A model-based study of the impact of managed services and the spawning of applications in broadband networks

  • Authors:
  • Debasis Mitra;Qiong Wang

  • Affiliations:
  • Bell Labs, Alcatel-Lucent, Murray Hill, NJ;Bell Labs, Alcatel-Lucent, Murray Hill, NJ

  • Venue:
  • Proceedings of the 24th International Teletraffic Congress
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Whether to allow a broadband service provider to supplement basic best-effort service by high quality managed service at a premium price has been hotly debated, especially in the context of network neutrality. Offering managed service provides delay-sensitive customers with a better option of guaranteed quality of service, but may allow the service provider to starve the best-effort service of bandwidth so as to force best-effort users to subscribe to more expensive managed service. We study this issue through models in which the library of applications available to all broadband users is key to usage and decisions made by users and service provider, and, importantly, for investigating how the innovations process leading to new network applications, affects, and is affected by, the service provider's bandwidth allocation. First, we hold the number of applications fixed and potential users weigh utility and cost to decide whether to subscribe to the broadband network, and if so, whether to use best-effort service applications that are free but associated with congestion-dependent delay or to pay a per-use fee to use managed service. Likewise, the service provider provisions bandwidth for the two services and sets the per-use fee for managed service to maximize its profit. Next, we model innovations that spawn new applications to originate from use of best-effort service. Cognizant of the separation of time constants in application generation and bandwidth provisioning, we assume that the service provider undertakes a sequence of myopic optimizations, in each of which it takes the number of applications as fixed by the preceding decision. We find that the innovations process serves as a stabilizer that keeps bandwidth allocation in balance: with fewer applications, the provider will sponsor more best-effort service and more usage leads to more new applications. When the number of applications becomes excessively large, the provider is induced to cut back on bandwidth for best-effort service to provide more managed service.