A Dynamic Application Partitioning and Offloading Framework to Enhance the Capabilities of Transient Clouds Using Mobile Agents

Mobile cloud computing has emerged as a prominent area of research, a natural extension of cloud computing that proposes to offer solutions for enhancing the capabilities of smart mobile devices commonly plagued by resource constraints. As one of its promising models, transient clouds aim to address the internet connectivity shortfall inherent in most solutions through the formation of ad hoc networks by devices in close proximity, then the offloading some computations (Cyber Foraging) to the created cloud. However, transient clouds, at their current state, have several limitations, concerning their expansion on a local network having a large number of devices and the management of the instability of the network due to the constant mobility of the devices. Another issue is the fact code partitioning and offloading are not addressed to fit the need of such networks, thereby rendering the distributed computing mechanism barely efficient for the Transient Cloud. In this study, we propose a transient cloud-based framework that exploits the use of multi-agent systems, enabling a dynamic partitioning and offloading of code, and facilitating the movement and the execution of code partition packets in a multi-hop ad-hoc mesh network. When created and deployed, these intelligent mobile agents operate independently or collaboratively and adapt to the continual entry and exit of devices in the neighbourhood. The integration of these trending concepts in distributed computing within a framework offers a new architecture for resource-sharing among cooperating devices that addresses the varied issues that arise in dynamic environments