Intelligent reflecting surface (IRS) has been increasingly considered in mobile edge computing (MEC), assisting smart terminals (STs) in offloading computationally-intense tasks to base stations (BSs). This paper presents a new IRS-assisted MEC framework, which jointly optimizes the local CPU frequencies of the STs, the receive beamformers of the BS, the ST offloading schedules, and the IRS phase configuration, to minimize the energy consumption of the STs. To this end, we reveal that the optimal CPU frequency is time-invariant for each ST. Under flat-fading channels, the IRS phases and the receive beamformers of the BS can be then decoupled from the offloading schedules. Based on this structure, we develop an alternating optimization to solve the IRS phase configuration and the receive beamformers, and then exploit the Lagrange duality method to solve the offloading schedules. We prove that the overall algorithm is guaranteed to compute a stationary point solution for the problem of interest with a low complexity. Under frequency-selective channels, we also develop a new alternating optimization algorithm to minimize the energy consumption, where manifold optimization is leveraged to effectively solve the IRS phase shifts. Numerical results show that the proposed algorithms are superior to existing techniques in terms of energy efficiency under both flat-fading and frequency-selective channels.

中文翻译：

---

智能反射面辅助移动边缘计算的能量最小化

智能反射面 (IRS) 在移动边缘计算 (MEC) 中得到越来越多的考虑，帮助智能终端 (ST) 将计算密集型任务卸载到基站 (BS)。本文提出了一种新的 IRS 辅助 MEC 框架，该框架联合优化了 ST 的本地 CPU 频率、BS 的接收波束形成器、ST 卸载时间表和 IRS 相位配置，以最小化 ST 的能耗。为此，我们揭示了每个 ST 的最佳 CPU 频率是时不变的。在平坦衰落信道下，可以将 IRS 相位和 BS 的接收波束形成器从卸载调度中解耦。基于这种结构，我们开发了一种交替优化来解决 IRS 相位配置和接收波束形成器，然后利用拉格朗日对偶方法求解卸载计划。我们证明了整个算法可以保证以低复杂度计算感兴趣的问题的固定点解决方案。在频率选择通道下，我们还开发了一种新的交替优化算法来最小化能耗，其中利用流形优化来有效地解决 IRS 相移。数值结果表明，所提出的算法在平坦衰落和频率选择信道下的能量效率方面都优于现有技术。我们还开发了一种新的交替优化算法来最小化能源消耗，其中利用流形优化来有效地解决 IRS 相移。数值结果表明，所提出的算法在平坦衰落和频率选择信道下的能量效率方面都优于现有技术。我们还开发了一种新的交替优化算法来最小化能源消耗，其中利用流形优化来有效地解决 IRS 相移。数值结果表明，所提出的算法在平坦衰落和频率选择信道下的能量效率方面都优于现有技术。