Grammar-based fuzzing is a practical and effective approach for uncovering vulnerabilities in interpreters that directly manipulate memory when parsing highly structured inputs. These fuzzers aim to generate well-formed, bug-triggering test cases using either formal grammars or ML-based techniques. To trace the development of these methods, this paper surveys 45 representative papers from top-tier conferences and Q1 journals (2007–2024), manually selected via a snowballing literature review. The survey balances foundational work and recent innovations, highlights key insights, and identifies existing solutions and gaps. It also outlines future challenges and opportunities, laying a solid foundation for continued progress in grammar-based fuzzing.

Zero-knowledge proofs(ZKPs) are powerful tools that have performed an important role in the development of secure cryptographic protocols for decades. Recently, there's been a spike of interest in them due to recent breakthroughs that have made them a viable option in the real world. These proofs have contributed to the evolution of cryptography and blockchain, offering unique properties such as succinctness, soundness, and the capability to verify the truth of a claim without spilling any additional details beyond confirming the claim's truthfulness. This paper surveys the main development regarding zero-knowledge proofs, starting from the basic principles and reaching the most recent and sophisticated research progress achieved by zero-knowledge proof(ZKP) system. We categorize zero-knowledge proofs into distinct types, each offering different trade-offs and benefits. This work also points out applications in various fields, such as blockchain, authentication systems, electronic voting and other systems.

Joint estimation of target range and direction of arrival (DoA) considers both range and direction information of targets at the parameter modeling level. By introducing mutual information (MI), it provides a more comprehensive description of the information structure in radar observations, quantitatively characterizing the sensing capability of radar systems from a statistical perspective and offering theoretical guidance for system design. Despite its significance, analytical investigations into closed-form approximations for jointly estimating multiple parameters in radar systems remain scarce. In this study, a Shannon information theory-based approach is employed to obtain a closed-form expression of MI associated with joint range and DoA estimation in multiple-input multiple-output (MIMO) radar systems. Utilizing the joint posterior probability density function (PDF), a closed-form approximation of the upper bound of MI is further derived, along with an information-theoretic performance metric—the entropy error (EE). Simulation results are presented to compare the MI and its upper bound, as well as EE and its lower bound, demonstrating the validity and accuracy of the proposed formulations.

The explosive increase in bandwidth-consuming services puts forward higher requirements on system capacity. Bandwidth limitation is one of the important obstacles to achieve higher capacity. It is a research hotspot to realize higher transmission rate under bandwidth limitation. Poly-binary modulation can be used to achieve a higher symbol rate by spectral shaping. In this paper, we proposed the modified poly-binary modulation scheme without encoding. The proposed scheme and different modulation schemes are compared in detail. The simulation results show that the achievable maximum bit rate of pre-coded (1+D)-four level pulse amplitude modulation (PAM-4) is over 110/200~Gb/s with 10/20~GHz bandwidth constraints. High-order poly-binary modulation can achieve higher transmission rates. The modified poly-binary modulation scheme without encoding shows great advantage in high-speed system.

Traditional superimposed pilot (SP) schemes suffer from high peak-to-average power ratio (PAPR) and struggle to estimate fractional Doppler shifts, which hurts the communication performance. To overcome these problems, we develop a low-PAPR sparse Bayesian learning channel estimation scheme for fractional Doppler (SBLCE-FD), utilizing a novel hybrid SP (HSP) structure. First, the developed SBLCE-FD scheme leverages the novel HSP structure to estimate integer delay/Doppler taps, strategically enhancing the sparsity of the pilot dictionary matrix to reduce computational overhead. Subsequently, leveraging the sparsity-enhanced dictionary matrix, the SBLCE-FD scheme localizes the range of the potential non-zeros elements in sparse channel gain vector. This process suppresses uncorrelated channel components, thereby mitigating cross-path interference. Furthermore, the minimum mean square error (MMSE)-based detection mitigates residual errors during data recovery, with iterative refinement progressively improving channel gain accuracy. Our analysis confirms the low-PAPR advantage of the HSP structure in the delay-Doppler domain. Simulations demonstrate the SBLCE-FD scheme with HSP structure outperforms conventional SBL and orthogonal matching pursuit (OMP) methods with other SP structures in PAPR, normalized mean squared error (NMSE), and bit error rate (BER).

Recently Reed-Muller codes based on iterative projection aggregation (IPA) decoding method have emerged as a promising candidate to meet the stringent requirements of ultra-reliable low-latency communications (URLLC) in next-generation wireless systems. However, the conventional IPA decoding algorithm still suffers from a high computational complexity, which severely hinders its application in time-and-resource-constrained scenarios. To address this challenge, we propose a novel Fast-IPA decoding algorithm that incorporates two key innovations: a hardware-friendly projection pruning strategy and a lightweight early termination criterion, both of which substantially reduce the computational burden of traditional IPA decoding. Furthermore, we develop a highly efficient and flexible pipelined architecture tailored for the proposed Fast-IPA algorithm. Synthesis results under TSMC 28~nm technology demonstrate that the proposed decoder can achieve remarkable performance, including up to 43.9$\times$ enhancement in area efficiency and 92.0% reduction in latency, while maintaining comparable error-correcting performance compared to the state-of-the-art IPA decoder.

Achieving high hardware efficiency in low-density parity-check (LDPC) decoders for 5G new radio (NR) standards remains challenging due to the complexity of logic and routing. This paper proposes a low-complexity stochastic decoder for 5G LDPC codes that achieves both high hardware efficiency and superior error-correction performance. We introduce a novel variable node (VN) update method utilizing comprehensive and inverse operations to minimize structural redundancy in conventional variable node units (VNUs) and improve the convergence accuracy of high-degree VNs. Additionally, we propose enhanced post-processing strategies to mitigate error floors, particularly in codes characterized by a large proportion of degree-one VNs. Simulation results show that the performance of the proposed stochastic decoder is comparable to that of the normalized min-sum algorithm (NMSA). The proposed decoder is implemented in a 90 nm CMOS process for the 5G LDPC code with a length of 3808 and a rate of 0.3. The design occupies 3.84 $\text{mm}^2$ of silicon area, uses 1.36M logic gates, and achieves a throughput of 8.14 Gbps at a bit error rate (BER) of $10^{-6}$ under 750 MHz. With an area efficiency of 2.12 $\text{Gbps/mm}^2$, our decoder outperforms current state-of-the-art implementations.

With the adoption of multi-band transmission in optical networks, fully upgrading all network links can lead to underutilization of spectrum resources on certain links due to constraints such as spectrum continuity. To address this issue, selectively upgrading only a subset of links to support multi-band transmission has emerged as a cost-effective alternative. However, whether such a partially link-upgraded approach can achieve comparable performance to the network with fully band-upgraded links remains an open and urgent question. For this, we propose the systematic analysis of partially link-upgraded multi-band networks. We first introduce different link upgrade methods to determine which links should be upgraded and propose a novel inter-band wavelength conversion strategy designed to optimize signal transmission across partially band-upgraded links. To evaluate and compare the performance of this strategy with alternative approaches including fully link-upgraded networks, we introduce a generalized signal-to-noise ratio (GSNR) estimation model under wavelength conversion and a network cost model. We then propose heuristic algorithms for different wavelength conversion strategies for lightpath establishment in the multi-band optical networks. Simulation results show that, the stage-based proportional link upgrade method enables rapid capacity expansion while effectively controlling the upgrade cost. In the partially link-upgraded network, the proposed inter-band wavelength conversion strategies can not only significantly enhance the network capacity, but also significantly reduce the network upgrade cost, thus achieving an effective trade-off between capacity enhancement and cost reduction. Moreover, according to the link upgrade proportions, the partial-band upgrades are particularly suitable for large and sparse networks.

Vortex beams have gained significant attention in underwater optical wireless communication (UOWC) systems due to their unique advantages, including the ability to carry orbital angular momentum, which enables efficient spatial multiplexing for high-capacity data transmission. Their resilience to oceanic turbulence further enhances the stability of UOWC systems. Integrating fiber optic devices with optical vortex beams is crucial in this context. This paper focuses on analyzing the fiber coupling efficiency (CE) of optical vortex beams under the effects of oceanic turbulence. In the study, the extended Huygens-Fresnel principle is utilized to analyze the propagation of optical vortex waves. The necessary optical parameters for calculating fiber CE, including the optical power captured by the receiver lens and the optical power entering the fiber core, are then analytically evaluated.

A highly efficient Doherty power amplifier (DPA) with a new band-pass power divider for better performance in wide frequency ranges has been introduced. The proposed DPA is a combination of a Class-F carrier amplifier utilizing Class-AB bias and a Class-F peaking amplifier with Class-C bias to allow for amplification with the highest efficiency, at levels of low-power and high-power. A noteworthy feature of the design is the bandpass power divider designed with interdigital structures and asymmetrical low-impedance lines. It removes DC-block capacitors, lowers insertion loss, and improves harmonic suppression. For the class-F power amplifiers, a new harmonic control circuit (HCC) has been introduced, which is designed based on the transistor's output capacitor and in the resonance conditions. The DPA was fabricated, with a maximum drain efficiency (DE) more than 69.26%, a 10.23 dB power gain and a $P_{\mathrm{out}}$ of 29.23 dBm. These advances in power efficiency and bandwidth make the DPA highly conducive to sub-6 GHz 5G NR Band n41 (2.496—2.690 GHz) and LTE Bands 7 and 41. High PAPR signals above 69% drain efficiency and 500 MHz bandwidth are supported by the DPA's design, which is in line with the requirements of these standards.

Low earth orbit (LEO) constellation composed of hundreds to thousands of satellites plays a significant role in providing global low-latency and high-bandwidth internet services especially for the remote areas. However, considering the current launching capability of rocket, i.e., tens of satellites per launch, LEO constellation may consume several years or more to complete its construction. Before the constellation is constructed completely, partially constructed constellation suffers from a long-term weak connectivity, which is hardly to provide real-time end-to-end services. Fortunately, partially constructed constellation may provide non-real-time delay-tolerant services (DTSs, e.g., some earth observation and monitoring applications) through employing disruption tolerant network protocols, which could avoid the waste of satellite resources and earn early revenue to accelerate the subsequent construction of LEO constellation. To assure the service capability of DTSs in the progressive construction of constellation, we first propose a novel satellite deployment strategy with maximizing number of potential contacts (MNPC) under deployment cost constraints. Then, after some satellites are deployed successfully, a time-correlated contact plan (TCCP) scheme under time and resource constraints is proposed to maximize number of temporal transmission paths that across multiple time slots. Simulation results show that MNPC and TCCP schemes effectively optimize the delivery success ratio and delay of DTSs compared with the previous works.

Edge caching (EC) has become a promising technology that minimizes last-hop traffic by caching popular content on edge servers (ESs). Given practical resource limitation, each ES only caches a portion of the contents and cannot satisfy all user requests. How to efficiently cache the contents and route the requests is a challenging problem in EC networks. First, switching cost incurred by the caching decision is temporal-correlated. Second, since contents and requests are dynamic, each ES should continue adjusting its contents. Third, it is desirable to have an online solution with provable performance guarantee and polynomial running time. Existing efforts are insufficient to address all the above challenges. This work models the content caching and request routing of multi-ESs as a joint optimization problem, to minimize the overall system cost including storage, service and switching. We design a new online framework $JoCR$ by taking advantage of online convex optimization, predictive control, and randomized rounding methods. Theoretically, $JoCR$ can realize an excellent competitive ratio compared with the offline optimum. Practically, extensive simulations verify the efficiency of $JoCR$ over multiple benchmarks in reducing operational cost.

To address the conflict between distribution network expansion and traditional communication constraints, dual-mode communication technology that integrates high-speed power line carrier and radio frequency communication is increasingly being adopted. Efficient network analysis is essential for optimizing deployment and improving communication efficiency, scalability, and system compatibility. However, complex factors such as traffic characteristics, transmission modes, and interference avoidance policies hinder accurate evaluation of the end-to-end (E2E) delay violation probability, posing challenges for performance assessment during deployment. To tackle these challenges, this paper proposes an E2E delay analysis method based on the martingale framework. The method first constructs and analyzes delay models for different traffic types under specified network scale and node state conditions. It then transforms node states into node service capabilities, and maps the service process from the time domain to the martingale domain for analytical evaluation. In addition, we analyze the impact of non-service link traffic on the performance of service links, and derive the E2E delay violation probability based on a serial service descriptor constructed under the martingale framework. Simulation results confirm the correctness of the proposed method and demonstrate its ability to achieve a tighter delay bound.

Anonymous communication network I2P leverages multi-layer encryption and multi-hop routing to effectively conceal user identities. However, conducting performance and security evaluations in real-world environments often faces challenges such as ethical concerns, lack of reproducibility, and complex operational requirements. To address these limitations, simulation-based test platforms have emerged as a viable alternative. Nevertheless, accurately replicating the complex and dynamic structure of the I2P network at a reduced scale remains a significant challenge. In this paper, we proposes a passive local topology modeling method for the I2P network, based on the data of real-world network database (netDb) from controlled I2P nodes. Our method employs a breadth-first edge traversal algorithm combined with a hash-based filtering mechanism to reconstruct a representative local network topology. While preserving key network characteristics, the model reduces overall network scale according to node performance metrics. The resulting model is instantiated within a network testbed built on the P-I2Prange framework. Experimental results demonstrate that the modeled network successfully reproduces several critical properties of real I2P, including netDb distribution, tunnel performance metrics (such as Time to First Byte and Time to Last Byte), and router traffic loads. Across different scales, more than 90% of router throughput errors remain below 15%. This research presents a scalable, reproducible, and ethically sound modeling framework for I2P, supporting realistic experimental studies in anonymous network research.

Censorship circumvention technology is an indispensable part of covert communication. However, many existing censorship circumvention tools rely on Web real-time communications (WebRTC), lacking mechanisms to facilitate client-proxy circumvention channel establishment. We innovatively created EthRTC, a censorship circumvention system, utilizing blockchain as a communication channel during connection to facilitate client-broker connections. We utilize the Ethereum network as a direct medium for exchanging bootstrap information, which can be detected in the transaction pool without requiring the transaction to be included in the Ethereum consensus. This introduces an acceptable time cost during the client's bootstrap phase. During the bypass traffic transmission phase, we utilize WebRTC's DataTunnel to transfer bypass traffic, allowing users to provide proxy services for other users and making efficient use of client-side resources. We demonstrated EthRTC's effective censorship circumvention for users in censored regions in the real world. We also compared the delay between the bootstrapping phase and accessing websites during the circumvention traffic transmission with Tor's anti-censorship plugin Snowflake. EthRTC's average bootstrapping delay is around 9 seconds higher than Snowflake's 22.14 seconds. Accessing websites has nearly identical latency, showing EthRTC's acceptable introduced delay and costs.

6G demands networks that can reason, explain, and adapt in heterogeneous, fast-changing environments. Existing paradigms—rule-driven automation, template-bound IBN, and black-box KDN—struggle with scalable reasoning, interpretability, and cross-domain generalization. This paper presents knowledge-driven autonomous network (KAN), a 6G architecture that elevates cognition to a first-class capability by introducing a dedicated cognitive plane to close the loop across semantic perception, hybrid inference (forward, backward, abductive), policy synthesis, and continual learning atop a unified knowledge base. In reproducible ns-3 evaluations spanning fault recovery, cross-domain generalization, and UAV-swarm optimization, KAN consistently outperforms baselines: recovery time 2.1 s (vs. SDN 8.3 s, IBN 5.0 s, KDN-DQN 3.8 s), strategy accuracy 94.1%, and an explainable reasoning depth of $8.7$. Under domain shifts, KAN maintains an average normalized score of $0.879 \pm 0.037$ with only 10.5% degradation from the Urban baseline. In 100-UAV swarms, it achieves the highest throughput ($0.914$) and energy efficiency ($0.882$), exceeding KDN-DQN by 3.2% in throughput and 6.1% in energy efficiency, while sustaining post-recovery stability of $0.92$ and a system-level convergence time of 20.1 s. By grounding intent in symbolic knowledge and compiling it into verified, device-ready configurations, KAN advances network autonomy from opaque automation to transparent, intent-aligned cognition.

Accurate prediction of cellular network traffic is essential for efficient resource management as mobile Internet and 5G services continue to expand rapidly. However, existing methods struggle to capture complex spatiotemporal dependencies and often perform poorly on out-of-distribution data. To address these challenges, we propose a forecasting framework that leverages a pretrained large language model (LLM) and combines structured prompt engineering with time-series reprogramming. Specifically, we construct sparse neighborhood graphs using dynamic time warping (DTW) to capture long-term spatial correlations, while adaptive edge weights are employed for short-term dynamics. Furthermore, spatiotemporal priors are encoded into prompt prefixes to guide the LLM's reasoning. By reprogramming the time series data, we avoid the need for extensive model fine-tuning. Experiments on a real-world dataset demonstrate that our method outperforms classical and recent graph-based baselines by reducing prediction error by over 10% on average, and that each component of the framework contributes to the overall performance. Notably, it exhibits strong zero-shot generalization capability in unseen regions, offering a robust and scalable solution for future networks.

The low-altitude intelligent network (LAIN) is gaining significant attention from academia and industry. Fortunately, LAIN naturally aligns with near-field communications in extremely large-scale MIMO (XL-MIMO) systems. By leveraging near-field beamfocusing, LAIN can precisely direct beam energy to unmanned aerial vehicles, while the additional distance dimension boosts overall spectrum efficiency. However, near-field communications in LAIN still face several challenges, such as the increase in signal processing complexity and the necessity of distinguishing between far and near-field users. Inspired by the large language models (LLM) with powerful ability to handle complex problems, we apply LLM to solve challenges of near-field communications in LAIN. The objective of this article is to provide a comprehensive analysis and discussion on LLM-empowered near-field communications in LAIN. Specifically, we first introduce fundamentals of LLM and near-field communications, including the key advantages of LLM and key characteristics of near-field communications. Then, we reveal the opportunities and challenges of near-field communications in LAIN. To address these challenges, we present a LLM-based scheme for near-field communications in LAIN, and provide a case study which jointly distinguishes far and near-field users and designs multi-user precoding matrix. Finally, we outline and highlight several future research directions and open issues.

Contention resolution diversity slotted aloha (CRDSA) can offer an attractive way to promote the system throughput of the prevailing satellite-oriented internet of things (IoT) networks. In this work, energy efficiency of the CRDSA scheme based on non-orthogonal multiple access (NOMA), namely CRDSA-NOMA, is investigated for satellite IoT scenarios with battery-operated user equipments. Under the statistical hypothesis on the received power level (RePL), the transmission success rate of CRDSA-NOMA is first derived, and an energy-efficient CRDSA-NOMA scheme is then proposed by solving the formulated optimization problem, where the optimal probability distribution of RePL is obtained by leveraging the concept of differential evolution. After that, the upper bound of the achievable energy efficiency is presented through asymptotic analysis. Additionally, a load back-off (LBO) is integrated with the proposed scheme to handle the improper load deviation induced by the inaccurate analytical results. A candidate implementation of the proposed scheme is finally provided in the IoT system using DVB-RCS2 standard, where a lookup table is built to minimize the signaling overhead. Numerical simulation results illustrate the effectiveness of the proposed scheme, which experiences more elastic behaviors in terms of energy efficiency, throughput, and PLR, as compared to the other existing schemes.

This paper investigates the opportunistic array radar deployment problem for UAV swarms and proposes an array algorithm based on fuzzy correlation opportunistic planning. The algorithm utilizes credibility theory and comprehensively accounts for the randomness in UAV position distribution and the uncertainty in their operational states. Fuzzy variables are used to model the complex and uncertain environment of UAV deployment. A planning model for UAV opportunistic array radar deployment is developed, and an intelligent optimization algorithm is designed by integrating fuzzy simulation and non-dominated sorting genetic algorithms to solve the model. Simulation experiments confirm the effectiveness and stability of the proposed algorithm regarding monitoring coverage and power consumption under different array configurations for UAV swarm opportunistic array radar in uncertain environments, providing valuable guidance for engineering practice.

Simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS) enables full-space wireless coverage; However, its application in ultra-dense networks is constrained by the energy efficiency of massive multiple-input multiple-output (MIMO) systems, antenna coupling effects, and the complexity of successive interference cancellation (SIC) in power-domain multiple access. To address these challenges, polarization index modulation (PIM) is proposed for STAR-RIS-aided MIMO systems. In this scheme, information is encoded across three dimensions: the index of the dual-polarized antenna element (DP-AE), polarization symbols, and a codeword selected from a STAR-RIS block lookup codebook that maps public bits to a paired transmission and reflection block. The data is partitioned into public and private information, where public information encompasses spatial and polarization indices along with the codeword index, while private information, designated for specific users, comprises dual-polarized amplitude and phase modulation (APM) symbols. Propagation-level decoupling is employed to suppress inter-user interference (IUI), enhance spectral efficiency, and facilitate low-complexity reception. Simulation results demonstrate improved bit error rate (BER) performance and increased user capacity.

Non-orthogonal multiple access (NOMA) has been regarded as a promising multiple access technique for sixth-generation (6G) communication systems. This article focuses on designing a joint resource allocation scheme, including user grouping and computation offloading, for mobile edge computing (MEC)-aided unmanned aerial vehicle (UAV) networks with integrated communication and computing, which are referred to as UAV-MEC networks. In particular, multiple UAVs within a group upload their information to the local ground controller by employing NOMA to support massive connectivity. Specifically, to meet the novel 6G requirements, a time-oriented sequential coalition formation (TSCF) algorithm is developed in terms of a non-transferable utility coalition formation game, aiming to maximize the system's energy efficiency. In addition, a DDPG method is designed to provide adaptive power allocation for temporary combinatorial user groups until the final stable groups are formed. Furthermore, simulation results show that the proposed TSCF algorithm achieves a significant gain compared with the other benchmarks, and the stability of the proposed algorithms can be verified.

This work proposes a covert transmission scheme in the context of integrated sensing and communication (ISAC) scenario, where the radar with the aid of an intelligent reflecting surface (IRS) is employed to achieve covert transmission and target detection simultaneously. Specifically, we jointly design the radar’s transmit beamforming vector and the phase shifts of the IRS to maximize the covert rate subject to a covertness constraint, a total power constraint, and a radar detection constraint, which is represented by mutual information. We resort to alternating optimization and the semidefinite relaxation method to solve a non-convex optimization problem to achieve the above design. Numerical results illustrate the effectiveness of the covert transmission system aided by the IRS and reveal the performance trade-off between radar detection and covert communication. For given the maximum transmit power, a higher quality of radar detection for the target will consume more power and thus lead to covert communication performance reduction, i.e., the performance trade-off between target detection and covert communication. It is worth noting that the covert rate is an increasing function of the transmit power, yet it is subject to a covertness constraint.