Computation-bandwidth trading for mobile edge computing

We consider the problem of mobile computation offloading to both an edge cloud and to peer devices, and propose bandwidth incentives for those peer devices. Based on this idea, we formulate a joint optimization of the share of computations that are offloaded to the cloud and to peers, the identity o...

Descripción completa

Detalles Bibliográficos
Autores: Gupta, Sabyasachi, Lozano Solsona, Angel
Tipo de recurso: artículo
Estado:Versión aceptada para publicación
Fecha de publicación:2019
País:España
Institución:Universitat Pompeu Fabra
Repositorio:Repositorio Digital de la UPF
OAI Identifier:oai:repositori.upf.edu:10230/37088
Acceso en línea:http://hdl.handle.net/10230/37088
http://dx.doi.org/10.1109/CCNC.2019.8651745
Access Level:acceso abierto
Palabra clave:Task analysis
Bandwidth
Cloud computing
Resource management
Optimization
Mobile handsets
Conferences
Descripción
Sumario:We consider the problem of mobile computation offloading to both an edge cloud and to peer devices, and propose bandwidth incentives for those peer devices. Based on this idea, we formulate a joint optimization of the share of computations that are offloaded to the cloud and to peers, the identity of those assisting peers, and the allocation of computational resources at the cloud. The solutions derived from this optimization allow reducing by roughly 40% the completion time of computation tasks at the mobile devices, thereby facilitating the operation of latency-demanding applications.