On Data Staging Strategies for Mobile Accesses to Cloud Services

In this paper, we design and develop time-efficient algorithms that identify vantage locations to stage a shared data item for the ease of mobile accesses. As mobile users will be charged on the use of communication bandwidth as well as on the fraction of the time the storage resources are kept busy on their respective data, we aim to provide such access for a class of mobile users using Cloud services with minimum total cost. To this end, we adopt a homogeneous cost model and network model used in [1], and consider this problem under a practical restriction that the maximum number of instant copies is bounded by a given constant k. We first study this optimal problem without constraints and then investigate the constrained version by limiting the relationship between the communication cost C and caching (storage) cost S of a copy of the requested data and point out the limitation of this method. For arbitrary C and S, we propose three heuristic algorithms to this problem. All the proposed algorithms are built upon the optimal solution (using dynamic programming (DP)) without constraints on the number of copies, but leverage different strategies to respect the constraint while minimizing the cost. We shown that they have the same time complexity but exhibit different actual performance in different situations. We validate our findings via extensive simulation studies.