The mainstream activity recognition technology depends on professional measurement equipment, which leads to the problem of difficult deployment and use. An activity identification technology based on the characteristics of existing WiFi hotspot received signal strength was proposed. The result shows that the proposed algorithm is capable of identifying the existence of a person in the room with 80% accuracy. And person's standing, walking and lying activity can be inferred with 95% accuracy. The walking direction can also be identified with 80% accuracy. The required signal of the proposed algorithm exists everywhere in daily life, and can be effectively used to identify indoor activities with low power consumption, and high precision.