Abstract: With the increasing of nodes in the network, the flooding based routing strategy on JXTA will yield a lot of redundant data packets in the network, which may result in network congestion, low routing efficiency, and more difficult in fitting the complex network environment. By combining ant colony algorithm and random walk algorithm, this paper proposes a dynamic routing algorithm, which changes the forwarding rules of routing resolving request information so as to reduce the forwarding number. Finally, the network simulation experiments are made via the Peersim network simulator, which show that the improved routing algorithm can significantly reduce the number of redundant packets and effectively enhance the routing efficiency of the JXTA network.