DAMMP: A Distributed Actor Model for Mobile Platforms

While mobile computing has seen a trend towards miniaturization and energy savings for a number of years, the available hardware parallelism in mobile devices has at the same time continued to increase. Overall, mobile devices remain resource constrained on power consumption and thermal dissipation. Aggregating the computing capabilities of multiple mobile devices in a distributed and dynamic setting, opens the possibilities for performance improvements, longer aggregate battery life and novel dynamic and distributed applications. In this paper, we propose a Distributed Actor Model for Mobile Platforms (DAMMP), which includes a) a mobile extension to the actor-based Distributed Selector (DS) programming model, along with a new implementation for mobile Android devices, b) an extension to the DS programming model that enables the programmer to react and adapt to dynamic changes in device availability, c) an adaptive mobile-to-server and mobile-to-mobile computation offloading model and its implementation on the Android platform, and d) creation of a dynamic network of heterogeneous Android devices using both Wi-Fi Soft AP and Wi-Fi Direct's peer to peer (P2P) network. We evaluate the DAMMP framework under ideal thermally-controlled usage conditions to show promising scalability and performance, and analyze the communication overhead of both Wi-Fi and Wi-Fi Direct when used as the communication layer for DAMMP. We also evaluate the impact of adaptive offload on device-level thermal dissipation in more realistic usage scenarios, thereby demonstrating possibilities for thermal control and power management that can be achieved at the application level with a distributed actor model. To the best of our knowledge, this work is the first cross-platform distributed actor/selector runtime system that can span mobile devices and distributed servers.