Multiple-input multiple-output wireless sensing systems achieve high-resolution detection through spatial diversity. However, they suffer from reliability degradation under low signal-to-noise ratio (SNR) conditions. Conventional matrix-based methods may discard critical target information embedded in inter-dimensional signal correlations due to dimension-reduction flattening operations. To overcome these limitations, this paper proposes a tensorial target detection (TTD) framework combining noise reduction and enhanced detection specifically designed for high-dimensional processing. Firstly, we propose a two-stage tensorial noise reduction (TNR) method based on the minimum mean square error criterion and the alternating least square iteration strategy to remove noise in high-order signal space. We further identify the sub-optimal performance caused by inter-dimensional noise coupling at the first stage of TNR, and resolve the issue via rank-constrained optimization for noise-target subspace separation at the second stage of TNR. Then, we develop an augmented tensorial detector based on cross-shaped spatial partitioning (CSP) to enhance detection performance, which jointly optimizes detection thresholds by adaptively refining noise estimation. Finally, field measurements confirm the TTD framework's operational validity, while in simulation the TNR method achieves 5 dB SNR improvement over 2D method, and the CSP-based detector delivers a 21% enhancement in detection probability over conventional approaches.

Video anomaly detection (VAD) is of great importance for a variety of real-time applications in video surveillance. Most deep learning-based anomaly detection algorithms adopt a one-class learning scheme to train a classifier using only normal data to distinguish between normal and abnormal events during the test phase. However, these methods, whether they are reconstruction or prediction models, commonly face the challenge of the model’s overly strong representation capability, which leads to excessive fitting of abnormal events and thus limits the performance of the model in diverse scenarios. To address these challenges, this work develops a novel residual autoencoder and generative adversarial network-based 3-D convolutional network, called READ3D-net, for anomaly detection. An adaptive multimodal pseudoanomaly generator is developed to simulate and generate diverse pseudoanomalies, aiming to enhance the model’s ability to extract the features with regard to “abnormal” behaviors while reducing the interference of background on the detection performance. In addition, residual structures are incorporated into the design of a reconstruction-based autoencoder model to enhance its feature extraction and discriminative capabilities. To further improve the model’s reconstruction ability on normal data, a dynamic generative adversarial strategy is proposed for effective feature learning. Extensive experiments conducted on public benchmark datasets demonstrate that the proposed model is more competitive than the state-of-the-art methods in video anomaly detection tasks, fully validating the effectiveness and practicality of the proposed approach.

Sensing function in integrated sensing and communication (ISAC) system concentrates on collecting and extracting information of the targets from noisy observations, which can assist positioning the users and enable a precise directional communication link. This paper deals with noise reduction via tensor ring (TR) decomposition and total variation (TV) for linear frequency modulated continuous-wave (FMCW) signals in the multiple-input multiple-output ISAC system. Specifically, TR decomposition is used to exploit the low-rankness and describe the global correlation among different dimensions of the high-order received signal. The noise suppression is addressed by the integration of a TV regularization and a Frobenius norm term to ensure sufficient signal-to-noise ratio (SNR). The corresponding optimization problem is solved using augmented Lagrange multiplier (ALM) and proximal alternating minimization. Simulation results illustrate that the proposed method improves denoising performance, leading to a higher output SNR of the target and a better detection probability.

Abstract This paper investigates an advanced effective signal processing technique to suppress noise, addressing a modern high‐performance detection in the field of radar sensing. To achieve a higher accuracy, the frequency modulated continuous wave radar is taken as a case study to derive the algorithm based on Karhunen ‐ Loève transform (KLT) before detection. KLT defines a linear projection of the signal statistics on the eigenfunctions domain, which makes the input‐dependent signals orthogonal to each other under new eigen‐basis and eigenvalues. The highest energy along slow time dimension of each range bin is concentrated in the transformed domain corresponding to the largest N eigenvalues. The performance of the algorithm is evaluated by different eigenvalue selection strategies. Numerical experiments are employed to obtain the relationship between signal‐to‐noise ratio and different eigenvalue selection strategies. Pertaining to the detection performance, constant false alarm ratio detector is applied to demonstrate the detection ability as a result of the processor by use of probability of detection ( P d ).

In this paper, we investigate the quantization and the feedback of downlink spatial covariance matrix for massive multiple-input multiple-output (MIMO) systems with cascaded precoding. Massive MIMO has gained a lot of attention recently because of its ability to significantly improve the network performance. To reduce the overhead of downlink channel estimation and uplink feedback in frequency-division duplex massive MIMO systems, cascaded precoding has been used to convert high-dimensional physical channels into low-dimensional effective channels. For the cascaded precoding, the inner precoder is determined by the downlink spatial covariance matrix, which is unknown in the base station (BS). To address this issue, we propose a spatial spectrum-based approach for the quantization and the feedback of the spatial covariance matrix. In this manner, the BS can obtain partial information on the downlink spatial covariance matrix. Our result shows that the inner precoder based on the proposed approach can be viewed as modulated discrete prolate spheroidal sequences and thus achieves much smaller spatial leakage than the traditional discrete Fourier transform submatrix-based precoding. Practical issues for the application of the proposed approach are also addressed in this paper.

In this paper, we investigate the low-overhead feedback for statistical channel information in massive multiple-input multiple-output (MIMO) systems. As a promising technique for next generation cellular networks, massive MIMO has gained a lot of attention due to its ability to significantly improve the network performance. To reduce the feedback overhead in frequency division duplex systems, cascaded precoding has been proposed for downlink transmission, where the outer precoder with reduced dimension is implemented through traditional limited feedback while the inner precoder is determined by the channel spatial covariance matrix. The size of the covariance matrix in massive MIMO is very large due to the huge number of antennas. In this paper, we propose a spectrum quantization approach to achieve efficient quantization of the covariance matrix with low- overhead feedback. In this approach, the quantization of the covariance matrix is conducted with respect to the corresponding spatial spectrum to avoid the complicated matrix calculation. Simulation results show that our approach can reduce the feedback overhead significantly at the cost of negligible performance degradation.

Large-scale antenna (LSA) has gained a lot of attention recently since it can significantly improve the performance of wireless systems. Similar to multiple-input multiple-output (MIMO) orthogonal frequency division multiplexing (OFDM) or MIMO-OFDM, LSA can be also combined with OFDM to deal with frequency selectivity in wireless channels. However, such combination suffers from substantially increased complexity proportional to the number of antennas in LSA systems. In this paper, we propose a low-complexity recursive convolutional pre-coding to address the issues above. The traditional ZF precoding is implemented through the recursive convolutional precoding in the time domain so that only one IFFT is required for each user and the matrix inversion can be also avoided. Simulation results show that the proposed approach can achieve the same performance as that of ZF but with much lower complexity.

In this paper, we study cooperative precoder design in two-tier networks, consisting of a macro-cell (MC) and several small-cells (SCs). By exploiting multiuser Vandermonde-subspace frequency division multiplexing (VFDM) transmission, an MC downlink can co-exist with cognitive SCs. In this paper, we first propose a cooperative cross-tier precoder (CTP) among the transmitters in the SCs to increase the transmitted dimension. The cooperative CTP allows us to use more efficient intra-tier precoder (ITP) in SCs to handle intracell interference and improve the throughput of the cognitive system. And then, three ITPs, a block-diagonal zero-forcing (BD-ZF) ITP, a capacity-achieving (CA) ITP, and a generalized MMSE channel inversion (GMI) ITP, are developed. Complexities of all CTPs and ITPs are discussed and compared. The overhead of channel state information (CSI) exchange is analyzed. Numerical results are presented to demonstrate the throughput improvement of the proposed schemes and to discover the impact of the imperfect CSI. From the complexity comparison and the numerical results, the GMI ITP offers a good tradeoff between complexity and throughput.

Large-scale antenna (LSA) has gained a lot of attention recently since it can significantly improve the performance of wireless systems. Similar to multiple-input multiple-output (MIMO) orthogonal frequency division multiplexing (OFDM) or MIMO-OFDM, LSA can be also combined with OFDM to deal with frequency selectivity in wireless channels. However, such combination suffers from substantially increased complexity proportional to the number of antennas in LSA systems. For the conventional implementation of LSA-OFDM, the number of inverse fast Fourier transforms (IFFTs) increases with the antenna number since each antenna requires an IFFT for OFDM modulation. Furthermore, zero-forcing (ZF) precoding is required in LSA systems to support more users, and the required matrix inversion leads to a huge computational burden. In this paper, we propose a low-complexity recursive convolutional precoding to address the issues above. The traditional ZF precoding can be implemented through the recursive convolutional precoding in the time domain so that only one IFFT is required for each user and the matrix inversion can be also avoided. Simulation results show that the proposed approach can achieve the same performance as that of ZF but with much lower complexity.

This study deals with channel parameter based channel estimation in time‐varying channels for orthogonal frequency division multiplexing in an uplink transmission. By modelling the uplink channel properly, the time‐varying ‘channel response’ can be determined by estimating the corresponding ‘channel parameters’. A novel algorithm is proposed in this study to estimate the channel parameters by exploiting the time‐frequency‐representation of the time‐varying channel response. Moreover, a decision‐directed receiver structure is adopted to cancel the impact of residual Doppler shifts. Theoretical analysis on the performance is given as well. The optimum power allocation ratio of training signal power to data signal power is also discussed. Simulations are performed, demonstrating the effectiveness of the proposed algorithm.

Iterative channel estimation (ICE) usually exploits soft information of unknown data symbols as references to improve estimation performance. This paper investigates ICE for orthogonal frequency division multiplexing (OFDM) over wireless channels. The optimum ICE is derived in terms of maximum a posteriori (MAP) criterion, which can be solved using fixed-point iteration (FPI). Furthermore, the derived MAP ICE is closely related to the well-known expectation-maximization (EM) estimation. We also demonstrate that the MAP ICE converges within only one step when the signal-to-noise ratio (SNR) is large through analysis and simulation results.

This paper investigates iterative channel estimation (ICE) for orthogonal frequency-division multiplexing (OFDM) with multiple transmit antennas. To improve performance of channel estimation, we exploit the soft information of unknown data symbols on both the expected transmit antenna and the interfering transmit antenna. Maximum a posteriori (MAP)-based ICE is derived and is implemented using the fixed-point iteration (FPI). For an OFDM system with multiple transmit antennas, the proposed MAP-based ICE suggests a harmonic-average-based soft symbol on the expected transmit antenna while an arithmetic-average-based soft symbol on the interfering transmit antennas. Similar to an OFDM system with a single transmit antenna, MAP-based ICE can achieve the Cramer-Rao bound (CRB) within only one iteration, when the signal-to-noise ratio (SNR) is large enough.

A timing synchronisation method for orthogonal frequency division multiplexing (OFDM) using improved cross-correlation (ICC) is proposed in this study. The proposed method is derived in terms of maximum-a-posteriori (MAP) criterion over an additive white Gaussian noise (AWGN) channel. In this way, the proposed ICC method is able to take into consideration the surrounded data signals, which is not considered in the conventional cross-correlation (CCC) method and thus the timing accuracy is improved accordingly. Compared to the CCC method, the ICC method only appends a correction term to the cross-correlator output, which can be easily implemented using an iterative method. Therefore the complexity of the proposed ICC method has no significant increase. Simulation results show that the proposed ICC method significantly outperforms the CCC based timing method.