4. Results of the experiment
In this section, we analyze and discuss the results obtained by means of numerical simulations with Network Simulator
2 based on the mobility patterns inferred by the experiment described in Section 3.
4.1. Scenario
In our experiment, we consider an emergency originated in one of two locations in Boston (see Fig. 2): (i) concert area,
Hatch Shell; (ii) shopping center, Prudential Center. For both the locations, we consider an enabling area radius equal to
100 m or 200 m and two target areas with radii of 1 km and 2 km, respectively. The enabling area radii and the target area
radii are modeled according to the lethal ranges and the minimal evacuation distances provided by the US Bureau of Alcohol,
Tobacco and Firearms [17], respectively.
We adopt the infection rate, i.e., the rate between the number of infected target devices and the total number of target
devices, as the performance metric. Clearly, due to human mobility some devices could enter or exit the target area during the
time. However, in our scenario a device remains a ‘target’ also after it leaves the target area, giving a conservative measure of
the performance that can be obtained. This assumption is reasonable, since each human that entered the target area at least
once is a potential target of the emergency. As an example, the emergency could be a terrorist bio-attack and the emergency
message could contain instructions on how to limit the injuries.
Since the speed of the alarm message diffusion deeply affects the amount of damage and/or number of injuries caused
by the emergency, we consider 30 min as the period of time available for alarm spreading and analyze the infection rate as
a function of time.