Cache-equipped Base-Stations (CBSs) is an attractive alternative to offload the rapidly growing backhaul traffic in a mobile network. New 5G technology and dense femtocell enable one user to connect to multiple base-stations simultaneously. Practical implementation requires the caches in BSs to be regarded as a cache server, but few of the existing works considered how to offload traffic, or how to schedule HTTP requests to CBSs. In this work, we propose a DNS-based HTTP traffic allocation framework. It schedules user traffic among multiple CBSs by DNS resolution, with the consideration of load-balancing, traffic allocation consistency and scheduling granularity of DNS. To address these issues, we formulate the user-traffic allocation problem in DNS-based mobile edge caching, aiming at maximizing QoS gain and allocation consistency while maintaining load balance. Then we present a simple greedy algorithm which gives a more consistent solution when user-traffic changes dynamically. Theoretical analysis proves that it is within 3/4 of the optimal solution. Extensive evaluations in numerical and trace-driven situations show that the greedy algorithm can avoid about 50% unnecessary shift in user-traffic allocation, yield more stable cache hit ratio and balance the load between CBSs without losing much of the QoS gain.

中文翻译：

---

移动边缘缓存中一致的用户流量分配和负载平衡

配备缓存的基站 (CBS) 是一种有吸引力的替代方案，可以卸载移动网络中快速增长的回程流量。新的 5G 技术和密集的 femtocell 使一个用户能够同时连接到多个基站。实际实现需要将 BS 中的缓存视为缓存服务器，但现有工作很少考虑如何卸载流量，或如何调度对 CBS 的 HTTP 请求。在这项工作中，我们提出了一个基于 DNS 的 HTTP 流量分配框架。它通过DNS解析在多个CBS之间调度用户流量，同时考虑DNS的负载均衡、流量分配一致性和调度粒度。为了解决这些问题，我们制定了基于 DNS 的移动边缘缓存中的用户流量分配问题，旨在最大限度地提高 QoS 增益和分配一致性，同时保持负载平衡。然后我们提出了一个简单的贪心算法，当用户流量动态变化时，它给出了一个更一致的解决方案。理论分析证明它在最优解的3/4以内。在数值和跟踪驱动情况下的广泛评估表明，贪婪算法可以避免用户流量分配中大约 50% 的不必要转移，产生更稳定的缓存命中率并平衡 CBS 之间的负载，而不会失去太多的 QoS 增益。