Performance and Energy Consumption Evaluation of Computation Offloading using CAOS D2D

Author(s)

dos Santos, Gabriel B. ; Trinta, Fernando A. M. ; Rego, Paulo A. L. ; Silva, Francisco A. ; de Souza, Jose N. ; IEEE

• Document type: Journal article
• Source: 2018 IEEE GLOBAL COMMUNICATIONS CONFERENCE (GLOBECOM); v. N/A, p. 7-pg., 2018-01-01.

Abstract

With the goal of simplifying the design and development of applications that use computation offloading, we developed the CAOS D2D platform to provide an abstraction layer for dealing with low-level tasks related to offloading of methods among Android mobile devices. This paper presents experiments performed to evaluate different aspects of the CAOS D2D, such as execution time and energy consumption of devices during method offloading. Besides that, we also performed experiments to evaluate the process of applications' dependency deployment. The experiments show that computation offloading can speedup the execution time by up to 4.55 times and reduce the power consumption by up to 88% in comparison to local executions of the same methods. (AU)

• FAPESP's process: 15/24144-7 - Technologies and solutions for enabling the cloud of things paradigm
• Grantee: José Neuman de Souza
• Support Opportunities: Research Projects - Thematic Grants