Miranda, Hugo Alexandre TavaresLima, Diogo2025-02-102025-02-102024-02-232023-09-18http://hdl.handle.net/10400.5/98265Today’s distributed mobile applications assume a permanent network connectivity to mediate the interactions between a large number of users, coordinated by resourceful servers located in Cloud datacenters. This centralization simplifies application state management, but creates a non-negligible physical distance between the application state and the users. As this distance extend for possibly thousands of kilometers, users are exposed to high latency and uncertain response times (jitter), due to propagation time and the competition for the available bandwidth on backbone networks. Fog Computing proposes to overcome this problem by approximating the Cloud from the users with the deployment of surrogate servers, providing computing power and memory to manage application state, at the edge of the network. A judicious distribution of application state fragments by surrogates gives to Fog Computing the potential to mitigate latency and reduce load in the backbone networks. However, to achieve it, two problems that could not be found in the centralized cloud approach must be addressed. Firstly, the geo-aware state deployment (GeoSD) problem focuses on being able to deploy each of the application state fragmented components on the surrogate at the most favourable location. The deployment algorithm must consider that, even within a simple application, state components do not behave equally. Secondly, the distributed environment problem is concerned with providing state consistency, in particular when operations involve application state components stored on two or more surrogates. Maintaining consistency is crucial to guarantee global application logic. This problem is further aggravated by the fact that surrogates are also susceptible to crash or suffer network partitions. This thesis presents an adaptive geographical-aware state deployment middleware that aggregates a geo-aware state deployment algorithm with a scalable distributed environment framework that supports data consistency and replication.engDistributed mobile applicationsFog computingGeographical state deploymentConsistencyLightweight groupsAplicações móveis distribuídasComputação na neblinaDisposição geográfica de dadosCoerênciaGrupos levesdoctoral thesis101512147