We open-sourced the ULOOF (User-Level Online Offloading) Framework, result of a long-lasting collaboration with UPEM, UFMG, UNIBO and Google.
The framework in its current form allows to optimize an arbitrary Android application, analyzing its method-level structure, determine which are offloadable in a composite off-line and run-time mode, and offloading methods to a nearby android server (e.g., a power-plugged powerful MEC host or personal computer/device) based on latency and energy consumption profiling of the applicaiton and the device using machine learning. Results show gains in energy consumption and execution time from 30% to 50%, roughly.
More details on the project and the code:
http://uloof.roc.cnam.fr
Related publications
José L.D. NETO, Se-young YU, Daniel F. MACEDO, José M.S. NOGUEIRA, Rami LANGAR, Stefano SECCI, “ULOOF: a User-Level Online Offloading Framework for Mobile Edge Computing“, IEEE Transactions on Mobile Computing, Vol. 17, No. 11, pp: 2660-2674, Nov. 2018. [Abstract]
Alessandro ZANNI, Se-young YU, Paolo BELLAVISTA, Rami LANGAR, Stefano SECCI, “Automated Selection of Offloadable Tasks for Mobile Computation Offloading in Edge Computing“, Proc. of 2017 Conference on Network and Service Management (CNSM), Nov. 26-30, 2017, Tokyo, Japan. Short paper. [Abstract]
