This study demonstrates the feasibility of the proactive received power prediction by leveraging spatiotemporal visual sensing information toward the reliable millimeter-wave (mmWave) networks.Since the received power on a mmWave link can attenuate aperiodically due to a human blockage, the long-term series of the future received power cannot be predicted by analyzing the received signals before the blockage occurs. We propose a novel mechanism that predicts a time series of the received power from the next moment to even several hundred milliseconds ahead. The key idea is to leverage the camera imagery and machine learning (ML). The time-sequential images can involve the spatial geometry and the mobility of obstacles representing the mmWave signal propagation. ML is used to build the prediction model from the dataset of sequential images labeled with the received power in several hundred milliseconds ahead of when each image is obtained. The simulation and experimental evaluations using IEEE 802.11ad devices and a depth camera show that the proposed mechanism employing convolutional LSTM predicted a time series of the received power in up to 500 ms ahead at an inference time of less than 3 ms with a root-mean-square error of 3.5 dB.Parts of this work were presented at the 85th IEEE Vehicular Technology Conference (VTC Spring) and the IEEE Consumer Communications and Networking Conference (CCNC). 2 Index Terms millimeter-wave communications, link quality prediction, proactive prediction, machine learning, supervised learning, depth image Received power (dBm) Grand truth (a) When the camera was at A low . Received power (dBm) Grand truth (b) When the camera was at A high .

No abstract

No abstract

This study develops a federated learning (FL) framework overcoming largely incremental communication costs due to model sizes in typical frameworks without compromising model performance. To this end, based on the idea of leveraging an unlabeled open dataset, we propose a distillation-based semi-supervised FL (DS-FL) algorithm that exchanges the outputs of local models among mobile devices, instead of model parameter exchange employed by the typical frameworks. In DS-FL, the communication cost depends only on the output dimensions of the models and does not scale up according to the model size. The exchanged model outputs are used to label each sample of the open dataset, which creates an additionally labeled dataset. Based on the new dataset, local models are further trained, and model performance is enhanced owing to the data augmentation effect. We further highlight that in DS-FL, the heterogeneity of the devices' dataset leads to ambiguous of each data sample and lowing of the training convergence. To prevent this, we propose entropy reduction averaging, where the aggregated model outputs are intentionally sharpened. Moreover, extensive experiments show that DS-FL reduces communication costs up to 99% relative to those of the FL benchmark while achieving similar or higher classification accuracy.

No abstract

No abstract

For millimeter-wave networks, this paper presents a paradigm shift for leveraging time-consecutive camera images in handover decision problems. While making handover decisions, it is important to predict future long-term performance-e.g., the cumulative sum of time-varying data rates-proactively to avoid making myopic decisions. However, this study experimentally notices that a time-variation in the received powers is not necessarily informative for proactively predicting the rapid degradation of data rates caused by moving obstacles. To overcome this challenge, this study proposes a proactive framework wherein handover timings are optimized while obstacle-caused data rate degradations are predicted before the degradations occur. The key idea is to expand a state space to involve timeconsecutive camera images, which comprises informative features for predicting such data rate degradations. To overcome the difficulty in handling the large dimensionality of the expanded state space, we use a deep reinforcement learning for deciding the handover timings. The evaluations performed based on the experimentally obtained camera images and received powers demonstrate that the expanded state space facilitates (i) the prediction of obstacle-caused data rate degradations from 500 ms before the degradations occur and (ii) superior performance to a handover framework without the state space expansion.

Full-duplex relaying (FDR), i.e., simultaneous transmission and reception using the same frequency channel at a radio relay, can be used to achieve a spectral efficiency higher than that in the case of half-duplex relaying (HDR) if loopback interference is well-managed. To achieve spectral efficiency that is higher than that achieved when using FDR and HDR separately, an optimal transmission-scheduling scheme for an FDR-HDR hybrid is proposed. The scheme is formulated as an optimization problem. The conditions required to achieve the maximum spectral efficiency are determined analytically. Numerical results confirm that the proposed scheme is superior to FDR and HDR.