Han Hu

Under the background of new engineering, the education department has put forward new demands for the teaching of engineering courses. Taking the core course in electronic information, “Principles of Communication,” as the entry point, this study analyzes the existing problems in current teaching practice and, in combination with students' learning characteristics, has comprehensively explored various aspects such as teaching philosophy, teaching methods, and assessment methods, innovatively constructing a dual-engine-driven teaching model powered by knowledge graphs and large-language models. Specifically, the knowledge graph systematically combs the knowledge system of communication principles and builds a disciplinary knowledge network; the large-language model realizes the dynamic adaptation of teaching resources through intelligent Q&A systems, personalized learning path recommendations, and virtual simulation dialogues. The study adheres to the student-centered approach and, through the “theory practice ideological and political education” three in one teaching reform path, aims to improve the course's teaching quality as well as students' comprehensive quality and innovation ability. Actual teaching practice has shown that this dualdriven teaching model effectively realizes the systematization of knowledge construction, the personalization of ability cultivation, and the subtle shaping of values, providing a promotable paradigm for the reform of engineering courses in the new era.

Thanks to low path-loss and high macro-diversity gain, the cell-free (CF) massive multiple-input multiple-output (MIMO) is regarded as a perspective technique to realize ultra-reliable and low-latency communication (URLLC) for future wireless systems. In this paper, we consider the emerging reconfigurable intelligent surface (RIS) to further improve the capability of CF massive MIMO system in URLLC provision. Under the finite channel blocklength regime, we formulate the sum spectral efficiency (SE) maximization problem to jointly design the active beamforming of access points (APs) and the passive beamforming of RISs. To resolve this non-convex problem, we establish a tight logarithmically lower bound based on the majorization minimization criterion to replace the complicated objective function. Then, we propose an alternating optimization algorithm to solve two subproblems and derive the closed-form solutions of both active and passive beamformings. Taking into account the extra power consumption of RISs, we utilize the novel performance metric named resource efficiency (RE) to investigate the trade-off between SE and energy efficiency (EE) in RIS-aided CF massive MIMO URLLC systems. Moreover, a variant bisection search algorithm is developed to maximize the system RE. Numerical results demonstrate the superiority of the proposed lower bound for optimizing the RIS-aided CF massive MIMO URLLC system. The proposed algorithms are also effective in improving the sum-SE and striking the SE-EE balance of RIS-aided CF massive MIMO URLLC systems.

In the time-division-duplexing (TDD)-based cell-free (CF) massive multiple-input multiple-output (MIMO) system, the channel reciprocity needs to be recovered via a reciprocity calibration operation due to the random circuit impact on the transceiver radio frequency. In this paper, we study the effect of the calibration error on the TDD CF massive MIMO system under non-ideal calibrations. Assuming the spatially correlated Ricean fading channel, we derive the closed-form expression of the downlink achievable rate, which takes both the channel estimation error and the calibration error into account. Some novel insights of the calibration error in the CF massive MIMO system are gathered from the analytical results. It is shown that the downlink achievable rate is more sensitive to the calibration error at the user side. In order to provide a uniformly good service for each user, we employ the geometric programming (GP) to solve the max-min power optimization problem to maximize the minimum user rate. Additionally, we utilize the scaled alternating direction method of multipliers to develop a calibration error-aware beamforming scheme to mitigate the impact of calibration errors, improving the downlink sum-rate. Numerical results demonstrate that the proposed GP-based algorithm significantly improves the 95%-likely per-user downlink achievable rate with a fast convergence behavior. Moreover, the proposed calibration error-aware beamforming scheme enhances the downlink sum-rate and outperforms other benchmark schemes.

When millimeter wave (mmWave) frequencies and reconfigurable intelligent surfaces (RISs) are used in cell-free (CF) massive multiple-input multiple-output (MIMO) systems to enhance the data rate, the energy efficiency (EE) may decrease due to the high power consumption of more antennas and RIS elements. In this paper, we aim to design a hybrid beamforming scheme to strike the balance between the spectral efficiency (SE) and EE of the RIS-aided mmWave CF massive MIMO system. To achieve this goal, we formulate the resource efficiency (RE) maximization problem to jointly design the analog and digital beamformings at access points (APs) and the reflect beamforming at RISs. Since the original problem is non-convex, we propose a golden-search-based alternating optimization algorithm to solve it in an iterative manner. The closed-form solutions of three beamformings are also derived. Numerical results demonstrate that the proposed hybrid beamforming scheme is effective in realizing the trade-off among the SE and EE in RIS-aided mmWave CF massive MIMO systems. Moreover, the proposed algorithm exhibits the fast convergence behavior.

Semantic communication is envisioned as a potential communication paradigm enabled by artificial intelligence and is promising to break the Shannon limit for future 6G networks. This paradigm benefits unmanned aerial vehicles (UAVs) to conserve communication resources and minimize latency by only transmitting task-relevant semantic information. However, resource allocation in the multiple collaborative UAV scenarios remains unexplored, particularly regarding multi-modal semantic communication. To tackle this challenge, this paper investigates a semantic-aware intelligent resource allocation method for multi-UAV-assisted semantic communication networks in the UAV image-sensing task-oriented scenario. A multi-modal semantic communication framework with multi-UAV relay collaboration is developed. At the semantic level, a novel quality of experience (QoE) and the transmission cost model are introduced, based on which a semantic-aware resource allocation problem is formulated, aiming to maximize QoE while minimizing the transmission cost by jointly optimizing the UAV trajectory, the spectrum bandwidth, the transmit power and the number of the transmitted semantic symbols. To deal with optimization challenges involving hybrid variables and coordination among UAVs, a multi-UAV hybrid decision-controlled deep reinforcement learning (DRL) scheme is proposed. Simulation results demonstrate the effectiveness of the proposed scheme compared with the benchmark schemes in achieving a good balance between the QoE and the transmission cost.

Thanks to the customization of channel propagation, reconfigurable intelligent surface (RIS)-aided cell-free (CF) massive multiple-input-multiple-output (MIMO) is recognized as a competitive candidate technique for the future communication system. However, only the limited performance gain can be afforded by passive RIS due to the double-fading effect in RIS-aided links. In this article, we consider the CF massive MIMO system with the assistance of the active RIS, which is capable of reflecting and amplifying the incident signal, to enable the Internet of Things network. We analyze the tradeoff between the number of active RIS reflecting elements (REs) and the amplification coefficient. Considering the power constraint at the active RIS, we formulate a sum-rate maximization problem to jointly optimize the user transmission power, the receive beamforming, and the RIS reflecting precoding. Since the original problem is nonconvex, we decouple it into three subproblems and then design an alternating optimization algorithm to solve them iteratively. Using the Lagrangian dual reformulation and generalized Rayleigh quotient theory, we derive the closed-form solutions for both the user transmission power and the uplink receive beamforming. We also develop a low-complexity method to acquire the RIS reflecting precoding based on the primal-dual subgradient theory. Compared to the passive RIS, the active RIS can significantly improve the system performance with fewer REs. Moreover, the proposed optimization scheme effectively mitigates the drawbacks of the active RIS under the high-transmission power regime and enhance its benefits. Finally, the proposed alternating optimization algorithm is validated by numerical results.

Semantic communications have been envisioned as a potential technique that goes beyond Shannon paradigm. Unlike modern communications that provide bit-level security, the eavesdropping of semantic communications poses a significant risk of potentially exposing intention of legitimate user. To address this challenge, a novel deep neural network (DNN) enabled secure semantic communication (DeepSSC) system is developed by capitalizing on physical layer security. To balance the tradeoff between security and reliability, a two-phase training method for DNNs is devised. Particularly, Phase I aims at semantic recovery of legitimate user, while Phase II attempts to minimize the leakage of semantic information to eavesdroppers. The loss functions of DeepSSC in Phases I and II are respectively designed according to Shannon capacity and secure channel capacity, which are approximated with variational inference. Moreover, we define the metric of secure bilingual evaluation understudy (S-BLEU) to assess the security of semantic communications. Finally, simulation results demonstrate that DeepSSC achieves a significant boost to semantic security particularly in high signal-to-noise ratio regime, despite a minor degradation of reliability.

In the context of Internet-of-Things, this paper considers a cell-free (CF) massive multiple-input multiple-output (MIMO) system for ultra-reliability and low latency communication (URLLC) assisted by multiple reconfigurable intelligent surfaces (RIS). We derive the closed-form expression of the downlink achievable rate under multiple correlated RISs and pilot contamination. To mitigate the impact of pilot contamination and improve the fairness among users, we minimize the maximum normalized mean-squared error (NMSE) of the channel estimation by jointly optimizing the pilot power coefficient and the RIS phase shifts. Due to the non-convexity of the original problem, we design an alternating optimization algorithm to solve the substitutable two sub-problems using fractional programming and sequential convex approximation. Numerical results validate the proposed algorithm in terms of decreasing the maximum user NMSE and converging. Moreover, the 95%-likely per-user downlink achievable rate is also improved.