Online stable matching as a means of allocating distributed resources
Academic Article
Overview
Research
Identity
Additional Document Info
Other
View All
Overview
abstract
In heterogeneous distributed systems, achieving optimality in both effective use of computational resources (e.g. throughput) and user satisfaction (e.g. response time) is an important unresolved problem. If the users of the system participate dynamically as consumers as well as donors of computational resources, the task of optimizing the exchange of these computational resources leads to a combinatorial problem. As a solution, we propose a novel algorithm, adaptive online stable matching (AOSM). We present experimental data which compare the performance of AOSM with the performance of two alternative algorithms, first-come first-served (FCFS) and fixed-k online.