Low duty cycle is proposed to reduce the energy consumption of WSNs (wireless sensor networks), thereby extending the lifecycle of WSNs. However, low duty cycle makes neighbor discovery extremely difficult. Especially considering the mobility of nodes, effective neighbor discovery is more challenging. In this work, a new neighbor discovery algorithm based on Continuous Torus Quorum is proposed to solve the neighbor discovery problem in asynchronous symmetric and asymmetric low duty cycle WSNs. A neighbor discovery probability is also provided to estimate efficiency of neighbor discovery algorithms in mobile scene. Furthermore, a simulation platform is developed to measure performance of neighbor discovery algorithms. Both theoretical analysis and simulation results reveal that Continuous-Torus-Quorum-based algorithm can achieve significant performance improvement over several classical heterogeneous neighbor discovery algorithms, such as Disco and U-Connect, in terms of energy efficiency, discovery delay and discovery probability in the symmetric and asymmetric scenes.