On the Characterization and Risk Assessment of AI-Powered Mobile Cloud Applications

Abstract Ultra-reliable low-latency communication supports powerful mission-critical applications such as artificial intelligence-enabled mobile cloud applications designed to deliver the quality of service and quality of experience to their users. However, whether existing security mechanisms are ready to address the risks emerging from these applications operating over ultra-fast 5G and 6G infrastructures is an open question. The complexity of finding answers to this question is partly due to the lack of means to measure software applications’ intelligence levels and partly due to the limitations of existing risk assessment approaches. In this paper, first, we propose an ability-based scale to characterize intelligent software applications. After that, we propose a semi-quantitative approach for threat modeling and risk analysis of intelligent software applications. Focusing on Android, we define three intelligent mobile cloud applications’ scenarios and demonstrate the feasibility of the proposed scale and approach. We perform their risk analyses for assessing the readiness of Android security mechanisms to mitigate their risks and identify open problems. We propose to rethink intelligent mobile cloud computing applications’ characterization and warn security experts to redesign their security mechanisms to serve evolving privacy, security, and trust requirements.