This paper proposes a new technique for soft information relaying, which is based on a soft decode–compress–forward relay protocol. The proposed system provides a means of using distributed low-density parity-check (LDPC) coding in conjunction with higher order modulation, such as pulse amplitude modulation (PAM) and quadrature amplitude modulation (QAM), which is effective even under poor source–relay link conditions. Ordinarily, such schemes suffer from error propagation to the destination caused by incorrect decoding at the relay when the signal-to-noise ratio (SNR) on the source–relay link is low; however, our system avoids this problem by generating soft versions of the additional (parity-bearing) PAM symbols for transmission from the relay. The proposed technique of soft compression does not suffer from parity log-likelihood ratios (LLRs) converging to zero, as do many soft re-encoding techniques for turbo and LDPC codes. In the case of Gray-coded PAM/QAM signaling, we also propose a method of performing exact expectation-based soft modulation with low computational complexity. Furthermore, we propose a new model, which we refer to as the soft scalar model, for the overall source-to-destination channel encountered by the constellation symbols, and this model is used at the destination to compute LLRs for joint decoding of the distributed LDPC code. Simulation results demonstrate that the proposed scheme can provide good coding gain, diversity gain, and spectral efficiency under poor source–relay SNR conditions.

A Soft Decode–Compress–Forward Relaying Scheme for Cooperative Wireless Networks

In this paper, we propose a new distributed coding structure with a soft input soft output (SISO) relay encoder for error-prone parallel relay channels. We refer to it as the distributed soft coding (DISC). In the proposed scheme, each relay first uses the received noisy signals to calculate the soft bit estimate (SBE) of the source symbols. A simple SISO encoder is developed to encode the SBEs of source symbols based on a constituent code generator matrix. The SISO encoder outputs at different relays are then forwarded to the destination and form a distributed codeword. The performance of the proposed scheme is analyzed. It is shown that its performance is determined by the generator sequence weight (GSW) of the relay constituent codes, where the GSW of a constituent code is defined as the number of ones in its generator sequence. A new coding design criterion for optimally assigning the constituent codes to all the relays is proposed based on the analysis. Results show that the proposed DISC can effectively circumvent the error propagation due to the decoding errors in the conventional detect and forward (DF) with relay re-encoding and bring considerable coding gains, compared to the conventional soft information relaying.

Distributed Soft Coding with a Soft Input Soft Output (SISO) Relay Encoder in Parallel Relay Channels

In this paper, error performance analysis of decode-and-forward (DF)-based cooperative vehicular networks with relay selection is investigated. All of the links in the system are modeled as cascaded Nakagami-m distributed random variables which provide a realistic description of an intervehicular channel. We employ virtual-noise (VN)-based demodulation, maximum likelihood (ML) demodulation, cooperative maximal ratio combining (C-MRC), and log-likelihood-ratio-based transmissions in mitigating error propagation effect encountered in DF cooperative networks. In order to obtain more effective solutions and improve the system performance, we introduce hybrid anti-error propagation approaches, namely VN-MRC, VN-ML, and VN-LLR in which relay selection criterion is derived by using conditional bit error rate (BER) expressions of VN technique while data detection is performed by utilizing maximal ratio combining (MRC), ML detection, C-MRC and LLR transmission methods. The performance analyses and the numerical results for cascaded Nakagami-m channels have shown that VN-LLR approach provides significant performance improvement with respect to the other considered techniques in DF cooperative vehicular systems with relay selection.

/pdf/relay-selection-for-df-based-cooperative-vehicular-systems-5d8kt2hzwo.pdf

Relay selection for DF-based cooperative vehicular systems

In this paper we propose a novel power adaptive network coding (PANC) for a non-orthogonal multiple-access relay channel (MARC), where two sources transmit their information simultaneously to the destination with the help of a relay. In contrast to the conventional XOR-based network coding (CXNC), the relay in PANC generates network coded symbols by considering the coefficients of the source-to-relay channels, and forwards each symbol with a pre-optimized power level. Specifically, by defining a symbol pair as two symbols from the two sources, we first derive the expression of symbol pair error rate (SPER) for the system. Noting that deriving the exact SPER are complex due to the irregularity of the decision regions caused by random channel coefficients, we propose a coordinate transform (CT) method on the received constellation to simplify the derivations of the SPER. Next, we obtain the optimal power level by decomposing it as a multiplication of a power scaling factor and a power adaptation factor. We prove that with the power scaling factor at the relay, our PANC scheme can achieve a full diversity gain, i.e., an order of two diversity gain, while the CXNC can achieve only an order of one diversity gain. In addition, we optimize the power adaptation factor at the relay to minimize the SPER at the destination by considering of the relationship between SPER and minimum Euclidean distance of the received constellation, resulting in an improved coding gain. Simulation results show that (1) the SPER derived based on our CT method can well approximate the exact SPER with a much lower complexity; (2) the PANC scheme with power adaptation optimizations and power scaling factor design can achieve a full diversity, and obtain a much higher coding gain than other network coding schemes.

Power Adaptive Network Coding for a Non-Orthogonal Multiple-Access Relay Channel

This paper proposes the novel technique of multilevel-threshold-based soft quantization (MLT-SQ) for a multiple-access relay system (MARS). The scheme is suitable for systems using binary phase-shift keying (BPSK) and network coding at the relay. In the proposed MLT-SQ protocol, the relay evaluates the reliabilities, which are expressed as log-likelihood ratios (LLRs), of the received signals from the two sources. It then computes the LLRs of the network-coded packet and quantizes these using a set of optimized multilevel thresholds, forwarding the resulting “ quantized soft symbols ” to the destination. We provide the derivation for the bit error rate (BER) at the destination, and based on this, we optimize the multilevel thresholds to minimize the BER. Simulation results are provided for the proposed MLT-SQ system, both without coding and for the case where low-density parity-check (LDPC) coding is employed. The proposed system achieves full diversity order. Compared with competing schemes, the performance of our system is superior in terms of BER when the same amount of channel state information (CSI) is exploited.

A Soft-Network-Coded Multilevel Forwarding Scheme for Multiple-Access Relay Systems

In this paper, we investigate novel soft mutual information forwarding (MIF) protocols in a two-way relay channel (TWRC), where two sources exchange information with the help of an intermediate relay. Based on the estimated signals from the two sources, the relay calculates the soft mutual information and then broadcasts it to the two sources. In particular, we propose two MIF protocols, namely, network-coded MIF (NC-MIF) and superposition-coded MIF (SC-MIF), which is suitable for different channel conditions. The expressions derived for the received signal-to-noise ratio (SNR) at the receivers reveal that, if both source-to-relay channels are in good condition, the NC-MIF outperforms the SC-MIF. Otherwise, the SC-MIF is superior to the NC-MIF. For the TWRC with varying channels, we further develop an adaptive scheme, which enables the dynamic switch between the two protocols, depending on the received SNR at the sources. Furthermore, the threshold that determines the switch of the protocols is developed as a close-form expression. Simulation results show that our adaptive scheme outperforms other protocols discussed in this paper over fading channels.

Novel Soft Information Forwarding Protocols in Two-Way Relay Channels

This letter considers the mutual information based soft forwarding (MIF) scheme for a memoryless parallel relay network in Rayleigh fading channels. The exact analytical expression for soft noise variance in Rayleigh fading channels is first given. A tight upper bound of the soft noise variance of the relayed signal and the input-output relationship at the relay is then derived. Furthermore, it is revealed that the MIF scheme in Rayleigh fading channels can achieve a full diversity order. Simulation results show that the MIF scheme yields about 0.8 - 1.8 dB signal-to-noise ratio (SNR) gains compared to the various full diversity-achieving schemes in the literature.

Soft Information Relaying in Fading Channels

Drying is a complex process of simultaneous heat, mass, and momentum transport phenomena with continuous phase changes. Numerical modelling is one of the most effective tools to mechanistically express the different physics of drying processes for accurately predicting the drying kinetics and understanding the morphological changes during drying. However, the mathematical modelling of drying processes is complex and computationally very expensive due to multiphysics and the multiscale nature of heat and mass transfer during drying. Physics-informed machine learning (PIML)-based modelling has the potential to overcome these drawbacks and could be an exciting new addition to drying research for describing drying processes by embedding fundamental transport laws and constraints in machine learning models. To develop such a novel PIML-based model for drying applications, it is necessary to have a fundamental understanding of heat, mass, and momentum transfer processes and their mathematical formulation of drying processes, in addition to data-driven modelling knowledge. Based on a comprehensive literature review, this paper presents two types of information: fundamental physics-based information about drying processes and data-driven modelling strategies to develop PIML-based models for drying applications. The current status of physics-based models and PIML-based models and their limitations are discussed. A sample PIML-based modelling framework for drying application is presented. Finally, the challenges of addressing simultaneous heat, mass, and momentum transport phenomena in PIML modelling for optimizing the drying process are presented at the end of this paper. It is expected that the information in this manuscript will be beneficial for further advancing the field.

/pdf/fundamental-understanding-of-heat-and-mass-transfer-2xu9y9th.pdf

Fundamental Understanding of Heat and Mass Transfer Processes for Physics-Informed Machine Learning-Based Drying Modelling

The burgeoning fields of machine learning (ML) and quantum machine learning (QML) have shown remarkable potential in tackling complex problems across various domains. However, their susceptibility to adversarial attacks raises concerns when deploying these systems in security sensitive applications. In this study, we present a comparative analysis of the vulnerability of ML and QML models, specifically conventional neural networks (NN) and quantum neural networks (QNN), to adversarial attacks using a malware dataset. We utilize a software supply chain attack dataset known as ClaMP and develop two distinct models for QNN and NN, employing Pennylane for quantum implementations and TensorFlow and Keras for traditional implementations. Our methodology involves crafting adversarial samples by introducing random noise to a small portion of the dataset and evaluating the impact on the models performance using accuracy, precision, recall, and F1 score metrics. Based on our observations, both ML and QML models exhibit vulnerability to adversarial attacks. While the QNNs accuracy decreases more significantly compared to the NN after the attack, it demonstrates better performance in terms of precision and recall, indicating higher resilience in detecting true positives under adversarial conditions. We also find that adversarial samples crafted for one model type can impair the performance of the other, highlighting the need for robust defense mechanisms. Our study serves as a foundation for future research focused on enhancing the security and resilience of ML and QML models, particularly QNN, given its recent advancements. A more extensive range of experiments will be conducted to better understand the performance and robustness of both models in the face of adversarial attacks.

Exploring the Vulnerabilities of Machine Learning and Quantum Machine Learning to Adversarial Attacks using a Malware Dataset: A Comparative Analysis

Bridges play a vital part in the transportation system by ensuring the connectedness of transportation systems, which is critical for a country’s social and economic prosperity by offering daily mobility to the people. However, according to the American Society of Civil Engineers (ASCE 2017), many U.S. bridges are in critical condition, raising safety issues, with 9.1 and 13.6 percent of the country’s 614,387 bridges, respectively, structurally defective, and functionally obsolete. Every day, 178 million people traverse these structurally defective bridges. Furthermore, the average annual failure rate is expected to be between 87 and 222. Bridge breakdowns have disastrous repercussions, and in many cases, result in death. While bridge authorities strive to improve bridge conditions, budget limits make it difficult to make cost-effective maintenance decisions. Bridge authorities distribute limited repair resources based on projected future bridge conditions. As a result, building a data-driven, autonomous, and effective bridge condition prediction model is critical for improving maintenance decision-making. In this paper, we present a novel bridge condition prediction framework using advanced Machine Learning (ML) algorithms on the National Bridge Inventory (NBI) dataset. The framework consists of two stages, where the most informative features from the NBI dataset are selected using the Recursive Feature Elimination process and in the 2nd step, ML classifiers are applied to the selected features for bridge condition prediction. The experimental results show that the proposed framework can effectively predict bridge conditions by producing highly accurate results in terms of accuracy, precision, recall, and f1-score.

M. A. Karim

Papers

Novel Soft Information Forwarding Protocols in Two-Way Relay Channels

Soft Information Relaying in Fading Channels

Fundamental Understanding of Heat and Mass Transfer Processes for Physics-Informed Machine Learning-Based Drying Modelling

Exploring the Vulnerabilities of Machine Learning and Quantum Machine Learning to Adversarial Attacks using a Malware Dataset: A Comparative Analysis

A Novel Machine Learning Based Framework for Bridge Condition Analysis