A selective forwarding attack in mobile wireless sensor networks is an attack that selectively drops or delivers event packets as the compromised node moves. In such an attack, it is difficult to detect the compromised node compared with the selective forwarding attack occurring in the wireless sensor network because all sensor nodes move. In order to detect selective forwarding attacks in mobile wireless sensor networks, a fog computing-based system for a selective forwarding detection scheme has been proposed. However, since the proposed detection scheme uses a single path, the energy consumption of the sensor node for route discovery when the sensor node moves is large. To solve this problem, this paper uses fuzzy logic to determine the number of multi-paths needed to improve the energy efficiency of sensor networks. Experimental results show that the energy efficiency of the sensor network is improved by 9.5737% compared with that of the existing scheme after 200 seconds when using the proposed scheme.