论文标题
网络性能估算器，以及用于物联网多媒体应用程序路由选择的应用程序
Network Performance Estimator with Applications to Route Selection for IoT Multimedia Applications
论文作者
论文摘要
储层计算是预测湍流的有力工具，其简单的架构具有处理大型系统的计算效率。然而，其实现通常需要完整的状态向量测量和系统非线性知识。我们使用非线性投影函数将系统测量扩展到高维空间，然后将其输入到储层中以获得预测。我们展示了这种储层计算网络在时空混沌系统上的应用，该系统模拟了湍流的若干特征。我们表明，使用径向基函数作为非线性投影器，即使只有部分观测并且不知道控制方程，也能稳健地捕捉复杂的系统非线性。最后，我们表明，当测量稀疏、不完整且带有噪声，甚至控制方程变得不准确时，我们的网络仍然可以产生相当准确的预测，从而为实际湍流系统的无模型预测铺平了道路。
Estimating the performance of multimedia traffic is important in numerous contexts, including routing and forwarding, QoS provisioning, and adaptive video streaming. This paper proposes a network performance estimator which aims at providing, in quasi real-time, network performance estimates for IoT multimedia traffic in IEEE 802.11 multihop wireless networks. To our knowledge, the proposed multimedia-aware performance estimator, or MAPE, is the first deterministic simulation-based estimator that provides real-time per-flow throughput, packet loss and delay estimates while considering inter-flow interference and multi-rate flows, typical of multimedia traffic. Our experimental results indicate that MAPE is able to provide network performance estimates that can be used by IoT multimedia services, notably to inform real-time route selection in IoT video transmission, at a fraction of the execution time when compared to stochastic network simulators. When compared to existing deterministic simulators, MAPE yields higher accuracy at comparable execution times due to its ability to consider multi-rate flows.
