Interference alignment is a new technique recently proposed in physical layer wireless communications research, which purposely align the interference among multiple users thus increase the efficiency of channel use. This paper proposes a new method to enhance throughput of multi-hop wireless networks by utilizing interference alignment at physical layer. At first, it takes a simple network with only 6 nodes to demonstrate that interference alignment can achieve higher throughput. Subsequently, an optimization problem is formulated to maximize multi-hop wireless network throughput, which is a nonlinear programming (NLP) problem. Among that, finding the current transmission links can be transformed into maximal cliques of a graph. Furthermore, a branch-and-bound framework is provided to obtain the achievable bound of the NLP problem. Numerical results are presented to validate the algorithm and offer insights on the throughput enhancement brought by the interference alignment technique.