Selective forwarding attacks in mobile wireless sensor networks are classified as threat attacks because the compromised node moves to various areas and attempts to attack. A fog computing-based system has been proposed to detect selective forwarding attacks. However, since the detection scheme uses a single path, the energy consumption of the sensor node for route discovery is large. To solve this problem, multipath can be used. However, in regions where the multipath setting is not needed, this setting causes the sensor node to consume energy unnecessarily. The proposed scheme improves the energy efficiency of the sensor network by using fuzzy logic to determine whether to use multipath. The experimental results demonstrate that the proposed scheme achieves an energy improvement of about 8.3374% over 200 seconds.