With the dynamic and mobile nature of ad hoc networks, links may fail due to topology changes. So, a major challenge in ad hoc networks is to search path from a source to a destination with an efficient routing method, which is also an important issue for delay-sensitive real-time application. A topology of the nodes interconnections in an ad hoc network can be modeled as an adjacency matrix. In this paper, we present a knowledge discovery agent to collect topology information and derive the adjacency matrix. Based on a sequence matrix calculation, we propose an effective routing algorithm Matrix- Based Backward path Search (MBBS) for route discovery from a destination to a source. The performance of MBBS and that of Breadth First Search (BFS) algorithm are compared. The research result proves that MBBS is more efficient in recognizing the route.