Energy Efficient Mobile Cloud offloading for Image Processing Applications using Transfer Learning

Processing real-time images in the digital era is a complex task that requires the use of more sophisticated tools and consumes more energy. With the introduction of Hybrid model, Mobile and Cloud Computing, called as Mobile and Cloud Computing (MCC), this becomes an easy task in which the end user connects using the virtual instance in the Cloud environment. MCC enable the execution of cloud applications through mobile terminals. However, the mobile devices cannot process high end applications as they have limited energy. We propose Mobile Cloud offloading technique that excludes the execution on the Mobile terminal by deploying the applications on Cloud. Identification and decision on the energy aware applications are playing a vital role to improve Quality of Services (QoS). The proposed model introduces the Transfer Learning- based Energy efficient Mobile Cloud offloading framework. The proposed model predicts whether to move the applications towards the Mobile Cloud environment. The proposed model uses InceptionResNetV2 as pre-trained transfer learning approach.. The CNN based classifier used to classify the type of application and the respective energy consumption ratio based on the pre-trained model. The proposed model reduces training period to identify the application and effectively handle the offload process to improve the QoS metrics.