The Random Access (RA) procedure is used to request channel resources for the uplink data transmission in the cellular-based massive Internet of Things (mIoT). To ease the RA failure and the network congestion, power ramping (PR) technique is used to step up the preamble transmit power after each unsuccessful RA attempt. In this paper, we develop a traffic aware spatio-temporal model to analyze the PR scheme in the mIoT network, where the Signal-to-Interference-and-Noise Ratio (SINR) outage and collision events jointly determine the traffic evolution and the RA success probability. Compared with existing literature only modelled collision from single cell perspective, we model both the SINR outage and the collision from the network perspective. Based on this analytical model, we derive the exact expression for the RA success probability to show the effectiveness of the PR scheme. Our results show that the geometry PR scheme with smooth increased transmission power is effective in heavy traffic scenario in terms of increasing the RA success probability.