Online Hedge Reservation for Diverse Plans and Competitive Analysis

Authors

  • Binghan Wu University of Sydney
  • Wei Bao University of Sydney
  • Dong Yuan University of Sydney
  • Bing Zhou University of Sydney

Keywords:

Continuous Planning, On-line And Real-time Domains

Abstract

In this paper, we investigate the plan reservation problem with diverse plans in mobile networks. The pricing scheme includes: 1) Pay-as-you-go (PAYG) payment; 2) All-in-one plan: an upfront fee is charged to cover data volume of a period of time; and 3) Directional plan: an upfront fee is charged to cover data volume of a specific app for a period of time. We investigate online plan reservation with competitive analysis, as the data volume is not known until an app is used. The problem is challenging as there are multiple directional plans and one all-in-one plan, creating a large decision space and complicated correlations among the decisions. We propose the Online Hedge Reservation (OHR) Algorithm to solve the problem and prove that it achieves e^beta/(e^beta-1) competitive ratio when each plan is valid till the end of each calendar month and 2e^beta/(e^beta-1) competitive ratio when each plan is valid for a full month, where beta is the ratio of prices of the directional plans and the all-in-one plan. This is an exciting neat extension of the competitive ratio e/(e-1) of the classic ski-rental problem. Finally, trace-driven simulation is conducted to further verify the advantages of the OHR Algorithm.

Downloads

Published

2021-05-17

How to Cite

Wu, B., Bao, W., Yuan, D., & Zhou, B. (2021). Online Hedge Reservation for Diverse Plans and Competitive Analysis. Proceedings of the International Conference on Automated Planning and Scheduling, 31(1), 396-404. Retrieved from https://ojs.aaai.org/index.php/ICAPS/article/view/15985