• cloud
• edge computing
• modelli stocastici
• opportunistic networks
• reti opportunistiche
• valutazione prestazioni

Opportunistic computing is the recent application of delay-tolerant networking to the creation of networks of mobile devices that give users the capability to share and access services provided by other (mobile) devices in proximity without using any cellular infrastructures. The importance of this paradigm becomes apparent given the ubiquitous proliferation of personal mobile devices in recent years. Opportunistic computing can also be used to realise service offloading, a recent trend in mobile networking research where resources on the edge of the cellular network are used in synergy with the cloud infrastructure. The importance of this application of opportunistic computing comes from the data traffic generated by mobile devices that, in the last few years, has been steadily increasing. While the development of LTE and LTE-A will boost cellular network capacity, it is unclear whether this would be enough to support the expected exponential increase in traffic demands in the medium term. Opportunistic techniques can contribute to solve this problem by offloading computation and data access to locally available devices, exploiting unused resources and balancing allocation of users requests to obtain an increase in service provisioning performances and avoiding network congestion.

This thesis brings contributions in two different scenarios: the first one is purely opportunistic with the detailing of a distributed system for the establishment and self-organization of mobile service provisioning. The system is established by each device autonomously collecting and using context information to individuate sequential compositions of resources for service provisioning and, thanks to a stochastic model, find the alternative that is expected to result in the lowest service provisioning time. In the second scenario, this thesis presents a solution for the integration of the opportunistic paradigm into a mobile edge system, where service provisioning is orchestrated between mobile devices and a remote cloud system thanks to the collaboration with network base stations local to the mobile devices. In the first scenario, experiments are presented to validate the decision algorithms and the stochastic model they rely on, while in the second scenario, we evaluate the performance gains obtained by using the opportunistic paradigm for service offloading in respect to traditional remote cloud systems.