**5. Conclusions and future work**

In this work, we use a tree structure in which the edges have weight. The advantage of our method is that any node in our weighted tree has a unique path, so the user's query against all of previous methods is not sent to the extra and unnecessary paths. We can directly reach the target node using a resource footprint which is stored in nodes.

Furthermore, for resource discovery we only use one bitmap in every node which is for the storing of information about its local resources and the resources of its children and descendant. Also it preserves a footprint of resources and if we need a resource which is available in its children or descendant, we can directly and without any referring to unnecessary and extra nodes, reach the target node. This method significantly reduces the system traffic and increases the performance of system.

We compare our algorithm with previous algorithms using simulations and results and show that the number of nodes visited in our resource discovery algorithm is less than that for other algorithms, and the difference would be significant with an increase in the number of nodes. Also the cost of update in our proposed algorithm is low.

In future, if we could a present a technique that could locate several heterogeneous resources (with different attributes) of a grid environment in smaller forms with a lower volume, or placed in one bitmap involving some factors, for example allocated costs for resources etc., we could improve the algorithm.
