Security-aware service migration for tactical mobile micro-clouds

摘要

Mobile micro-clouds (MMCs), which are small cloud-like infrastructures that can host services for mobile users, provide a promising approach to deliver data and computation to users in low bandwidth/high latency networks. In this work, we take a step towards security-aware service migration for multiple servers for MMC in tactical environments. As a fundamental model, we consider two users sharing a cluster of MMCs and moving between coverage areas of different MMCs according to random walk. We formulate the problem using a finite horizon Markov Decision Process (MDP). Due to the coupling between users due to security (and possibly interference) costs, we observe that the problem is significantly more complex compared with the non-security-aware single user service migration problem studied in the literature, even in the simplest case of two users. Accordingly, we consider efficient control algorithms which aim to minimize time average costs. Particularly, we propose a modified myopic policy which aims to reduce time average costs by steering operation towards settings which would reduce the likelihood of future security cost increases, by proactively keeping the services distant. Our numerical results demonstrate that our modified myopic policy outperforms other policies in terms of average cost for various settings.