A Novel Node Selection Method for Real-Time Collaborative Computation in Cloud

摘要

Most personal computers are underutilized in most time. Collaborative cloud computing (CCC) is an emerging paradigm, which can organize and integrate these idle computing or storage resources scattered around the world. In this paradigm, due to individual capacity limitation, multiple participants (nodes) need to cooperate with each other to complete complicated tasks. Considering the heterogeneities of different nodes, node selection problem becomes more pivotal and non-trivial. Actually, different nodes are usually geo-distributed and exhibit different service performance. In this paper, we propose a novel node selection method to determine the most appropriate resource providers who can cooperate to support real-time computing in CCC. In detail, our method consists of the following steps. Firstly, all participatory nodes are clustered based on the preferences about location information and credit scores. Secondly, a stable marriage model is applied to fulfill the two-sided matching problem efficiently. Finally, a certain number of reserving nodes are maintained to guarantee the service quality with charging no extra fee to the requester. Experimental results show that our method can achieve a good performance and decrease the matching time.