Position paper at the 9th International Conference on Information Fusion (July, 2006)
The military resource management problem involves timely distribution and placement of materiel, personnel, and sensor assets to accommodate mission requirements throughout the world. Moreover, the problem is highly dynamic in nature in the sense that requirements are changing almost at every moment due to changing situations in the underlying missions, especially in the current age of asymmetric warfare. The resource management problem of today’s US military is perhaps the largest and most complex optimization problem in terms of the number of variables representing various resource types and constraints relating the variables. The traditional solution using mass-based hierarchical distribution of resources, applicable to force-on-force conventional warfare, is fast becoming a concept of the past. This modern resource management problem needs to be considered within a network centric warfare (NCW) environment, and hence the term “sense and respond logistics” (S&RL). Given this paradigm, the issues of effective communication and coordination among various nodes within a netcentric environment replace the complexity of a centralized and monolithic search problem for traditional resource management.
In this paper, I view the resource management issue as intimately related to the problem of planning and scheduling of tasks, that is, one cannot effectively reason with resources in isolation. In the data fusion domain, this view translates to consideration of the collection management process during situation/threat assessment and courses-of-action generation processes. Within a NCW environment a node must proactively determine mission requirements based on the current situation and threat, and then coordinate with other nodes to meet those requirements via some communication mechanism (e.g. publish and subscribe) on the underlying infrastructure. The specific algorithmic approach I advocate for dynamically managing resources is essentially based on constraint programming (CP) techniques, where constraints are declaratively stated and placed on the types and quantities of resources at hand. The constraints can also be dynamically added or retracted from the system, thus are suitable to meet the demands of a dynamic resource management environment. The approach distinguishes between consumable and non-consumable resources, and in some situations views time as a special kind of resource. Here I illustrate the CP approach to resource management in NCW environments in terms of two examples. The first example is related to surveillance asset management and the second is related to logistics.
For More Information
(Please include your name, address, organization, and the paper reference. Requests without this information will not be honored.)