In this paper, we consider the secure multi-user downlink communication scheme against eavesdroppers in Integrated Sensing and Communication (ISAC) systems, where the ISAC base station (BS) transmits ISAC waveform to serve communication users and sense targets, and artificial noise (AN) is employed to guarantee the transmission secrecy. In particular, we study the robust ISAC beamforming design under the assumption of imperfect legitimate users’ and eavesdroppers’ channel state information (CSI). Specifically, we aim to minimize the squared error between designed and desired beampatterns under the probabilistic outage constraints concerning the secrecy performance as well as power budget constraint. To deal with the non-convex problem, we utilize the semidefinite relaxation (SDR) technique and Bernstein-type inequality to convert the problem to a tractable form. Numerical results show that the proposed robust beamforming design algorithm is able to approach the performance under perfect CSI, and illustrates the trade-off between sensing and secrecy performances.

The rapid development of Internet of Things gives rise to the emergence of delay-sensitive and computation-intensive applications. Due to the inherent long delay of cloud computing and the limited resources at end devices, mobile edge computing is considered as a promising approach to meet the stringent delay requirement of such demanding applications. To handle the massive connection of Internet of Things, 5G network is shifting toward heterogenous architecture where each end device can access more than one edge server (e.g., base stations and access points). In the presence of multiple edge servers, this paper investigates the interesting problem of how to exploit heterogenous computation resources at network edge to achieve the best energy efficiency among multiple end devices while satisfying their delay requirements. We study a computation offloading management problem by jointly considering heterogeneous computation resources, latency requirements, power consumption at end devices, and channel states. The formulated energy minimization problem falls into the category of mixed-integer and nonlinear program. To solve it efficiently, we decompose the original problem into two subproblems and propose an iterative solution framework to solve for transmission power allocation strategy and computation offloading scheme. Through simulation results, we show that the proposed solution is competitive when compared with the optimal solution. Moreover, we leverage the optimal solutions to analyze the impact of computation resource distribution on energy consumption and computation offloading decision.

In this letter, we explore the problem of DOA estimation using neural networks for two closely spaced sources. Since the traditional high-resolution techniques based on classical algorithms cannot achieve high-accuracy DOA estimation in the presence of two closely spaced sources, especially at low signal-to-noise ratios (SNR), we propose a novel DOA estimation method based on a cascaded neural network to address this problem. Specifically, this network comprises two parts: the SNR classification network and the DOA estimation network. The latter network contains two estimation subnetworks, which are appropriate for different SNRs by training with noisy data and activated by the output of the SNR classification network. Simulation results demonstrate that the estimation performance of our proposed method achieves much better than that of the existing algorithms under various conditions, especially for the scenes with low SNRs or small snapshot number.

In this paper, we investigate the relay selection problem for two-way simultaneous wireless information and power transfer (SWIPT) enabled cognitive radio networks. The system we consider includes a pair of primary users and multiple secondary user transceivers, where the secondary transmitters are energy-constrained and employ the decode-and-forward scheme to assist the transmission between the primary users. Specifically, in order to improve the spectrum efficiency of the considered system, we design a novel two-way communication protocol for SWIPT-enabled cognitive radio networks. Furthermore, different from the traditional relay selection scheme, under this protocol, we propose two data-driven relay selection methods based on the neural network for the fixed number of relays and the variable number of relays, respectively. The simulation results demonstrate that the proposed methods can achieve better performance than the traditional relay selection as well as the traditional machine learning methods, and approach to the theoretical optimal value in most cases.

The vertical industry member and the Internet of Things (IoT) business are spurring heated discussions on the evolution of future mobile communication. The emerging non-terrestrial network (NTN) plays increasingly important role in addressing service and coverage demand boom, whilst they still envisage technical and performance bottlenecks. Besides, traditional communication paradigm is constrained by Shannon theorem and fails to exploit the semantic features. In this article, we review the classic 6G NTN IoT networks and semantic communication in terms of structure, challenges, and applications. Then we observe that the superiority of semantic communication in low signal-to-noise ratio (SNR) regime is complementary to 6G NTN. We further propose an adaptive semantic-empowered 6G NTN IoT structure supporting communication paradigm switching. Finally, the open issues are discussed.

In this letter, we propose a deep learning-based method for the direction of arrival (DOA) estimation in the low signal-to-noise ratio (SNR) scenario. Specifically, the DOA estimation is modeled as a multi-label classification task, and a novel dual class token Vision Transformer (DCT-ViT) is designed to fit it. Different from the classical ViT architecture with a single class token, the DCT-ViT includes two class tokens which are located at the beginning and end of the latent vector sequence, respectively. This architecture enables enhanced information mining and feature extraction from the array signal data in order to improve the accuracy of DOA estimation. Furthermore, a single DCT-ViT model can accommodate different source numbers by leveraging a training dataset with different numbers of sources. Simulation results illustrate that our proposed method outperforms existing methods in the low SNR scenario, including classical model-based and other deep learning-based methods.