Um Mecanismo de tomada de decisão de offloading relacionando custo energético e complexidade algorítmica

Abstract

Mobile Cloud Computing is a novel paradigm that uses remote infrastructure to improve resources of mobile devices. However, if MCC turns possible extend energy and computational capabilities of such devices, it also brings new challenges about where a task should be executed. Here we propose a offloading decision mechanism to reduce energy and decrease execution time considering algorithm complexity from a given procedure and the type of wireless connection such as 3G or WiFi. The proposal was validated and test by simulation and real implementation with a framework MCC. We implemented four existing algorithm (Binary Search (O(logn)), Linear Search (O(n)), Bubble Sort (O(n^2))and Matrix Multiplication (O(n^3)). The results show that procedures with complexity below (O(n^2)) with any I/O byte size and any network type do not consume energy enough to be executed remotely. Otherwise it is necessary to send information to execute a procedure remotely. We also observed that our proposed mechanism provides an economy of battery energy until 8% comparing with other offloading decision technique.