The line-of-sight multiple-input multiple-output (LOS-MIMO) has emerged as a potential solution for increasing spectral efficiency in dense urban microwave links, given limited spectrum resources. Phase variation and power attenuation may introduce uncertainty to the channel estimation, affecting channel capacity and the performance of interference cancellation by zero-forcing receivers. This paper investigates the impact of this uncertainty in the presence of rain. Commercial backhaul links (CMLs) in cellular networks are not only essential for data transmission but have also proven useful for rainfall monitoring. They are now emerging as a new opportunistic integrated sensing and communication (OISAC) application for weather sensing. This paper also studies the use of LOS-MIMO backhaul technology for rain rate estimation. The availability of multiple data streams allows the number of rain estimation values to increase linearly with the minimum number of transmit and receive antennas in the MIMO link, Additionally, the potential for LOS-MIMO microwave links to retrieve parameters related to rain drop size distribution based on measurement data is also explored. This new backhaul solution shows great potential to be used for near-ground environmental monitoring and weather prediction studies, particularly with the advent of the big data era.

In this paper, we pay our attention towards secure and robust communication in the presence of a Reconfigurable Intelligent Surface (RIS)-enhanced mobile eavesdropping attacker in Multiple-Input Multiple-Output (MIMO) wireless networks. Specifically, we first provide a unifying framework that generalizes specific intelligent wiretap model wherein the passive eavesdropper configured with any number of antennas is potentially mobile and can actively optimize its received signal strength with the help of RIS by intelligently manipulating wiretap channel characteristics. To effectively mitigate this intractable threat, we then propose a novel and lightweight secure communication scheme from the perspective of information theory. The main idea is that the data processing can in some cases be observed as communication channel, and a random bit-flipping scheme is then carefully involved for the legitimate transmitter to minimize the mutual information between the secret message and the passive eavesdropper’s received data. The Singular Value Decomposition (SVD)-based precoding strategy is also implemented to optimize power allocation, and thus ensure that the legitimate receiver is not subject to interference from this random bit-flipping. The corresponding results depict that our secure communication scheme is practically desired, which does not require any a prior knowledge of the eavesdropper’s full instantaneous Channel State Information (ICSI). Perfect acquisition of ICSI is clearly always not affordable, which is further exacerbated by the RIS involved and the potential mobility of the passive eavesdropper that leads to unavoidable fast fading channels. Furthermore, we consider the RIS optimization problem from the eavesdropper’s perspective, and provide RIS phase shift design solutions under different attacking scenarios. Finally, the optimal detection schemes respectively for the legitimate user and the eavesdropper are provided, and comprehensive simulations are presented to verify our theoretical analysis and show the effectiveness and robustness of our secure communication scheme across a wide range of attacking scenarios.

A lightweight Sybil attack detection method based on an improved received signal strength indicator (RSSI) and collaborative node verification is presented to ensure the integrity of data collected from wireless sensor networks (WSNs) for machine learning applications. Sybil attacks can tamper with data, leading to misleading model outputs and reducing the algorithmic performance. To address this challenge, the proposed approach employs an adaptive moving average–Kalman hybrid filtering (AMA-KF) algorithm to optimize RSSI measurements by mitigating multipath interference and environmental noise. The filtered RSSI is then incorporated into a multi-node cooperative positioning mechanism that detects Sybil nodes through spatial signal consistency analysis. Under this mechanism, detection strategies are dynamically adapted to network types: distance consistency is verified in fixed-power networks, while RSSI ratios are analyzed in adjustable-power networks. Simulation results demonstrate a detection rate of 94% and a false positive rate of only 3%, outperforming existing methods. This approach offers a lightweight and effective solution to ensure reliable data acquisition in WSNs for trustworthy machine learning model training.

Multiple-symbol noncoherent detection (MSND) with the aid of Neural Networks (NNs) for low-density parity-check (LDPC) coded multiple phase shift keying (MPSK) signals is studied for Unmanned aerial vehicle (UAV) communications. In the traditional MSND scheme, the number of the candidate sequences grows exponentially with respect to the length of the symbol observation period. Implementing the optimal bit log-likelihood ratio (LLR) for decoding is challenging, even when the observed symbol period is two. In this paper, we first proposed an improved scheme to reduce the number of the candidate sequences by phase combination, the phase is uniformly quantized into L discrete values. We find that the performance requirements can be well met when the phase quantization order is only 4. Then we utilize Back Propagation neural networks (BPN) to compute the bit LLR. To enhance the training efficiency of our NNs and achieve better performance, we also uniformly quantize the carrier phase offset (CPO) into discrete states. The decoding convergence is accelerated significantly compared to the improved traditional scheme. The complexity is reduced to a certain extent within the acceptable range of performance loss.

The decision fusion rule and global optimality is studied for serial distributed detection in multihop multirelay wireless sensor networks (WSNs) under the end–edge–cloud orchestration. In particular, a multihop relay node serial distributed detection configuration is considered. Then, the optimal decision fusion rule is derived, and the detection probability and false alarm probability of the distributed detection system are given. Third, the suboptimal decision fusion rule under different conditions for the multihop relay channel is represented. Furthermore, in order to solve the high energy consumption and bandwidth limitation problems of WSNs, we consider the global optimization for serial distributed detection systems and obtain the sufficient condition. Finally, under different communication conditions, the detection performance of the system has been kept optimal when we increase the number of sensors in series. We validate the conclusions based on the numerical results. The sufficient condition for optimality detection of serial distributed detection multirelay sensor network system is satisfied then the system can achieve optimal detection performance.

The parallel distributed detection is studied for mobile wireless sensor networks (MWSNs) in the presence of Byzantine attacks in entirely unknown environment or complicated environment from the perspective of the information theory, where we pay most of our attention toward design, analysis, and evaluation of the attack strategy. In particular, the multihop relay network and the harsh wireless communication condition, e.g., the dynamic and entirely unknown channel, are taken into consideration in our configuration. Second, the conditions that the optimal attacking strategy should satisfy is analyzed and developed under different attacking scenarios. Third, the minimum attacking power is developed for the Byzantines to blind the fusion center (FC). Furthermore, the optimal attacking strategies are developed when no prior information of the system is known for the Byzantines. Finally, the traditional four typical attack strategies are evaluated, and we find that the fraction of Byzantines is the only factor that affects the reliable data fusion when the network size and the attacking strategy are fixed. The extensive simulation is conducted to verify our design, analysis, and evaluation of the attack strategy.

With the continuous advancement of wireless technologies, the design of cost-effective and efficient communication solutions to support large-scale Internet of Vehicles (IoV) systems has become increasingly critical. Reconfigurable Intelligent Surfaces (RIS), a cutting-edge technology in wireless communication, presents a promising solution to this challenge. This paper focuses on leveraging RIS to address the key obstacles encountered in vehicle communication. We propose an IoV communication system in which RIS is employed as the transmitter, operating under Nakagami-m fading channels. The basic system model consists of a radio frequency (RF) transmitter and an RIS, both serving as the transmission source. The signal is transmitted to a relay, which subsequently forwards it to multi-antenna-equipped vehicles via an amplify-and-forward (AF) scheme. We provide a thorough performance analysis of the system, deriving expressions for the outage probability and the average channel capacity. Furthermore, we validate the theoretical results through extensive simulations. Our findings suggest that RIS has the potential to significantly enhance the performance of IoV systems.

The technological advancements in diffusion models (DMs) have demonstrated unprecedented capabilities in text-to-image generation and are widely used in diverse applications. However, they have also raised significant societal concerns, such as the generation of harmful content and copyright disputes. Machine unlearning (MU) has emerged as a promising solution, capable of removing undesired generative capabilities from DMs. However, existing MU evaluation systems present several key challenges that can result in incomplete and inaccurate assessments. To address these issues, we propose UnlearnCanvas, a comprehensive high-resolution stylized image dataset that facilitates the evaluation of the unlearning of artistic styles and associated objects. This dataset enables the establishment of a standardized, automated evaluation framework with 7 quantitative metrics assessing various aspects of the unlearning performance for DMs. Through extensive experiments, we benchmark 9 state-of-the-art MU methods for DMs, revealing novel insights into their strengths, weaknesses, and underlying mechanisms. Additionally, we explore challenging unlearning scenarios for DMs to evaluate worst-case performance against adversarial prompts, the unlearning of finer-scale concepts, and sequential unlearning. We hope that this study can pave the way for developing more effective, accurate, and robust DM unlearning methods, ensuring safer and more ethical applications of DMs in the future. The dataset, benchmark, and codes are publicly available at this link.

The M-ary distributed decision fusion is studied for multi-hop amplify-and-forward Wireless Sensor Networks (WSNs), the implementation framework is given, and the performance analysis is developed. In particular, we first propose an M-ary distributed decision fusion configuration, wherein the multi-hop relay network is involved, and the relay node only forward information from their neighbors. Furthermore, the optimal decision rule with the explicit and exact form is derived, and the implementation structure is depicted. Our results show that, the Hamming Distance (HD), which is well-known in information theory, is finally involved in the optimal decision metric. Then, we derive the suboptimal decision fusion algorithms for three scenarios. Firstly, we achieve an interesting rule in the form of a Maximum Ratio Combining (MRC), wherein the local channel is considered to be ideal. When the idea of Equal Gain Combining (EGC) is directly followed, we propose the Minimum Hamming Distance Summation (MHDS) rule. We also give a Selective Combining (SC) statistic, wherein only received observation with the optimal quality is selected for decision fusion, and we correspondingly term this criterion as the Minimum Hamming Distance (MHD) rule. Secondly, we achieve a Chair-Varshney rule when it is assumed that the crossover probability of relay Binary Symmetric Channel (BSC) is small enough. The simple majority-based statistic is developed when the homogeneous multi-hop relay WSNs is considered. A statistic with the similar form of SC is also developed. Thirdly, when each relay BSC’s crossover probability is relatively large, we also propose the suboptimum fusion statistic in an analog form to the MRC and EGC, respectively. Our results show that the MHDS rule can be achieved via optimum decision rule in the first or third scenarios. The performance assessment is determined both Monte Carlo simulation and analysis.

Activation maximization (AM) refers to the task of generating input examples that maximize the activation of a target class of a classifier, which can be used for class-conditional image generation and model interpretation. A popular class of AM method, GAN-based AM, introduces a GAN pre-trained on a large image set, and performs AM over its input random seed or style embeddings, so that the generated images are natural and adversarial attacks are prevented. Most of these methods would require the image set to contain some images of the target class to be visualized. Otherwise they tend to generate other seen class images that most maximizes the target class activation. In this paper, we aim to tackle the case where information about the target class is completely removed from the image set. This would ensure that the generated images truly reflect the target class information residing in the classifier, not the target class information in the image set, which contributes to a more faithful interpretation technique. To this end, we propose PROBEGAN, a GAN-based AM algorithm capable of generating image classes unseen in the image set. Rather than using a pre-trained GAN, PROBEGAN trains a new GAN with AM explicitly included in its training objective. PROBEGAN consists of a class-conditional generator, a seen-class discriminator, and an all-class unconditional discriminator. It can be shown that such a framework can generate images with the features of the unseen target class, while retaining the naturalness as depicted in the image set. Experiments have shown that PROBEGAN can generate unseen-class images with much higher quality than the baselines. We also explore using PROBEGAN as a model interpretation tool. Our code is at https://github.com/csmiler/ProbeGAN/.

Current deep neural networks (DNNs) are vulnerable to adversarial attacks, where adversarial perturbations to the inputs can change or manipulate classification. To defend against such attacks, an effective and popular approach, known as adversarial training (AT), has been shown to mitigate the negative impact of adversarial attacks by virtue of a min-max robust training method. While effective, it remains unclear whether it can successfully be adapted to the distributed learning context. The power of distributed optimization over multiple machines enables us to scale up robust training over large models and datasets. Spurred by that, we propose distributed adversarial training (DAT), a large-batch adversarial training framework implemented over multiple machines. We show that DAT is general, which supports training over labeled and unlabeled data, multiple types of attack generation methods, and gradient compression operations favored for distributed optimization. Theoretically, we provide, under standard conditions in the optimization theory, the convergence rate of DAT to the first-order stationary points in general non-convex settings. Empirically, we demonstrate that DAT either matches or outperforms state-of-the-art robust accuracies and achieves a graceful training speedup (e.g., on ResNet-50 under ImageNet). Codes are available at https://github.com/dat-2022/dat.

Recent works have developed several methods of defending neural networks against adversarial attacks with certified guarantees. However, these techniques can be computationally costly due to the use of certification during training. We develop a new regularizer that is both more efficient than existing certified defenses, requiring only one additional forward propagation through a network, and can be used to train networks with similar certified accuracy. Through experiments on MNIST and CIFAR-10 we demonstrate improvements in training speed and comparable certified accuracy compared to state-of-the-art certified defenses.

Contrastive learning (CL) can learn generalizable feature representations and achieve state-of-the-art performance of downstream tasks by finetuning a linear classifier on top of it. However, as adversarial robustness becomes vital in image classification, it remains unclear whether or not CL is able to preserve robustness to downstream tasks. The main challenge is that in the self-supervised pretraining + supervised finetuning paradigm, adversarial robustness is easily forgotten due to a learning task mismatch from pretraining to finetuning. We call such challenge 'cross-task robustness transferability'. To address the above problem, in this paper we revisit and advance CL principles through the lens of robustness enhancement. We show that (1) the design of contrastive views matters: High-frequency components of images are beneficial to improving model robustness; (2) Augmenting CL with pseudo-supervision stimulus (e.g., resorting to feature clustering) helps preserve robustness without forgetting. Equipped with our new designs, we propose AdvCL, a novel adversarial contrastive pretraining framework. We show that AdvCL is able to enhance cross-task robustness transferability without loss of model accuracy and finetuning efficiency. With a thorough experimental study, we demonstrate that AdvCL outperforms the state-of-the-art self-supervised robust learning methods across multiple datasets (CIFAR-10, CIFAR-100, and STL-10) and finetuning schemes (linear evaluation and full model finetuning).

Although the full multiple-symbol detection (MSD) for IEEE 802.15.4c multiple phase shift keying (MPSK) receivers gives much better performance than the symbol-by-symbol detection (SBSD), its implementation complexity is extremely heavy. We propose a simple MSD scheme based on two implementation-friendly but powerful strategies. First, we find the best and second-best decisions in each symbol position with the standard SBSD procedure, and the global best decision is frozen. Second, for the remaining symbol positions, only the best and second-best symbol decisions, not all the candidates, are jointly searched by the standard MSD procedure. The simulation results indicate that the packet error rate (PER) performance of the simplified MSD scheme is almost the same as that of the full scheme. In particular, at PER of 1 × 10 − 3 , no more than 0.2 dB performance gap is observed if we just increase the observation window length N to 2. However, the number of decision metrics needed to be calculated is reduced from 256 to 2. Thus, much balance gain between implementation complexity and detection performance is achieved.

Recent works have empirically shown that there exist adversarial examples that can be hidden from neural network interpretability (namely, making network interpretation maps visually similar), or interpretability is itself susceptible to adversarial attacks. In this paper, we theoretically show that with a proper measurement of interpretation, it is actually difficult to prevent prediction-evasion adversarial attacks from causing interpretation discrepancy, as confirmed by experiments on MNIST, CIFAR-10 and Restricted ImageNet. Spurred by that, we develop an interpretability-aware defensive scheme built only on promoting robust interpretation (without the need for resorting to adversarial loss minimization). We show that our defense achieves both robust classification and robust interpretation, outperforming state-of-the-art adversarial training methods against attacks of large perturbation in particular.

In this paper, real outdoor measurements were conducted to examine the performance of spatially separated 2×2 line-of-sight multiple-input multiple-output (LOS-MIMO) backhaul deployments at millimeter-wave frequency in city environment. The performance of the backhaul link transmission and received signal level are analyzed over a period in both sunny and rainy weather. We study the variations of signal propagation due to rain and compare the performance with theoretical rain model given by ITU. The impact of wet antenna attenuation and other uncertainties are analyzed based on the measurement data.

Millimeter-wave technology with its large bandwidth is the fastest growing small-cell backhaul solution for next generation wireless networks. E-band can provide twice 5 GHz bandwidth, offering 10 GHz aggregate spectrum (71-76 GHz and 81-86 GHz) and enable Gbps data rates. In this paper, E-band radio performance in line- of-sight (LOS) path conditions was recorded and correlated with weather statistic.

In this study, the authors propose a simple and robust near‐optimal single differential coherent detection scheme for IEEE 802.15.4 binary phase shift keying (BPSK) receivers. The detection process is initiated by a bootstrap processor, which is responsible for estimating and compensating the residual carrier frequency offset effect in the sample from the autocorrelator output. In particular, for this bootstrap processor, a simple estimator with only one addition operation is provided. This low complexity benefits from that the high signal‐to‐noise ratio and a trigonometric approximation are combined, i.e. , for simplification of the full estimation scheme. Further, to achieve near optimal performance as well as robustness to carrier frequency offset, the observation space is subdivided into four equi‐angular regions to decrease the estimation error introduced by the involved approximation. The proposed method is validated through detailed physical layer (PHY) simulations according to the IEEE 802.15.4 standard. The simulation results demonstrate that, compared with the full estimator, the proposed algorithm can acquire a fairly accurate estimation, with almost no loss of reliability and robustness, whereas complexity reduction is also achieved.

In this paper, we pay our attention towards the noncoherent demodulation aspect of binary phase shift keying (BPSK) receivers for IEEE 802.15.4 wireless sensor networks (WSNs), and a carrier frequency offset invariant as well as error-floor free multiple-symbol differential detection (MSDD) strategy is proposed over the flat fading channel. This detector is an alternative to the multiple-symbol detector that has been considered almost exclusively in the past. In this new configuration, the receivers do not perform chip-level precompensation as in conventional scheme but bit-level postcompensation. That is, the bit-level autocorrelation operation is first implemented with the “raw” chip sample, and then the carrier frequency offset effect (CFOE) embedded in the achieved statistic is compensated. Correspondingly, the cumulative error in the detection metric is decreased so much that the pervasive error floor for the conventional MSDD scheme is suppressed. Also, complexity efficient estimators for the MSDD scheme are reinvestigated, analyzed, and summarized. Simulation results demonstrate that this new detection strategy may achieve rather more encouraging gain from differential and spread spectrum coding than the conventional single differential coherent detection (SDCD) scheme. The pervasive error floor is also eliminated as compared with conventional MSDD scheme even if the most simple estimator is configured under large bit observation length. Then, much transmitting energy may be saved for each chip symbol, which is practically desired for transmit-only nodes in WSNs.

An asymmetric zero correlation zone (A-ZCZ) sequence set consists of multiple ZCZ sequence subsets and it is a generalized version of conventional zero correlation zone (C-ZCZ) sequence set. In this paper, a novel construction of A-ZCZ sequence sets is presented by interleaving a given CZCZ sequence set. One of the most important properties of the proposed A-ZCZ sequence set is the cross-correlation function between arbitrary sequences belonging to different sequence subsets, which has quite a large zero-cross-correlation zone (ZCCZ). The optimal A- ZCZ sequence set can be obtained by choosing an appropriate shift sequence and optimal C-ZCZ sequence set.

Many approaches to build a wiretap channel (WTC) by multi‐input multi‐output system have been introduced. Different from those approaches, the authors propose a feedback method combined with the low‐density parity‐check (LDPC) codes for building the WTC I (WTC‐I) to achieve unconditional security under single‐input single‐output system by the soft decision decoding. The novel approach establishes the WTC‐I on both the binary symmetric channel and binary input additive white Gaussian noise channel. In order to keep the eavesdropper be fully ignorant about the secret information, randomness is added to the feedback signals from the destination by taking advantage of feedback. In addition, the message to be sent is encoded by the LDPC codes such that it can be correctly decoded by a legitimate receiver. Furthermore, the secret information transmission capacity can be improved by the soft decision decoding.

This study investigates the framework of associating multiple‐input–multiple‐output (MIMO) beamforming with secure code to achieve unconditional secure communications in the wireless passive eavesdropping environment. The schemes are based on a two‐step method under Wyner's wiretap channel model. First, with MIMO transmit beamforming, one can utilise the spatial degree of freedom to cripple eavesdroppers’ interceptions even when he does not know the eavesdropper's channel state information. Consequently, by taking the threshold characteristics of the secure code, the legitimate receivers will continue to extend an average bit error rate advantage over eavesdroppers when they share similar conditions (background noise power and channel gains). By this way, the proposed system could achieve almost zero information obtained by the eavesdroppers while still keeping rather lower error transmissions for the main channel. A profound theoretical analysis for the MIMO advantage channel and the exact closed‐form expressions of secrecy outage probability for the secure code joint system are presented. The authors launch extensive experiments to verify the proposed security systems and demonstrate its feasibility and implement ability.

In this study, a modified channel‐independent weighted bit flipping (CIWBF) decoding algorithm is proposed for low‐density‐parity‐check codes. This modification, based on the ‘additive offset term adjustment’ in the reliability of the check sum, can also be applied to the reliability ratio weighted bit flipping algorithm. Simulation results show that the performance of the author's improved modified CIWBF algorithm is better than that of CIWBF algorithm about 0.45 dB at bit‐error rate of 10 −5 over an additive white Gaussian noise channel without much complexity increase.