Reduction In Computational Complexity Research Articles

Joint Azimuth, Elevation and Delay Estimation for Single Base Station Localization in 3D IIoT

Integrated sensing and communication (ISAC) in the Industrial Internet of Things (IIoT) presents unique challenges in terms of localization techniques. While three-dimensional (3D) environments offer extra challenges to enhanced accuracy and realism, research in this area remains limited. To bridge this gap, we propose a novel localization technique assisted by a single base station (BS) in 3D IIoT scenarios. Our approach employs the MUltiple SIgnal Classification (MUSIC) algorithm to jointly estimate the angle of arrival (AoA) in azimuth and elevation, as well as the time of arrival (ToA). Compared to conventional multi-BS-assisted or MUSIC-based algorithms, our technique offers flexibility, easy implementation, and low computational cost. To improve performance, we integrate the Taylor-series into the iterative process after a MUSIC-based joint azimuth, elevation angle and delay estimation (JAEDE), resulting in a significant 99% reduction in computational complexity compared to a two-step MUSIC-based approach utilizing coarse-fine grid searching. Through numerical simulations, we compare our algorithm with three other MUSIC-based joint or separate estimation approaches, demonstrating its superior performance in azimuth angle of arrival (AoAz), elevation angle of arrival (AoAe), TOA, and overall location estimation across varying signal-to-noise ratio (SNR) conditions.

Model order reduction methods applied to neural network training

AbstractNeural networks have emerged as powerful and versatile tools in the field of deep learning. As the complexity of the task increases, so do size and architectural complexity of the network, causing compression techniques to become a focus of current research. Parameter truncation can provide a significant reduction in memory and computational complexity. Originating from a model order reduction framework, the Discrete Empirical Interpolation Method is applied to the gradient descent training of neural networks and analyze for important parameters. The approach for various state‐of‐the‐art neural networks is compared to established truncation methods. Further metrics like L2 and Cross‐Entropy Loss, as well as accuracy and compression rate are reported.

Multiscale Entanglement Renormalization Ansatz: Causality and Error Correction

Computational complexity reduction is at the basis of a new formulation of many-body quantum states according to tensor network ansatz, originally framed in one-dimensional lattices. In order to include long-range entanglement characterizing phase transitions, the multiscale entanglement renormalization ansatz (MERA) defines a sequence of coarse-grained lattices, obtained by targeting the map of a scale-invariant system into an identical coarse-grained one. The quantum circuit associated with this hierarchical structure includes the definition of causal relations and unitary extensions, leading to the definition of ground subspaces as stabilizer codes. The emerging error correcting codes are referred to logical indices located at the highest hierarchical level and to physical indices yielded by redundancy, framed in the AdS-CFT correspondence as holographic quantum codes with bulk and boundary indices, respectively. In a use-case scenario based on errors consisting of spin erasure, the correction is implemented as the reconstruction of a bulk local operator.

A DCNN-LSTM based human activity recognition by mobile and wearable sensor networks

Systems for recognizing human activities are being developed as part of a larger framework that will allow for continuous monitoring of human behavior in intelligent home environments for elderly care, ambient assisted living, rehabilitation, sports injury detection, surveillance, and entertainment. The most challenging component of the sensor-based human activity detection pipeline for mobile and wearable devices is extracting relevant features. Using mobile and wearable sensors, the most challenging step in the pipeline for recognizing the behaviors of humans is extracting relevant features. The algorithm's performance is affected by feature extraction, which also reduces computation complexity and time. However, existing methods of recognizing human activity rely on hand-crafted characteristics that cannot handle complicated behaviors. There are several known systems for recognizing individual activities, including deep learning, but more must be developed to recognize transitions between actions. To solve these challenges, a new deep learning-based system is proposed for recognizing human actions in this research. Initially, a fuzzy logic-based genetic algorithm (FGA) is used for extracting features from sensor data. The introduced feature selection approach, called PSOACO, aims to enhance the particle swarm optimizer (PSO) algorithm's performance through the ant colony optimizer (ACO) operators. Then, the deep learning approach, named Deep Convolutional Neural Network and Long Short Term Memory (DCNN-LSTM), is used to classify the activities. F1 Score, Recall, Precision, and Accuracy are used to analyze the implementation's effectiveness. The overall recognition accuracy of the proposed model for UCI-HAR, PAMAP2, and WISDM datasets is 99.92%, 99.86%, and 99.94 %. According to the implementation, the HAR with privacy performs better than existing methods done on the Python platform.

Large Separable Kernel Attention: Rethinking the Large Kernel Attention design in CNN

Visual Attention Networks (VAN) with Large Kernel Attention (LKA) modules have been shown to provide remarkable performance, that surpasses Vision Transformers (ViTs), on a range of vision-based tasks. However, the depth-wise convolutional layer in these LKA modules incurs a quadratic increase in the computational and memory footprints with increasing convolutional kernel size. To mitigate these problems and to enable the use of extremely large convolutional kernels in the attention modules of VAN, we propose a family of Large Separable Kernel Attention modules, termed LSKA. LSKA decomposes the 2D convolutional kernel of the depth-wise convolutional layer into cascaded horizontal and vertical 1-D kernels. In contrast to the standard LKA design, the proposed decomposition enables the direct use of the depth-wise convolutional layer with large kernels in the attention module, without requiring any extra blocks. We demonstrate that the proposed LSKA module in VAN can achieve comparable performance with the standard LKA module and incur lower computational complexity and memory footprints. We also find that the proposed LSKA design biases the VAN more toward the shape of the object than the texture with increasing kernel size. Additionally, we benchmark the robustness of the LKA and LSKA in VAN, ViTs, and the recent ConvNeXt on the five corrupted versions of the ImageNet dataset that are largely unexplored in the previous works. Our extensive experimental results show that the proposed LSKA module in VAN provides a significant reduction in computational complexity and memory footprints with increasing kernel size while outperforming ViTs, ConvNeXt, and providing similar performance compared to the LKA module in VAN on object recognition, object detection, semantic segmentation, and robustness tests.

Improving surface defect detection with context-guided asymmetric modulation networks and confidence-boosting loss

Segmentation networks based on deep learning are widely used in the field of surface defect detection to ensure product quality. However, due to the complexity of defects and limited datasets, it is difficult for segmentation networks to achieve good performance in single-shot predictions. Therefore, we propose a novel context-guided asymmetric modulation network (called CAMNet) to improve the segmentation performance of existing methods. Inspired by the idea of cascading, CAMNet employs a coarse-to-fine segmentation framework that treats coarse predictions as priors to refine the extracted deep features. For this purpose, two asymmetric feature modulation modules, APM and ACM, are constructed in the spatial and channel dimensions, respectively. They are specially designed to leverage the rich context to aggregate and update features for fine prediction, and they are also lightweight and efficient, allowing for a significant reduction in computational complexity without sacrificing performance. In addition, a confidence-boosting loss is proposed to further widen the performance gap between the two prediction stages of CAMNet. Extensive experiments on three industrial datasets (PCB, Magnetic-tile, and DAGM2007) confirm the effectiveness of our method, showing that CAMNet can consistently achieve performance gains across different baselines and input scales. For example, CAMNet improves the IoU scores of FCN and GCNet on DAGM2007 by 5.92% and 4.98%, respectively.

Performance-Aware NILM Model Optimization for Edge Deployment

Non-Intrusive Load Monitoring (NILM) describes the extraction of the individual consumption pattern of a domestic appliance from the aggregated household consumption. Nowadays, the NILM research focus is shifted towards practical NILM applications, such as edge deployment, to accelerate the transition towards a greener energy future. NILM applications at the edge eliminate privacy concerns and data transmission-related problems. However, edge resource restrictions pose additional challenges to NILM. NILM approaches are usually not designed to run on edge devices with limited computational capacity, and therefore model optimization is required for better resource management. Recent works have started investigating NILM model optimization, but they utilize compression approaches arbitrarily without considering the trade-off between model performance and computational cost. In this work, we present a NILM model optimization framework for edge deployment. The proposed edge optimization engine optimizes a NILM model for edge deployment depending on the edge device’s limitations and includes a novel performance-aware algorithm to reduce the model’s computational complexity. We validate our methodology on three edge application scenarios for four domestic appliances and four model architectures. Experimental results demonstrate that the proposed optimization approach can lead up to a 36.3% average reduction of model computational complexity and a 75% reduction of storage requirements.

Beamforming Design in Short-Packet Transmission for URLLC in Cell-Free Massive MIMO System

Low latency, high reliability, and high data rate are three performance indicators of Ultra-reliable and low latency communication (URLLC) systems, but they are not able to be met simultaneously. Based on the determined latency and reliability requirements, this paper investigates the problem of the weighted sum rate (WSR) in cell-free massive multi-input multi-output (MIMO) system with multiple access points (APs). User-centric access point selection is considered to further reduce the computational complexity for AP used with acceptable performance loss. The transmitting beamforming is designed to mitigate inter-user interference so as to maximize the WSR under the backhaul and transmitting power constraints. Since the proposed problem is non-convex, two algorithms are proposed to solve the problem under the general SINR regime and the high SINR regime, respectively. For the former, we first approximate the objective function with the difference of convex (DC) programming and reformulate the problem into a semi-definite relaxation (SDR) form. Afterward, auxiliary variables are introduced in accordance with the successive convex approximation (SCA) method to further transform the nonconvex problem into a convex one. In the end, the sub-optimal WSR is obtained by iteratively solving a final convex problem. For the latter, the well-known WMMSE method is applied to address the WSR problem. Numerical results demonstrate the efficiency of our proposed algorithm, e.g., the maximum performance gain of the proposed algorithm is about 25% compared to the classical MMSE method. And the user-centric approach achieves 8 times computational complexity reduction with 4% performance loss.

Nonlinear distortion mitigation of DML-based OFDM optical systems with non-orthogonal DFT-precoding.

We propose a non-orthogonal discrete Fourier transform (DFT) matrix precoding scheme for the mitigation of nonlinear distortion induced by the interaction between laser chirp and fiber dispersion in a directly modulated laser (DML)-based orthogonal frequency division multiplexing (OFDM) transmission system. Compared with conventional OFDM, the proposed method can decrease the peak-to-average power ratio (PAPR) and significantly reduce the nonlinear distortion without sacrificing spectral efficiency (SE). The cascaded binary-phase-shift-keying iterative detection (CBID) algorithm is used to eliminate the inter-carrier interference (ICI) that is purposely induced by the non-orthogonal precoding. The performance of the proposed scheme is experimentally evaluated, achieving ∼0.4-dB sensitivity improvement at the KP4-forward error correction (FEC) threshold over the T/2-spaced third-order Volterra nonlinear equalizer (VNLE). Meanwhile, compared to the VNLE, the reduction in computational complexity of one OFDM frame is 90% for multiplication and 88.32% for addition in this work.

An optimal load curtailment method for operational risk assessment of integrated electricity, natural gas and heating systems

The rapid development of gas-fired generators and combined heat and power (CHP) units promotes the wide application of integrated electricity, natural gas and heating systems (IEGHSs). IEGHSs bring many opportunities to energy supply systems, but also produce huge challenges for ensuring secure and reliable energy supply. Therefore, it is indispensable to investigate the risk assessment for IEGHS operations, especially considering the complexity of system operations. First, a steady-state system model of IEGHSs is developed and linearized taking into account system operation characteristics, including generator operation constraints and the flexibility of CHP units. To reduce computation complexity, an optimal load curtailment method using a mixed integer linear programming algorithm is proposed, which considers both energy types and consumer importance of IEGHSs. Moreover, a contingency set based on the N-1 guideline is constructed taking into account the failures of IEGHSs with high probability. Additionally, the risk indices of IEGHSs are developed in two hierarchies: contingency level and system level. Furthermore, the operational risk of IEGHSs under the constructed contingency set is quantitatively evaluated using the proposed load curtailment method and the developed risk indices. Finally, case studies demonstrate the necessity of operational risk assessment and the effectiveness of the proposed method.

Hidden conditional random field-based equalizer for the 3D-CAP-64 transmission of OAM mode-division multiplexed ring-core fiber communication.

As a key technique for achieving ultra-high capacity optical fiber communications, orbital angular momentum (OAM) mode-division multiplexing (MDM) is affected by severe nonlinear impairments, including modulation related nonlinearities, square-law nonlinearity and mode-coupling-induced nonlinearity. In this paper, an equalizer based on a hidden conditional random field (HCRF) is proposed for the nonlinear mitigation of OAM-MDM optical fiber communication systems with 20 GBaud three-dimensional carrierless amplitude and phase modulation-64 (3D-CAP-64) signals. The HCRF equalizer extracts the stochastic nonlinear feature of the OAM-MDM 3D-CAP-64 signals by estimating the conditional probabilities of the hidden variables, thereby enabling the signals to be classified into subclasses of constellation points. The nonlinear impairment can then be mitigated based on the statistical probability distribution of the hidden variables of the OAM-MDM transmission channel in the HCRF equalizer. Our experimental results show that compared with a convolutional neural network (CNN)-based equalizer, the proposed HCRF equalizer improves the receiver sensitivity by 2 dB and 1 dB for the two OAM modes used here, with l = + 2 and l = + 3, respectively, at the 7% forward error correction (FEC) threshold. When compared with a Volterra nonlinear equalizer (VNE) and CNN-based equalizer, the computational complexity of the proposed HCRF equalizer was found to be reduced by 30% and 41%, respectively. The bit error ratio (BER) performance and reduction in computational complexity indicate that the proposed HCRF equalizer has great potential to mitigate nonlinear distortions in high-speed OAM-MDM fiber communication systems.

The arithmetic of triangular Z-numbers with reduced calculation complexity using an extension of triangular distribution

Information that people rely on is often uncertain and partially reliable. Zadeh introduced the concept of Z-numbers as a more adequate formal construct for describing uncertain and partially reliable information. Most existing applications of Z-numbers involve discrete ones due to the high complexity of calculating continuous ones. However, the continuous form is the most common form of information in the real world. Simplifying continuous Z-number calculations is significant for practical applications. There are two reasons for the complexity of continuous Z-number calculations: the use of normal distributions and the inconsistency between the meaning and definition of Z-numbers. In this paper, we extend the triangular distribution as the hidden probability density function of triangular Z-numbers. We add a new parameter to the triangular distribution to influence its convexity and concavity, and then expand the value's domain of the probability measure. Finally, we implement the basic operations of triangular Z-numbers based on the extended triangular distribution. The suggested method is illustrated with numerical examples, and we compare its computational complexity and the entropy (uncertainty) of the resulting Z-number to the traditional method. The comparison shows that our method has lower computational complexity, higher precision and lower uncertainty in the results.

ICpG-Pos: an accurate computational approach for identification of CpG sites using positional features on single-cell whole genome sequence data.

The investigation of DNA methylation can shed light on the processes underlying human well-being and help determine overall human health. However, insufficient coverage makes it challenging to implement single-stranded DNA methylation sequencing technologies, highlighting the need for an efficient prediction model. Models are required to create an understanding of the underlying biological systems and to project single-cell (methylated) data accurately. In this study, we developed positional features for predicting CpG sites. Positional characteristics of the sequence are derived using data from CpG regions and the separation between nearby CpG sites. Multiple optimized classifiers and different ensemble learning approaches are evaluated. The OPTUNA framework is used to optimize the algorithms. The CatBoost algorithm followed by the stacking algorithm outperformed existing DNA methylation identifiers. The data and methodologies used in this study are openly accessible to the research community. Researchers can access the positional features and algorithms used for predicting CpG site methylation patterns. To achieve superior performance, we employed the CatBoost algorithm followed by the stacking algorithm, which outperformed existing DNA methylation identifiers. The proposed iCpG-Pos approach utilizes only positional features, resulting in a substantial reduction in computational complexity compared to other known approaches for detecting CpG site methylation patterns. In conclusion, our study introduces a novel approach, iCpG-Pos, for predicting CpG site methylation patterns. By focusing on positional features, our model offers both accuracy and efficiency, making it a promising tool for advancing DNA methylation research and its applications in human health and well-being.

Time-Efficient Multiplex Orbital Angular Momentum Beam Synthesis With Arbitrary-Shaped Array Using Multipole Moment Expansion

Despite the significant interest in pattern synthesis over the decades, the beam synthesis of orbital angular momentum (OAM) has received relatively little attention. In this paper, we present a time-efficient method for OAM beam synthesis based on multipole moment expansion. The approach introduces the multipole moment term as a transforming level, which separates the excitation coefficients and radiation pattern. We evaluate the performance of our proposed method through several examples and demonstrate its high capability to synthesize multiplex OAM beams with arbitrary-shaped arrays, ensuring reasonable accuracy. The theoretical analysis shows a significant improvement in computational efficiency and a great reduction in computational complexity compared to conventional synthesis methods. Finally, we design, fabricate, and measure a double-ring concentric circular patch antenna array using our proposed method, and the measured results confirm the validity of our approach. The work provides a valuable contribution to the field of OAM beam synthesis and offers a promising avenue for future research.

A Novel Real-Time Algorithm for Optimizing Train Speed Profiles Under Complex Constraints

Energy-efficient speed profile optimization of train operations is challenging due to complex constraints including passage points and speed limits. Moreover, it is critical to perform real-time re-optimization of speed profiles within a short control period. To speed up the speed profile re-optimization with complex constraints, we introduce various train characteristics into the shortest path problem with time windows (SPPTW) from the literature, and formulate energy-efficient optimization as a train-based SPPTW (TRAIN-SPPTW). This reduces the solution space and leads to reduced computation complexity. To further reduce computational time for solving the Train-SPPTW, we take advantage of the preprocessing + query structure used in existing pulse algorithms (PA) on one hand, and further overcome their limitations in computation efficiency on the other hand by proposing a tour-adaptive partial-bounding pulse algorithm (TPPA) to shorten the total time of the preprocessing and query stages. We use a simulation study on the tracks of the Dongguan-Huizhou intercity railway in China to demonstrate that the proposed TPPA reduces the computational time by at least 60 % compared to three existing algorithms without sacrificing the quality of the solution.

Lightweight marine biological target detection algorithm based on YOLOv5

Underwater object detection currently faces many challenges, such as the large number of parameters in existing object detection models, slow inference speed, blurring of underwater images, and aggregation of small targets, making it difficult to conduct efficient underwater object detection. This paper proposes a lightweight underwater object detection algorithm based on YOLOv5.The method uses depth-wise separable convolution instead of ordinary convolution to reduce the number of parameters and computational complexity. A C3 module based on Ghost convolution is designed to further compress the model size and improve the computational speed. In the feature extraction stage, a RepVgg module based on structural reparameterization is used to convert the multi -branch structure into a single-branch structure in the inference stage, improving the feature extraction ability of the model and increasing the inference speed. A Rep-ECA module is designed to embed the efficient channel attention module ECANet into the RepVGG module, selecting more effective channel information and improving the model’s feature extraction ability for small objects in blurred images, thereby improving detection precision. Experimental results show that in the URPC underwater object detection dataset, the proposed algorithm has a 39% lower model parameter count compared to the original model, a 42% reduction in computational complexity. The model can achieve a frame rate of 85 on a single Nvidia GTX 1080ti GPU, which is a 24% improvement over the original model, while mAP reaches 85.1%, a 1.1% improvement over the original model. The algorithm can improve the detection precision and achieve lightweight, which lays a foundation for the deployment of underwater equipment.

Analysis of the Recursive Locally-Adaptive Filtration of 3D Tensor Images

This work is focused on the computational complexity (CC) reduction of the locally-adaptive processing of 3D tensor images, based on recursive approaches. As a basis, a local averaging operation is used, implemented as a sliding mean 3D filter (SM3DF) with a central symmetric 3D kernel. Symmetry plays a very important role in constructing the working window. The presented theoretical approach could be adopted in various algorithms for locally-adaptive processing, such as additive noise reduction, sharpness enhancement, texture segmentation, etc. The basic characteristics of the recursive SM3DF are analyzed, together with the main features of the adaptive algorithms for filtration of Gaussian noises and unsharp masking where the filter is aimed at. In the paper, the ability of SM3DF implementation through recursive sliding mean 1D filters, sequentially bonded together, is also introduced. The computational complexity of the algorithms is evaluated for the recursive and non-recursive mode. The recursive SM3DF also suits the 3D convolutional neural networks which comprise sliding locally-adaptive 3D filtration in their layers. As a result of the lower CC, a promising opportunity is opened for higher efficiency of the 3D image processing through tensor neural networks.

Computationally efficient direction of arrival estimation using adaptive grid selection

AbstractThe authors propose a computationally efficient approach to estimate the directions of arrival of far‐field sources impinging on a sensor array. The proposed estimator is formed using a sparse reconstruction framework, employing a novel adaptive grid selection technique to reduce the dimensionality of the used dictionary matrix. The method further makes use of a SPICE‐inspired dictionary to adaptively select an appropriate regularisation parameter and to select the active grid set. An adaptive stepsize selection strategy is also introduced to reduce computation complexity further. The proposed method allows for accurate direction‐of‐arrival (DOA) estimates using even a single snapshot, also from weak sources. Numerical simulations indicate that the method offers preferable performance in comparison to existing state‐of‐the‐art DOA estimation algorithms.

A deep reinforcement learning for energy efficient resource allocation Intelligent Reflecting Surface (IRS) driven Non-Orthogonal Multiple Access Beamforming (NOMA-BF)

This paper proposes a deep reinforcement learning centralized intelligent reflective surface (IRS) assisted non-orthogonal multiple access beamforming (NOMA-BF) system to improve the capacity gains and energy efficiency of next generation wireless communication networks. A single IRS is deployed at base station (BS) and consists of with R reflecting elements which is used to communicate with N single receiver antennas, referred to as centralized deployment of IRS- assisted NOMA. In the proposed central control scheme, we consider the practical situation where the instantaneous Channel State Information (CSI) of the channel between the IRS and the users is unknown. However, its long-term average is known and therefore makes the system more practical. Consequently, we exploit deep reinforcement technique learning technique to investigate the phase shift design and to tackle optimization problem. The energy efficiency problem is formulated as a non-convex optimization problem and is solved using deep and reinforcement learning algorithms. The first order Taylor approximation and difference of convex (DC) programming method using MATLAB tool box has been adopted to obtain the optimal values for power allocation coefficients. For interference mitigation, efficient user clustering along with beamforming algorithm is exploited. The NOMA user interference is tackled efficiently through the proposed algorithms that compute power allocation coefficients along with desirable phase shifts. The closed form expressions for achievable rates and SINR are derived under the required constraints. Simulation results demonstrate that proposed deep learning and reinforcement learning algorithms achieve improved performances in terms of sum rate, energy efficiency and coverage with lower computational complexity and symbol error rate. The computational complexity reduction has been proved both analytically and graphically and is in agreement.

Empirical validation of task-related component analysis reformulation for computational complexity reduction

The task-related component analysis (TRCA) is a data-driven method for extracting reproducible components across multiple data segments from multivariate data. TRCA has been proven effective in enhancing the signal-to-noise ratio of neuroimaging data in previous studies. However, its original form requires a computational cost of O(N2) to compute the sum of cross-covariance matrices, indicating that the computational time increases rapidly as the number of data segments N increases. This study aims to empirically validate that the reformulated form of TRCA can reduce its computational complexity. This study estimated that the reformulation can reduce the computational complexity from O(N2) to O(N) without distorting its outputs by reducing the number of matrix multiplications in computing the sum of cross-covariance matrices. A series of computer simulations using computer-generated synthetic, functional near-infrared spectroscopy (fNIRS), and electroencephalogram (EEG) data were conducted to verify the theoretical computation complexity and consistency of the outputs between the original and reformulated TRCA. The computation reduction was validated through an experimental comparison of the original and reformulated TRCA with synthetic and real data of various sizes (e.g., data segments, dimensions, and lengths). In addition, the two algorithms output almost exactly identical results with only floating-point errors. The reformulation would considerably benefit practical applications, especially when applying the TRCA to large-scale computations. The code is available at https://github.com/mnakanishi/TRCA-SSVEP.