Using Matrix indexes for Resource Discovery in Grid Environment Leyli Mohammad khanli Assistance Professor CS Department University of Tabriz Tabriz, Iran Saeed Kargar M.S. student Tabriz Branch Islamic Azad University Tabriz, Iran Ali Kazemi Niari M.S. student Tabriz Branch Islamic Azad University Tabriz, Iran E-mail address: l-khanli@tabrizu.ac.ir E-mail address: saeed.kargar@gmail.com E-mail address: a.kazemi.n@gmail.com Abstract There are many resources and services in grid environment which may be distributed geographically. So, resource discovery methods with lower update and discovery costs are of vital significance. The centralized methods have the advantage of improved management of nodes, but in a grid with large number of nodes and resources, a bottleneck is caused in the server, which reduces the system efficacy significantly. Decentralized methods proposed till now are not able to discover the required resources of the users and update their environment without any additional traffic and referral to unnecessary nodes. In this paper, we use a matrix in each node to reserve the local resources and the information of children to access the requested resources of the user. In our algorithm, each node uses indexes of resources in its matrix to present available resources to the parent node. The user delivers the indexes of the resource, which contains the least bits, to one of the tree nodes. The main advantage of our method is that the unnecessary and extra nodes are not visited in both discovery and update phase; i.e. not only reduces the discovery cost but also lowers the update costs. Our proposed approach is an efficient and easy method. The simulation results indicate that our method is the most optimal one compared to other methods considering the visited nodes in the discovery and updating as well as the lower traffic. 1. Introduction Today, a huge number of heterogeneous computational resources which are geographically distributed can be connected using grid. In the grid environment, the numbers of users and requested resources, variety of resources, number of available nodes etc, are large, so the centralized methods [1-3] are not suitable for such environments [6]. The other methods are distributed approaches. The algorithms which use such methods face difficulties in the grid with wide environments. Flooding-based and random-based are amongst these methods. Such methods are used in most resource discovery algorithms but with large number of resources and types in the grid environment, problems increase. Some recently distributed methods use tree structures in their algorithms. They are able relieve most disadvantages available in the previous methods. "A resource discovery tree using bitmap for grids" [4] and "FRDT" [5] are amongst these methods. In the current methods, some unnecessary visited nodes cause extra traffic and reduce system efficacy in discovery or updating phase. In this paper, like [5], we use a weighted tree for resource discovery. We use these weights as a unique path for resource discovery. Furthermore, we use a matrix in each node to reserve the local resources and the weight of children to access the requested resources of the user. Below are presented the works done in the field of resource discovery in grid in section 2. Section 3 discusses proposed method in detail. Section 4 includes simulation results, comparing our method with previous approaches and their results. Finally, section 5 presents conclusion and further works to improve the current method. 2. Related work In recent years, increasing computation needs have made researchers create a huge processing environment by connecting distributed resources, i.e., the current link and its infrastructure appear as a single powerful processing system with high storage capacity. It is likely that large number of heterogeneous computational resources connect to each other through personal computers to powerful clusters [7]. So, resource discovery methods are of vital importance. Several