This paper considers a problem of real-time control of uplink transmissions over dual links. In the problem, a UE is configured with dual links. We propose a joint control algorithm to choose a primary link and activate or deactivate each link in every slot. The activation of a link incurs uplink transmission power. Switching a primary link also incurs cost. The control algorithm aims to jointly minimize the transmission power and primary link switching cost, while minimizing transmission queues to reduce delays. The problem introduces several constraints on transmission power allocation and link activation. In particular, the secondary link can be activated only when it has non-empty transmission queues and the primary link is activated. For the primary link, no such restriction is imposed. These heterogeneous transmission constraints require the decisions to treat a time-varying decision set. Furthermore, the primary link switching cost may enforce the previous decisions to skew the current decision. To tackle such difficulties, we use Lyapunov optimization techniques and develop a control algorithm. We investigate the algorithm structure from the decisions made for special cases. We reveal that the primary link selection has a Shortest-Queue with Hysteresis structure. We also show that link activation control largely relies on transmission efficiency. We analyze the performance bound of our algorithm in terms of cost and queues to show that our algorithm can achieve an optimal target value. Finally, we present simulation results to verify the performance of our algorithm.

本文考虑了双链路上行链路传输的实时控制问题。在该问题中，UE配置了双链路。我们提出了一种联合控制算法来选择主链接并激活或停用每个时隙中的每个链接。链路的激活会导致上行链路传输功率。切换主链路也会产生成本。该控制算法旨在共同最小化传输功率和主链路切换成本，同时最小化传输队列以减少延迟。该问题引入了对传输功率分配和链路激活的几个约束。特别地，次链路只有在传输队列非空且主链路被激活时才能被激活。对于主链路，不施加此类限制。这些异构传输约束需要决策来处理随时间变化的决策集。此外，主链路切换成本可能会强制执行先前的决定以歪曲当前的决定。为了解决这些困难，我们使用了李雅普诺夫优化技术并开发了一种控制算法。我们从为特殊情况做出的决定中研究算法结构。我们发现主链接选择具有具有滞后结构的最短队列。我们还表明链路激活控制很大程度上依赖于传输效率。我们在成本和队列方面分析了我们算法的性能界限，以表明我们的算法可以实现最佳目标值。最后，我们给出了仿真结果来验证我们算法的性能。