File(s) under permanent embargo
Opportunistic mobile crowd computing: Task-dependency based work-stealing
conference contribution
posted on 2023-02-13, 03:27 authored by SS Nagesh, Niroshinie FernandoNiroshinie Fernando, Seng LokeSeng Loke, A Neiat, Pubudu PathiranaPubudu PathiranaMobile devices are ubiquitous, heterogeneous and resource constrained. Execution of complex tasks in mobile devices are resource demanding and time-consuming, forcing developers to offload portions of the complex task to cloud or edge computing resources. Task offloading becomes increasingly challenging due to intermittent Internet connectivity, remote resource unavailability, high costs, latency, and limited energy of the mobile device. A mobile device user is typically surrounded by other mobile devices, which can be leveraged to collaboratively compute a resource-intensive task. With the help of a work sharing framework, it is feasible for devices to communicate and collaborate. However, some mobile devices are incapable of computing complex portions of the task, and some can compute in accelerated mode. In this demonstration, we introduce Honeybee-T a collaborative mobile crowd computing framework that uses a work-stealing algorithm. The algorithm allows work sharing with collaborating devices based on devices' computational ability and task-dependencies. The experiments show that by employing Honeybee-T framework, when compared to monolithic execution of a large compute-intensive task, there is a considerable performance gain, as well as energy savings.
History
Pagination
775-777Publisher DOI
ISBN-13
9781450391818Title of proceedings
Proceedings of the Annual International Conference on Mobile Computing and Networking, MOBICOMEvent
ACM MobiCom '22: The 28th Annual International Conference on Mobile Computing and NetworkingPublisher
ACMUsage metrics
Categories
No categories selectedKeywords
Licence
Exports
RefWorks
BibTeX
Ref. manager
Endnote
DataCite
NLM
DC