Distributing Aggregate Computations on top of Akka Actors

In the context of the Internet of Things, development of large-scale, adaptive systems usually focuses on the behavior of the single device. Aggregate programming is a paradigm that provides an alternative approach, in which the basic unit of computing is a cooperating collection of devices, instead of a single device. scafi is a Scala framework for aggregate programming, and provides an Akka-based platform for aggregate applications, supporting both peer-to-peer and server-based networks. Moreover, scafi offers a simulator module for the simulation of an aggregate system. The work described in this thesis consists in the analysis of scafi, in the partial re-engineering of its internal actor platform, and in the development of new features. The main goal is to enhance the flexibility of scafi in a distributed context, promoting its adoption for programming spatial systems. First of all, communication between distributed nodes is enabled, by defining a JSON-based serialization strategy, which promotes interoperability. A hybrid platform is also introduced, exploiting a peer-to-peer communication between devices, with a central unit that manages all the relevant space related information. This platform fills the main gap of the peer to-peer approach in a distributed environment: tracking of remote devices. Moreover, a code mobility approach is implemented, allowing the assignment of new programs to devices, at runtime. Lastly, the concept of monitoring a distributed aggregate system emerged, leading to the development of a graphical user interface, observing the devices in a running system. In this thesis, I present the new architecture and API of the actor platform of scafi, designed with the aim of ensure a more flexible approach for the development of distributed applications with aggregate computing.