Optimal Decisions for Canadian Military Airlift Problem


  1. Boukhtouta, A.
  2. Berger, J.
  3. Powell, W.B.
  4. Bouziane-Ayari, B.
Corporate Authors
Defence R&D Canada - Valcartier, Valcartier QUE (CAN);Princeton Univ, Princeton NJ (US)
A modeling and algorithmic framework for analyzing and simulating the military airlift is presented in this report. Central to the framework is the modeling of the flow of information and decisions. A series of models demonstrating how different levels of information can be represented when making a decision are presented. These information sets are simulated to demonstrate their impact on costs and throughput. A detailed comparison of classical optimization and the current framework is presented. Using historical data for the airlifts conducted by the Canadian Air Force, a series of simulations were conducted to test the effect of uncertainty in customer demands as well as aircraft failures. It is demonstrated in this report that this effect is reduced when adaptive learning is used in the simulation process. There are a number of sources of randomness that arise in military airlift operations: random demands, aircraft failures, weather delays, etc. The cost of uncertainty can be difficult to estimate because it is difficult to give the mathematical expression of this cost. However, it is easy to overestimate this cost if we use simplistic decision rules. The military airlift problem, considered in this report, is modeled as a dynamic program, and solved using approximate dynamic programming. The experiments show that even approximate solutions produce decisions that substantially reduce the effect of uncertainty.

Il y a un résumé en français ici.

airlift problem;approximate dynamic programming;simulation;uncertainties
Report Number
DRDC-VALCARTIER-TR-2010-517 — Technical Report
Date of publication
01 Sep 2010
Number of Pages
Electronic Document(PDF)

Permanent link

Document 1 of 1

Date modified: