Zhiyong Feng’s research while affiliated with Beijing University of Posts and Telecommunications and other places

With the growing demand for seamless connectivity and intelligent communication, the integration of artificial intelligence (AI) and sixth-generation (6G) communication networks has emerged as a transformative paradigm. By embedding AI capabilities across various network layers, this integration enables optimized resource allocation, improved efficiency, and enhanced system robust performance. This paper presents a comprehensive overview of AI and communication for 6G networks, with a focus on their foundational principles, inherent challenges, and future research opportunities. We first review the integration of AI and communications in the context of 6G, exploring the driving factors behind incorporating AI into wireless communications, as well as the vision for the convergence of AI and 6G. The discourse then transitions to a detailed exposition of the envisioned integration of AI within 6G networks, divided into three progressive stages. The first stage, AI for network, focuses on employing AI to augment network performance, optimize efficiency, and enhance user service experiences. The second stage, network for AI, highlights the role of the network in facilitating and buttressing AI operations and presents key enabling technologies. We compare wireless network large models with conventional large language models (LLMs), and identify key design principles and components for building wireless network architectures. In the final stage, AI as a service, it is anticipated that future 6G networks will innately provide AI functions as services, supporting application scenarios like immersive communication and intelligent industrial robots. Specifically, we define the quality of AI service, which refers to a framework for measuring AI services within the network. We further summarize the standardization process of AI for wireless networks, highlighting key milestones and ongoing efforts. In addition, we analyze the critical challenges faced by the integration of AI and communications in 6G. Finally, we outline promising future research opportunities that are expected to drive the development and refinement of AI and 6G communications.

Integrated sensing and communication (ISAC) plays a crucial role in the Internet of Vehicles (IoV), serving as a key factor in enhancing driving safety and traffic efficiency. To address the security challenges of the confidential information transmission caused by the inherent openness nature of wireless medium, different from current physical layer security (PLS) methods, which depends on the \emph{additional communication interference} costing extra power resources, in this paper, we investigate a novel PLS solution, under which the \emph{inherent radar sensing interference} of the vehicles is utilized to secure wireless communications. To measure the performance of PLS methods in ISAC-based IoV systems, we first define an improved security performance metric called by transmission reliability and sensing accuracy based secrecy rate (TRSA\_SR), and derive closed-form expressions of connection outage probability (COP), secrecy outage probability (SOP), success ranging probability (SRP) for evaluating transmission reliability, security and sensing accuracy, respectively. Furthermore, we formulate an optimization problem to maximize the TRSA\_SR by utilizing radar sensing interference and joint design of the communication duration, transmission power and straight trajectory of the legitimate transmitter. Finally, the non-convex feature of formulated problem is solved through the problem decomposition and alternating optimization. Simulations indicate that compared with traditional PLS methods obtaining a non-positive STC, the proposed method achieves a secrecy rate of 3.92bps/Hz for different settings of noise power.

A near-field motion parameter estimation method is proposed. In contract to far-field sensing systems, the near-field sensing system leverages spherical-wave characteristics to enable full-vector location and velocity estimation. Despite promising advantages, the near-field sensing system faces a significant challenge, where location and velocity parameters are intricately coupled within the signal. To address this challenge, a novel subarray-based variational message passing (VMP) method is proposed for near-field joint location and velocity estimation. First, a factor graph representation is introduced, employing subarray-level directional and Doppler parameters as intermediate variables to decouple the complex location-velocity dependencies. Based on this, the variational Bayesian inference is employed to obtain closed-form posterior distributions of subarray-level parameters. Subsequently, the message passing technique is employed, enabling tractable computation of location and velocity marginal distributions. Two implementation strategies are proposed: 1) System-level fusion that aggregates all subarray posteriors for centralized estimation, or 2) Subarray-level fusion where locally processed estimates from subarrays are fused through Guassian product rule. Cram\'er-Rao bounds for location and velocity estimation are derived, providing theoretical performance limits. Numerical results demonstrate that the proposed VMP method outperforms existing approaches while achieving a magnitude lower complexity. Specifically, the proposed VMP method achieves centimeter-level location accuracy and sub-m/s velocity accuracy. It also demonstrates robust performance for high-mobility targets, making the proposed VMP method suitable for real-time near-field sensing and communication applications.

In the evolution towards the forthcoming era of sixth-generation (6G) mobile communication systems characterized by ubiquitous intelligence, integrated sensing and communication (ISAC) is in a phase of burgeoning development. However, the capabilities of communication and sensing within single frequency band fall short of meeting the escalating demands. To this end, this paper introduces a carrier aggregation (CA)-enabled multi-input multi-output orthogonal frequency division multiplexing (MIMO-OFDM) ISAC system fusing the sensing data on high and low-frequency bands by symbol-level fusion for ultimate communication experience and high-accuracy sensing. The challenges in sensing signal processing introduced by CA include the initial phase misalignment of the echo signals on high and low-frequency bands due to attenuation and radar cross section, and the fusion of the sensing data on high and low-frequency bands with different physical-layer parameters. To this end, the sensing signal processing is decomposed into two stages. In the first stage, the problem of initial phase misalignment of the echo signals on high and low-frequency bands is solved by the angle compensation, spatial filtering and cyclic cross-correlation operations. In the second stage, this paper realizes symbol-level fusion of the sensing data on high and low-frequency bands through sensing vector rearrangement and cyclic prefix adjustment operations, thereby obtaining high-precision sensing performance. Then, the closed-form communication mutual information (MI) and sensing Cramér-Rao lower bound (CRLB) for the proposed ISAC system are derived to explore the theoretical performance bound with CA. Simulation results validate the feasibility and superiority of the proposed ISAC system.

Integrated sensing and communication (ISAC) is expected to become a crucial component of the sixth-generation (6 G) networks owing to its outstanding spectrum management capability. However, improving the cooperative sensing capabilities of multiple ISAC user equipments (ISAC-UEs) in complex interference environment presents a significant research challenge. This paper focuses on the multi-user cooperative sensing in uplink orthogonal frequency division multiplexing (OFDM) ISAC system. By utilizing the stochastic geometry, we model the distribution of communication UEs (COM-UEs) as a one-dimensional Matern hard-core point process (1-D MHCP), and derive a closed-form expression for interference power. To further enhance cooperative sensing accuracy while maintaining quality of service (QoS) in communication, we perform waveform optimization by jointly optimizing the weighted range-velocity Cramer–Rao lower bound (CRLB) subject to communication data rate (CDR) and subcarrier power ratio (SPR) constraints. This approach involves selecting the optimal subcarriers for sensing and allocating the corresponding power on each subcarrier for communication and sensing subsystems. By employing the convex relaxation and the cyclic minimization algorithm (CMA), we decompose the complex optimization problem into three sub-problems, simplifying the original NP-hard problem into a solvable one via a cyclic optimization framework. The simulation results validate the effectiveness of our optimization strategy, and evaluate the influence of CDR and SPR constraints using the CRLB and root mean square error (RMSE).

Inverse synthetic aperture radar (ISAR) sensing is poised to become a key technology in integrated sensing and communication (ISAC) systems due to its high cross-range resolution for extended targets. However, existing ISAR-based orthogonal frequency-division multiplexing (OFDM) ISAC works are limited and primarily rely on discrete Fourier transform (DFT) techniques, which suffer from inherent resolution constraints, making them inadequate for resolving closely spaced scatterers. We investigate ISAR sensing within OFDM ISAC systems from a novel perspective. First, we model and derive the frequency domain reconstruction of ISAR OFDM received signals, proving that the main parameters range and cross-range of scatterers are decoupled. Building on this insight, we extend and improve two one-dimensional multiple signal classification (1D-MUSIC) algorithms to independently estimate them. Specifically: 1) We consider all possible scatterer relationships and comprehensively classify scatterers into uncorrelated and correlated by providing clear definition; 2) Correlated scatterers cause both range and cross-range steering matrices to be rank-deficient and both affect the rank of autocorrelation matrices. Therefore, we propose a spatial smoothing range estimation (SSR) method and a spatial smoothing cross-range estimation (SSC) method to restore the rank of steering matrixes; 3) 1D estimation cannot effectively associate the estimated range and cross-range values, and correlated scatterers cause the number of spectral peaks obtained by SSR and SSC methods to be less than the actual number of scatterers. Therefore, we develop a novel point-matching method to pair the decoupled range and cross-range values for each scatterer, while also accurately estimating the total number of scatterers. Finally, the numerical results demonstrate that: 1) The range and cross-range spatial spectrum peaks achieved by the proposed SSR and SSC methods are 10dB higher than those obtained with the conventional 1D-MUSIC; 2) The point-matching method exhibits performance comparable to the 2D-DFT concerning normalized spectrum and coordinates reconstruction; 3) In terms of NMSE, the proposed algorithms significantly outperform the traditional algorithm, achieving NMSE values that are 3 to 4 orders of magnitude lower, as well as maintain robust performance as the number of scatterers increases.

The urgent need for higher sensing accuracy and transmission reliability in the Internet of Vehicles (IoV), along with the limited availability of spectrum resources, integrated sensing and communication (ISAC) tends to operate in higher frequency bands. This shift, however, introduces severe sensing-communication coupling interference. Due to the openness nature of the wireless medium, ISAC enabled IoV faces more significant security concerns. Centered at the communication interference, traditional physical layer security (PLS) schemes aim to maximize the channel quality difference between legitimate and eavesdropping channels to ensure the perfect secrecy. However, transforming directly current PLS methods into secure communication in ISAC-assisted IoV systems faces two significant challenges, because of sensing-communication coupling interference and strong directional beamforming caused by higher frequency bands. To address these challenges, this paper formulates an optimization problem aimed at maximizing the average secrecy rate by jointly designing the transmission power allocation and radar sensing power allocation, while ensuring the desired sensing accuracy and transmission reliability. To solve this non-convex optimization problem, we introduce the block coordinate descent (BCD) and successive convex approximation (SCA) methods. Experimental results demonstrate that, compared to optimizing transmission power or radar sensing power individually, joint optimization significantly improves the average secrecy rate upon convergence, with increases of at least 265% and 441%, respectively.

Automatic modulation classification (AMC) is paramount within the Artificial Intelligence of Things (AIoT) realm for consumer electronics, offering advantages such as efficient spectrum utilization, heightened communication reliability and security, and an enhanced user experience. Addressing the challenges posed by variable signal-to-noise ratio (SNR) conditions, this paper introduces SEMIN (SNR-Enhanced Modulation Insight Network), a novel deep learning architecture aimed at significantly improving classification accuracy, particularly in high SNR scenarios. By integrating SNR-aware training and a unique combination of cross-entropy and center loss functions, SEMIN adeptly balances spatial and temporal feature extraction through convolutional neural networks (CNNs) and bidirectional gated recurrent units (BiGRUs). Comprehensive evaluations showcase the superior performance of the proposed SEMIN model, achieving an accuracy rate above 93% in high SNR conditions and surpassing existing methods. This outcome not only underscores the effectiveness of the proposed SEMIN model in modulation classification but also establishes a new benchmark for future research and application in relevant fields.

Exploring the mutual benefit and reciprocity of sensing and communication (S&C) functions is fundamental to realizing deeper integration for integrated sensing and communication (ISAC) systems. This paper investigates a novel communication-assisted sensing (CAS) system within 6G perceptive networks, where the base station actively senses the targets through device-free wireless sensing and simultaneously transmits the estimated information to end-users. In such a CAS system, we first establish an optimal waveform design framework based on the rate-distortion (RD) and source-channel separation (SCT) theorems. After analyzing the relationships between the sensing distortion, coding rate, and communication channel capacity, we propose two distinct waveform design strategies in the scenario of target impulse response estimation. In the separated S&C waveforms scheme, we equivalently transform the original problem into a power allocation problem and develop a low-complexity one-dimensional search algorithm, shedding light on a notable power allocation tradeoff between the S&C waveform. In the dual-functional waveform scheme, we conceive a heuristic mutual information optimization algorithm for the general case, alongside a modified gradient projection algorithm tailored for the scenarios with independent sensing sub-channels. Additionally, we identify the presence of both subspace tradeoff and water-filling tradeoff in this scheme. Finally, we validate the effectiveness of the proposed algorithms through numerical simulations.

In a time division broadcast localization and synchronization (TDBLAS) system, moving user nodes (UNs) with the clock offset and the clock drift usually use the sequential one-way time-of-arrival (TOA) and Doppler shift measurements from the anchor nodes (ANs) to resolve the joint localization and synchronization (JLAS) problem in the presence of the AN position error. The objective function of the maximum likelihood (ML) method to solve the above JLAS problem in a TDBLAS system is high-dimensional nonlinear and nonconvex. Thus, the existing iterative method for solving the above ML estimation problem usually encounters issues like local minima or non-convergence when an accurate initial guess is missing. In this paper, we propose a new semidefinite programming (SDP) method to address this issue, which can guarantee the global optimal solution without requiring initialization. We introduce the optimization variable and formulate a nonconvex constrained weighted least square (CWLS) minimization problem. Subsequently, we propose a novel semidefinite relaxation (SDR) approach to relax the complex and nonconvex CWLS problem into a convex and tractable SDP problem. Through theoretical estimation error analysis, we demonstrate that the CWLS solution can achieve the Cramér-Rao lower bound (CRLB) under small Gaussian noise. Simulation results in a 2D scenario show that the proposed SDP method reaches the CRLB under small Gaussian noise. Compared to the conventional iterative method, the proposed SDP method exhibits greater robustness, achieving the global optima without requiring initialization under small Gaussian noise.