We consider the problem of storing segments of encoded versions of content
files in a set of base stations located in a communication cell. These base
stations work in conjunction with the main base station of the cell. Users move
randomly across the space based on a discrete-time Markov chain model. At each
time slot each user accesses a single base station based on it's current
position and it can download only a part of the content stored in it, depending
on the time slot duration. We assume that file requests must be satisfied
within a given time deadline in order to be successful. If the amount of the
downloaded (encoded) data by the accessed base stations when the time deadline
expires does not suffice to recover the requested file, the main base station
of the cell serves the request. Our aim is to find the storage allocation that
minimizes the probability of using the main base station for file delivery.
This problem is intractable in general. However, we show that the optimal
solution of the problem can be efficiently attained in case that the time
deadline is small. To tackle the general case, we propose a distributed
approximation algorithm based on large deviation inequalities. Systematic
experiments on a real world data set demonstrate the effectiveness of our
proposed algorithms. Index Terms: Mobility-aware Caching, Markov Chain, MDS
Coding, Small-cell Networks

该研究主要研究如何在具有移动性的小型网络中存储内容文件的编码版本，并提出了分布式逼近算法来解决该问题。在实验中，使用基于大偏差不等式的方法使得研究结论在实际应用中更加精准。