Optimal Allocation of Pooled Servers for Situation Awareness Applications

U.C. Kozat, M.A. Fecko, and S. Samtani (USA)

Keywords

Server allocation; situation awareness; non-linear opti mization.

Abstract

We defined a new reliability architecture called Dynamic Survivable Resource Pooling (DSRP) that utilizes a virtual backbone for ad hoc network as a dynamic structure for hosting a naming system. This paper formulates the net work model and the problem statement for resource alloca tion for Situation Awareness data applications. The servers (Pool Elements or PEs) represent the data-processing units. Information source points (Pool Users or PUs) are respon sible both for appropriately classifying the priority level of their data and for requesting a server pool with this infor mation. The naming system provides the PUs with a set of PEs with a weight vector that defines a weighted round robin scheduling among the PEs. We formulate the PE se lection for Situation Awareness as an optimal matching of PU requests with PE workloads in terms of rewards and pricing through non-linear optimization.

Important Links:



Go Back