Lee Young-Seok, Lee Ki-Hun, Jung Bang Chul
Department of Electronics Engineering, Chungnam National University, Daejeon 34134, Korea.
Sensors (Basel). 2020 Nov 12;20(22):6464. doi: 10.3390/s20226464.
In this paper, a novel beamforming technique is proposed as the over-the-air computation (AirComp) framework in a multiple-input multiple-output (MIMO) Internet-of-things (IoT) network consisting of multiple IoT sensors (STAs) and a single access point (AP). We assume that each IoT device has the channel state information (CSI) from itself to the AP and the AP has the global CSI of all IoT devices. We consider the mean squared error (MSE), which represents the reliability of function computation, as a performance metric. In short, each IoT device exploits maximum-ratio transmission (MRT) as a transmit beamforming technique to improve MSE performance by taking full advantage of multiple transmit antennae. Moreover, for the receive beamforming, we first consider a receive antenna selection (RAS) technique as the simplest beamforming method at the AP. Then, a semi-definite relaxation (SDR) method and a successive convex approximation (SCA) algorithm are considered and compared with each other in terms of MSE. Finally, we propose a novel two-step beamforming algorithm to further improve the MSE performance of the aforementioned techniques. We have numerically verified through computer simulations that the proposed framework has an improved MSE performance of about 6dB compared to the conventional single-input multiple-output (SIMO) AirComp, even with only two transmit antennae, and the modified MRT outperforms the other transmit beamforming techniques. Furthermore, the proposed receive beamforming technique, a two-step algorithm, shows the best performance in terms of MSE compared to prior studies.
在本文中，一种新颖的波束成形技术被提出，作为由多个物联网传感器（STA）和单个接入点（AP）组成的多输入多输出（MIMO）物联网（IoT）网络中的空中计算（AirComp）框架。我们假设每个物联网设备都有从自身到AP的信道状态信息（CSI），并且AP拥有所有物联网设备的全局CSI。我们将表示函数计算可靠性的均方误差（MSE）作为性能指标。简而言之，每个物联网设备利用最大比传输（MRT）作为发射波束成形技术，通过充分利用多个发射天线来提高MSE性能。此外，对于接收波束成形，我们首先将接收天线选择（RAS）技术视为AP处最简单的波束成形方法。然后，考虑了半定松弛（SDR）方法和逐次凸逼近（SCA）算法，并在MSE方面进行了相互比较。最后，我们提出了一种新颖的两步波束成形算法，以进一步提高上述技术的MSE性能。我们通过计算机仿真进行了数值验证，即使只有两个发射天线，所提出的框架与传统单输入多输出（SIMO）AirComp相比，MSE性能也提高了约6dB，并且改进后的MRT优于其他发射波束成形技术。此外，所提出的接收波束成形技术，即两步算法，与先前的研究相比，在MSE方面表现出最佳性能。
