Channel reassignment is to assign again on the assigned channel resources in order to use the channel resources more efficiently. Channel reassignment in the Software-Defined Networking (SDN) based Internet of Things (SDN-IoT) is a promising paradigm to improve the communication performance of the network, since it allows software-defined routers (SDRs) with the help of SDN controller to appropriately schedule the traffic loads to meet the better transaction of corresponding channels in one link. However, the existing channel reassignment works have many limitations. In this paper, we develop a joint multi-channel reassignment and traffic control framework for the core backbone network in SDN-IoT. Comparing to classic performance metrics, we design a more comprehensive objection function to maximize the throughput and to minimize packet loss rate and the time delay by scheduling the appropriate traffic loads to corresponding channels in one link. We develop a Multi-Agent Deep Deterministic Policy Gradient (MADDPG)-based traffic control and multi-channel reassignment (TCCA-MADDPG) algorithm to optimize the objection function to achieve traffic control and channel reassignment. To tackle the dynamics and complexity of the core backbone network, we use the traffic prediction result as the part of the channel state information. In order to make better use of the time continuity of the channel state, we add an LSTM layer to the neural network in the experiment to capture the timing information of the channel. Simulation results show that the proposed algorithm converges faster and outperform existing methods.

中文翻译：

---

SDN-IoT中核心骨干网的联合流量控制和多通道重新分配：一种多代理深度强化学习方法

信道重新分配是在分配的信道资源上再次分配，以便更有效地使用信道资源。基于软件定义的网络（SDN）的物联网（SDN-IoT）中的信道重新分配是提高网络通信性能的一种有希望的范例，因为它允许借助SDN控制器的软件定义的路由器（SDR）来执行以下任务：适当地调度流量负载，以满足一个链接中相应频道的更好交易。但是，现有的频道重新分配工作具有许多局限性。在本文中，我们为SDN-IoT中的核心骨干网开发了一个联合的多通道重新分配和流量控制框架。与传统的效果指标相比，我们设计了一种更全面的异议功能，通过将适当的流量负载调度到一条链路中的相应通道，来最大化吞吐量并最大程度地减少丢包率和时间延迟。我们开发了一种基于多代理深度确定性策略梯度（MADDPG）的流量控制和多通道重新分配（TCCA-MADDPG）算法，以优化异议功能以实现流量控制和通道重新分配。为了解决核心骨干网的动态和复杂性，我们将流量预测结果用作信道状态信息的一部分。为了更好地利用通道状态的时间连续性，我们在实验中向神经网络添加了一个LSTM层，以捕获通道的时序信息。