Deep Convolutional Neural Network with Modified Fuzzy C-Means Clustering Based Mobile Applications Offloading to Clouds

摘要

The emerging mobile cloud has prolonged the vista of application development as well as organization with techniques, such as, code offloading, which is taken for protecting the energy and intensifying responsiveness of mobile devices, the technique till now have so many confronts associatingwith usefulness. An accurately offloading mobile application to cloud environment is affianced by means of using a Deep Convolutional Neural Network (DCNN). Primarily, in this research, important features that include bandwidth, latency, memory, data size, availability, and these featureswere extracted from mobile devices. Here, so as to reduce the computational cost, by means of Modified Fuzzy C-implies Clustering (MFCMC) algorithm, the features containing similar info are collected as alike cluster. Subsequent to clustering, by means of the energy of cloud servers utilizingDCNN, the thread-based offloading is affianced with precise as well as trade-off decision making system for improving the parts of computational execution in addition to energy-efficiency of the mobile device. By utilizing Karush–Kuhn–Tucker (KKT) conditions, the code offloadingchoice is made, a nonlinear optimization solver, which gets free off the weight of handling a heavyweight linear optimization problem, opposing to former works. DCNN based model is proposed for more accurately examining the consuming execution time by presuming regarding the dynamic conductof nature as well as application use. The test results illustrate that the execution of presented method attained desirable results while matched up with the previous Properly Offloading Mobile Applications to Clouds (POMAC) and adjusted Modified POMAC (M-POMAC) w.r.t execution time, predictionaccuracy and energy saving.