Extending Queuing Networks to Assess Mobile CrowdSensing Application Performance

The widespread and pervasive diffusion of smart devices is boosting Internet of Things and contribution-based paradigms. In particular, Mobile Crowdsensing (MCS), due to its big potential of sharing and collecting large population of contributors-devices, is acquiring interest. Devices such as smartphones and in-vehicle ones are equipped with different sensors and actuators are able to probe data about the physical environment. In a typical MCS scenario, data produced by sensors are sent to the remote server, where they are collected and processed by the applications. In order to exploit the MCS paradigm in large-scale business contexts it is required to guarantee the quality of service. Therefore, techniques and tools able to represent and evaluate MCS system quality attributes such as performance and energy consumption are required. To model MCS system is quite challenging since not only the number of users but also the number of contributors may vary. In this paper, we propose to adopt queuing networks, a well-known formalism able to deal with large number of requests, to address this issue. In particular we introduce and implement a new policy allowing the number of server to be variable. The proposed model is then adopted in the evaluation of an example, providing interesting insights on contribution, provisioning and usage impacts in terms of performance and energy consumption.