Abstract. Hypercube network is one of the most important multiprocessing mechanisms in parallel computing. At present, many network topologies are proposed by using hypercube characteristics. Such as Peer-to-Peer networks. The MPPR algorithm proposes in this paper is an improvement for the routing algorithm of HyperD (a labeling schema Peer-to-Peer networks), and the god is to find a better path in the effective period of time. The algorithm consists of two steps in which querying at all paths that satisfy the conditional probability ratio and then select the optimal path. The algorithm can find better paths in small time fluctuation range, thus it saves overhead of the network.
IntroductionThe hypercube is highly parallel, and the fusion is very strong. At the same time , it can combine with the topological structure of many other types of network, such as rings, linear vector and grid. These can be embedded in a hypercube in most applications, so the research on the hypercube structure and routing algorithm has never stopped [6]. The query routing algorithm can be divided into three categories: (1) the routing path from the source node (u) to the destination node (v) ; (2) node u that broadcasting to all other nodes of the path; (3) paths between any two points in a hypercube. In the conventional case, these three problems are satisfactorily solved.The HyperD concept was first presented at [2,5].HyperD was originally designed as a topological model for the dynamic distributed federated database (DDFD) [3]. The HyperD protocol assigns the label settings to each participating node [7,2]. Then these labels are used to determine which nodes will form virtual connections. Each label represents a position in a virtual hypercube, so the nodes in the virtual network connect to form a topology which resembles a hypercube. Finally, labels are used to facilitate communication between nodes using known optimal hypercube communication methods [7]. This paper mainly studies the routing algorithm which is the route communication using HyperD model, and the model was proposed by Andi Toce .etc [5].Improving the HyperDRouting routing algorithm between two points given [7].Contribution of this paper:1) The advantages and disadvantages of existing HyperDRouting algorithms are analyzed. 2) Adopting the idea of compromise, putting forward the improved routing algorithm. 3) Comparing the efficiency and effectiveness of the algorithm through a large number of experiments.