A Wireless Sensor Network (WSN) requires a high level of robust and fault tolerant sensing and actuating capabilities, specially when the application aims to gather delicate and urgent data with reasonable latency. Hence, verifying the behavior properties under the presence of faults remains an important step in developing an application over a WSN. A comprehensive study on characterization and understanding of all the possible faults is required in order to generate and inject 'any' known error to the system. In order to ensure appearance of all the faults and possible bugs in the system, conception and developing a fault injector which generates and injects any requested fault to the system is promising. This becomes more important and critical when the fault happens very rarely, while due to Murphy's law it happens certainly along the network life. Considering that occurrence of faults depends heavily on the specifications of the use case, in this paper we concentrate on a sensor network which aims to detect the presence of vehicles on parking lots. We try to categorize and characterize the faults driven by this system as the first step of developing a fault injector.