Human behavior is characterized by a huge variability. Accordingly, automatic recognition and mathematical modeling of human activities are very difficult tasks even in relatively simple environments. Sensor technologies embedded into everyday living spaces provide us an unprecedented extent of information on people behavior, but these pervasive environments are often hard to set up and raise privacy issues. We present a Monte Carlo simulation, which may be used as tool for developing and testing behavior models. We demonstrate the model simulating the office life in one of our laboratories, and compare the results to actual measurements obtained with a sensor network.