Compared with traditional half-duplex wireless systems, the application of emerging full-duplex (FD) technology can potentially double the system capacity theoretically. However, conventional techniques for suppressing self-interference (SI) adopted in FD systems require exceedingly high power consumption and expensive hardware. In this paper, we consider employing an intelligent reflecting surface (IRS) in the proximity of an FD base station (BS) to mitigate SI for simultaneously receiving data from uplink users and transmitting information to downlink users. The objective considered is to maximize the weighted sum-rate of the system by jointly optimizing the IRS phase shifts, the BS transmit beamformers, and the transmit power of the uplink users. To visualize the role of the IRS in SI cancellation by isolating other interference, we first study a simple scenario with one downlink user and one uplink user. To address the formulated non-convex problem, a low-complexity algorithm based on successive convex approximation is proposed. For the more general case considering multiple downlink and uplink users, an efficient alternating optimization algorithm based on element-wise optimization is proposed. Numerical results demonstrate that the FD system with the proposed schemes can achieve a larger gain over the half-duplex system, and the IRS is able to achieve a balance between suppressing SI and providing beamforming gain.

翻译：与传统的半双工无线系统相比，新兴的全双工（FD）技术在理论上可使系统容量翻倍。然而，FD系统中用于抑制自干扰（SI）的传统技术需要极高的功耗和昂贵的硬件。本文考虑在FD基站（BS）附近部署智能反射面（IRS），以缓解同时接收上行用户数据和向下行用户发送信息时的自干扰问题。目标是通过联合优化IRS相移、BS发射波束成形以及上行用户发射功率，最大化系统的加权和速率。为直观展示IRS在消除其他干扰后对自干扰消除的作用，我们首先研究一个包含一个下行用户和一个上行用户的简单场景。针对所提出的非凸问题，提出了一种基于逐次凸逼近的低复杂度算法。对于包含多个下行和上行用户的一般情况，提出了一种基于逐元素优化的高效交替优化算法。数值结果表明，采用所提方案的全双工系统相比半双工系统可获得更大增益，且IRS能够在抑制自干扰与提供波束成形增益之间实现平衡。