Optimal Transmission Power and Controller Design for Networked Control Systems Under State-Dependent Markovian Channels
Optimal Transmission Power and Controller Design for Networked Control Systems Under State-Dependent Markovian Channels
复制标题
DOI：
10.1109/tac.2022.3181758
复制
发表时间：
2022-10
影响因子：
6.8
通讯作者：
B. Hu;T. Tamba
中科院分区：
文献类型：
--
作者：
B. Hu;T. Tamba
本文考虑了工业网络控制系统的协同设计问题，以确保此类系统的稳定性和效率特性。由于工业环境中的无线通信不仅受到阴影衰落的影响，而且还受到与其周围环境相关的随机干扰的影响，因此该问题特别具有挑战性。本文首先介绍了一种新的状态依赖马尔可夫信道（SD-MC）模型，明确捕捉状态依赖的工业无线通信系统的功能，通过定义所提出的模型的转移概率作为环境的状态和传输功率的函数。在SD-MC模型下，给出了状态依赖衰落信道下非线性控制系统的最大允许传输间隔的充分条件，使系统具有期望渐近稳定性和几乎处处渐近稳定性.基于这些稳定性条件，协同设计问题，然后制定为一个约束多项式优化问题（CPOP），这可以有效地解决使用半定规划方法的情况下，一个两状态SD-MC模型。这样的CPOP的解决方案表示最优控制和功率策略，其在无限时间范围内优化平均预期联合成本，同时尊重稳定性约束。对于一般的SD-MC模型，本文进一步证明了次优解可以从所考虑的CPOP的线性规划公式中获得。仿真结果表明了所提出的协同设计方案的有效性。
This article considers a codesign problem for industrial networked control systems to ensure both stability and efficiency properties of such systems. This problem is particularly challenging due to the fact that wireless communications in industrial environments are not only subject to shadow fading, but also stochastically correlated with their surrounding environments. This article first introduces a novel state-dependent Markov channel (SD-MC) model that explicitly captures the state-dependent features of industrial wireless communication systems by defining the proposed model’s transition probabilities as a function of both environments’ states and transmission power. Under the SD-MC model, sufficient conditions on Maximum Allowable Transmission Interval are presented to ensure both asymptotic stability in expectation and almost sure asymptotic stability properties of a nonlinear control system with state-dependent fading channels. Based on these stability conditions, the codesign problem is then formulated as a constrained polynomial optimization problem (CPOP), which can be efficiently solved using semidefinite programming methods for the case of a two-state SD-MC model. The solutions to such a CPOP represent optimal control and power strategies that optimize the average expected joint costs in an infinite time horizon while respecting the stability constraints. For a general SD-MC model, this article further shows that suboptimal solutions can be obtained from linear programming formulations of the considered CPOP. Simulation results are given to illustrate the efficacy of the proposed codesign scheme.