论文标题
具有严格延迟约束的最佳云网络控制
Optimal Cloud Network Control with Strict Latency Constraints
论文作者
论文摘要
通过分布式计算网络（例如，移动边缘计算），及时交付了资源密集型和对潜伏期敏感的服务（例如工业自动化，增强现实）正在引起人们越来越多的关注。由于现有研究提供的平均延迟性能保证的不足，我们集中于在每包中提供下一代实时服务的关键目标，同时将整体云网络资源成本降至最低。我们介绍了一个新颖的排队系统，该系统能够跟踪数据包的寿命，并使用严格的截止日期约束将最佳云网络控制问题形式化。在说明在由于终生到期之前将数据包交付到目的地之前的主要挑战之后，我们构建了同等的配方，其中放松的流量保护允许利用Lyapunov优化，从而得出一种近乎最佳的完全分布的算法，以解决原始问题。数值结果验证了理论分析，并显示了与最新的云网络控制相比，提出的控制策略的出色性能。
The timely delivery of resource-intensive and latency-sensitive services (e.g., industrial automation, augmented reality) over distributed computing networks (e.g., mobile edge computing) is drawing increasing attention. Motivated by the insufficiency of average delay performance guarantees provided by existing studies, we focus on the critical goal of delivering next generation real-time services ahead of corresponding deadlines on a per-packet basis, while minimizing overall cloud network resource cost. We introduce a novel queuing system that is able to track data packets' lifetime and formalize the optimal cloud network control problem with strict deadline constraints. After illustrating the main challenges in delivering packets to their destinations before getting dropped due to lifetime expiry, we construct an equivalent formulation, where relaxed flow conservation allows leveraging Lyapunov optimization to derive a provably near-optimal fully distributed algorithm for the original problem. Numerical results validate the theoretical analysis and show the superior performance of the proposed control policy compared with state-of-the-art cloud network control.
