Investigating mobile crowdsensing application performance

Mobile Crowdsensing (MCS) is an emerging distributed paradigm lying at the intersection between the Internet of Things and the volunteer/crowd-based approach. MCS applications are usually deployed on contributing nodes such as smart devices and mobiles, equipped by sensing resources that sample the physical environment and provide the sensed data, once filtered, aggregated and preprocessed, to the MCS application server. The MCS opportunistic approach unlocks new form of pervasive, participatory sensing applications, acquiring interests also in business contexts that call for adequate techniques and tools to drive architects and developers in MCS application design. Aim of this paper is to evaluate the performance of an MCS application though a stochastic model able to stochastically represent the overall MCS environment, thus providing a valid support to MCS application development. The Petri nets formalism is used due to its expressiveness and the capabilities to represent complex, dependent, non-Markovian, phenomena usually characterizing MCS environments. A specific MCS application is then evaluated to demonstrate the effectiveness of the proposed technique on a real case study.