In this paper, we propose a concept of subnet of maximum controllability based on the model of network controllability, and set up the navigation search model based on the subnet of maximum controllability, called NMSMC. The strategy of adding links that is based on the subnet of maximum controllability is to solve, with the minimum cost, the terminating search, the problem that arises from no way for particles to search in the directed network. Based on the subnet of maximum controllability to deploy navigation nodes,the search time of the whole network can be made close to the average shortest path of the navigation network,which the number of navigation nodes is only 2% of the total nodes. The experimental results of the ER and SF networks show that the search efficiency is strongly correlated with the network controllability. The better the controllability, the less the adding links are, which can lead to the fact that the more the navigation nodes are distributed in the network, the more the search efficiency of the network can be enhanced.