Abstract
Deep learning is a powerful technology that enables intelligent data processing in the smart healthcare domain. Inspired by the tremendous processing power of cloud computing, the training process and the model repository of deep learning are moved to the cloud. Cloud-assisted deep learning applications enable smart mobile users to experience quick predictive results. Health professionals use smart mobile devices to convey recordings of the patient and to receive the best inference results. The mobility of these devices causes severe performance degradation as it increases the distance between its current location and the edge cloud where the virtual machines are provisioned. Therefore, mobility-based resource provisioning to identify a suitable server based on deadline constraints, available resources, and cost metrics is crucial. This paper proposes a proximity-based resource provisioning technique that guarantees minimal delay in obtaining inference results with a local mobile cloud system. The proposed technique comprises two algorithms (a) deadline-based initial resource provisioning and (b) resource migration and provisioning at suitable cloudlet during location change. The proposed technique is implemented in a mobile cloud platform running the inference method of a smart mobile healthcare application. The performance results show that the proposed technique outperforms the state-of-the-art techniques in terms of the response time, deadline meeting percentage, and system utilization.