Federated learning (FL) allows UAVs to collaboratively train a globally shared machine learning model while locally preserving their private data. Recently, the FL in edge-aided unmanned aerial vehicle (UAV) networks has drawn an upsurge of research interest due to a bursting increase in heterogeneous data acquired by UAVs and the need to build the global model with privacy; however, a critical issue is how to deal with the non-independent and identically distributed (non-i.i.d.) nature of heterogeneous data while ensuring the convergence of learning. To effectively address this challenging issue, this paper proposes a novel and high-performing FL scheme, namely, the hierarchical FL algorithm, for the edge-aided UAV network, which exploits the edge servers located in base stations as intermediate aggregators with employing commonly shared data. Experiment results demonstrate that the proposed hierarchical FL algorithm outperforms several baseline FL algorithms and exhibits better convergence behavior.

Automatic road extraction from unmanned aerial vehicle (UAV) imagery has been one of the major research topics in the area of remote sensing analysis due to its importance in a wide range of applications such as urban planning, road monitoring, intelligent transportation systems, and automatic road navigation. Thanks to the recent advances in Deep Learning (DL), the tedious manual segmentation of roads can be automated. However, the majority of these models are computationally heavy and, thus, are not suitable for UAV remote-sensing tasks with limited resources. To alleviate this bottleneck, we propose two lightweight models based on depthwise separable convolutions and ConvMixer inception block. Both models take the advantage of computational efficiency of depthwise separable convolutions and multi-scale processing of inception module and combine them in an encoder–decoder architecture of U-Net. Specifically, we substitute standard convolution layers used in U-Net for ConvMixer layers. Furthermore, in order to learn images on different scales, we apply ConvMixer layer into Inception module. Finally, we incorporate pathway networks along the skip connections to minimize the semantic gap between encoder and decoder. In order to validate the performance and effectiveness of the models, we adopt Massachusetts roads dataset. One incarnation of our models is able to beat the U-Net’s performance with 10× fewer parameters, and DeepLabV3’s performance with 12× fewer parameters in terms of mean intersection over union (mIoU) metric. For further validation, we have compared our models against four baselines in total and used additional metrics such as precision (P), recall (R), and F1 score.

In this letter, we consider the device-to-device (D2D) communication in cellular networks, where both the cellular user (CU) and D2D receiver (DR) decode information and harvest the energy. Unlike the conventional D2D communications, we figure out that it is possible for the D2D pair to communicate without degrading the performance of the CU. Based on this observation, two schemes changing time duration of the D2D transmission are investigated for D2D rate maximization subject to maintaining the CU performance. By analytically comparing the performance of the two schemes, we obtain useful insights into the D2D transmission in simultaneous wireless information and power transfer (SWIPT) cellular networks. Furthermore, utilizing this analysis, a closed-form scheme is proposed for fading channels.

This article investigates the problem of preamble detection for the long-range (LoRa) modulation and analyzes its performance. For analysis, the inevitable multiuser interference is reasonably modeled as correlated noise. First, the optimal preamble detector is derived and the best preamble is designed to maximize the detection probability while achieving a target value of the false alarm rate. To reduce the required computational complexity and to gain more insights, the preamble detection problem is further investigated for two important special scenarios with: 1) a large spreading factor (SF) and 2) uncorrelated noise, respectively. The analysis is then extended to the case in the presence of phase offset. Our work reveals that the existing preamble detection methods are strictly suboptimal. In addition, various useful and interesting engineering insights into preamble detection with LoRa modulation are provided from the derived results, and the theoretical performance limit of the preamble detection with LoRa modulation is quantified. Extensive numerical results demonstrate the effectiveness and superiority of the proposed scheme. Particularly, the proposed scheme outperforms the existing schemes by more than 10 dB in terms of signal-to-noise ratio (SNR) at the detection probability of 80% and the target false alarm rate of 10%.

Device-to-device (D2D) communications and simultaneous wireless information and power transfer (SWIPT) technologies are key to Internet of Things (IoT) to achieve massive connectivity and to prolong battery lifetime, respectively. In this article, we propose novel schemes for transmit power adaptation to maximize the average data rate and to minimize the outage probability of D2D communications over fast and slow-fading channels, respectively, where the D2D network coexists with a SWIPT-based IoT cellular network operating with the time-switching protocol. Also, different from the existing works, circuit power consumption for information decoding is taken into account in this article. To solve the formulated nonconvex power adaptation problems, we first derive tractable lower bound and upper bound on the average data rate and outage probability of cellular user (CU), respectively. Then, based on the conservative approximation, power adaptation solutions are determined. Intriguingly, we show that in the proposed schemes, the performance tradeoff between information decoding and energy harvesting at the CU is not degraded despite the reception of the D2D signals (i.e., interfering signals). Furthermore, deriving and analyzing closed-form bounds on the D2D performance, we obtain useful insights into the D2D performance improvement by the proposed power adaptation schemes. Numerical results demonstrate that the proposed schemes outperform the baseline scheme, and verify the obtained insights.

In this letter, we consider the device-to-device (D2D) communication in downlink cellular networks, where the cellular user (CU) both decodes information and harvests the energy. Unlike the conventional D2D communications, we figure out that it is possible for the D2D pair to communicate without degrading the performance of the CU. Based on this observation, D2D transmit power control schemes are proposed for D2D rate maximization subject to maintaining the CU performance. In a decentralized way, the proposed power control schemes are performed at the D2D transmitter (DT) with instantaneous channel state information (CSI) of only the link between the D2D pair. For performance analysis, an asymptotic upper bound is provided on the D2D rates of the proposed schemes. Numerical results confirm the effectiveness of the proposed schemes.