Home is Where the (Fast) Internet is: Flat-rate Compatible Incentives for Reducing Peak Load

摘要

Internet traffic to homes is surging, driven by the demand for rich content and the proliferation of home networks. This creates a huge problem for ISPs since residential customers expect the certainty of a fixed bill while ISPs do not want to upgrade backhaul equipment frequently in the absence of extra revenue streams. We consider simple variants on existing flat-rate schemes that will enable homes to self-select a portion of their peak hours traffic and move it to non-peak hours to benefit from offered incentives. We present a well-defined formulation of the problem and characterize its computational complexity. We show that a simple fractional algorithm achieves the optimal traffic reallocation and is realizable with small modifications to existing infrastructure. The fractional model also captures the reality that homes may be willing to move a fraction of their delay-tolerant traffic in response to appropriate incentives. Using trace-driven simulations based on well-accepted utility models and actual backbone traffic from a large ISP, we demonstrate that our incentive scheme can substantially lower peak congestion while still satisfying the increased demand of home networks.