• crowdsensing
• mobile
• injection
• coverage
• integrity
• probabilistic

Mobile Crowdsensing is a technique that leverages to the sensing capabilities of user's devices to gather data. Each device generates and shares a coverage map, which is a probability vector, where each element defines the probability of covering a specific location.
This thesis discusses two mobile crowdsensing paradigms:
1. Social-Oblivious Model: the location tagging is performed in a classical distributed manner
2. Social-Aware mode: opportunistic interactions among the users are used to swap. As the number of swaps increases, the correlation between each node and its coverage map decreases.
The thesis also introduces two attack scenarios where a node acts as an attacker is able to inject false mobility data. The aim is to tamper the analysis that will be performed by the server on the collective mobility data received from all nodes. We also assume the attacker has complete knowledge of other coverage maps.
The two attack scenarios are as follows:
1. The attacker identifies a data distribution that is statistically different from the rest of the nodes and injects it into the server.
2. During the swap mechanisms, the attacker injects false mobility data, altering its coverage map with Gaussian noise characterized by mean M and variance W.