Enabling Battlefield Decision-Making in the Tactical Cloud
This research will explore the use of cloud computing to get closer to the reality of the Warfighter having the “right” information, at the “right” time, at the “right” place, and displayed in the “right” format. Using cloud computing in battlefield scenarios can be challenging for several reasons: (1) Applications have different levels of complexity and deadlines in terms of time-to-solution. (2) Environmental constraints may limit the connectivity of computational resources and, thus, the number of accessible cloudlets. (3) Mobile electronic devices are necessarily resource-poor relative to static client and server hardware. Given a battlefield scenario, this research aims to identify the tactical cloudlets that can be formed to execute the specified application within the given time constraints. This may involve movement of computational resources, consideration of the reliability of communication paths, and selection of appropriate computing resources.
To support battlefield computational services, our research is focused on (1) algorithmic support for the specification of tactical cloudlets that can service the computational needs of Warfighters and (2) the development of performance-engineered solutions that have decreased resource demands and, thus, can be migrated to mobile computing platforms. This work will facilitate the development of battlefield cloud services/application solutions and will provide essential information needed to select the computing resources that populate tactical cloudlets.
Main Objectives: This research is focused on computational cloudlets (subsets of a larger cloud) that are comprised of static and mobile resources, which vary w.r.t. execution-time and power/energy performance due to architectural differences, interconnected via MANets, i.e., Mobile Ad-hoc Networks. In this context a cloudlet is formed and reformed in an ad-hoc nature based on the resources reachable via a MANet. Effective use of computational cloudlets to move data between resources and execute jobs, each with a deadline in terms of time-to-solution, involves research in cloud management, power management, and application migration, in particular research in:
1. Discovery of reachable resources and, thus, cloudlets, including resources that will/could be reachable in specific amounts of time, i.e., cloudlets that could be formed when specific mobile resources come into range;
2. Selection of appropriate, available computational resources;
3. Scheduling of computational tasks to resources considering not only computing capabilities and loads, but also reliability in terms of power/energy capacity;
4. Interconnection of resources using MANet or hybrid protocols that take into consideration power/energy consumption and reliability of communication paths; and
5. Techniques that improve job performance in terms of execution-time and power/energy consumption and, thus, facilitate migration of applications to locally-available resources.
Data processing can enable battlefield decision-making and planning. However, due to situational constraints of the Warfighter’s Ecosystem, e.g., terrain, level of threat, value of target, etc., such tasks are often associated with deadlines in terms of time-to-solution. Given the nature of mobile computing platforms, which are necessarily resource poor relative to static client and server hardware, the computational complexity of such a task and its target time-to-solution may be beyond the computing capability and/or power/energy capacity of locally available resources. In these cases, given remote static and mobile resources, the research mentioned above can be used to form tactical cloudlets for this purpose. Given factors, e.g., terrain, that constrain resource connectivity, this may include realizing the movement of resources to enable cloudlet formation.