pricing as an incentive mechanism to encourage voluntary disconnections (soft delay). Prices adopted during congestion is completely time-dependent and varies according to the real-time network congestion level. Specifically, the price of each time slot is obtained by solving a maximization problem of network utilization. If the available capacity is still not enough to satisfy the minimum transmission rate requirements of all on-going users, ISP forcibly delays connections one by one also on a priority basis (hard delay). Simulation results show that the proposed approach can significantly shift demand from peak time to off-peak time. When baseline demand is set to be 25% over the capacity of the link (i.e., peak time), 28.2% users voluntarily delay their connections for 0.23 hour under the price incentives, and only 0.84% users are forcibly delayed by ISP.