In modern wireless systems, the process of acquiring channel state information (CSI) has become increasingly important and challenging due to the conflicting needs of expanding antenna arrays and achieving high spectrum efficiency. Recent solutions leverage joint source and channel coding (JSCC) realized by an auto-encoder structure to enable a hybrid optimization of CSI compression and transmission, which shows great potential to improve the effectiveness of CSI feedback. However, the gap between research and realization is still big. This article proposes a comprehensive JSCC method for CSI feedback, including a technique solution for eliminating the impacts of residential fading and radio frequency ($\mathbf{R F}$) imperfections, and a protocol design to support the realization of the method in practical cellular systems. Furthermore, a hardware proof-ofconcept (PoC) platform was developed to validate the JSCC-based CSI feedback, and the results demonstrate the superiority of the method in realistic environments.

MOTIVATION: The classification of high-throughput protein data based on mass spectrometry (MS) is of great practical significance in medical diagnosis. Generally, MS data are characterized by high dimension, which inevitably leads to prohibitive cost of computation. To solve this problem, one-bit compressed sensing (CS), which is an extreme case of quantized CS, has been employed on MS data to select important features with low dimension. Though enjoying remarkably reduction of computation complexity, the current one-bit CS method does not consider the unavoidable noise contained in MS dataset, and does not exploit the inherent structure of the underlying MS data. RESULTS: We propose two feature selection (FS) methods based on one-bit CS to deal with the noise and the underlying block-sparsity features, respectively. In the first method, the FS problem is modeled as a perturbed one-bit CS problem, where the perturbation represents the noise in MS data. By iterating between perturbation refinement and FS, this method selects the significant features from noisy data. The second method formulates the problem as a perturbed one-bit block CS problem and selects the features block by block. Such block extraction is due to the fact that the significant features in the first method usually cluster in groups. Experiments show that, the two proposed methods have better classification performance for real MS data when compared with the existing method, and the second one outperforms the first one. AVAILABILITY AND IMPLEMENTATION: The source code of our methods is available at: https://github.com/tianyan8023/OBCS. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Grant-free non-orthogonal multiple access (NOMA) is considered as one of the most important methodologies for the machine-type communications (MTC). In the field of MTC, compressed sensing based multi-user detection (CS-MUD) has been recognized as an excellent candidate for joint user activity and data detection, since many users sporadically transmit short-size data packets at low rates. This article focuses on the CS-MUD problem in the up-link grant-free NOMA scenario, where users are (in)-active randomly in each time slot yet with high temporal correlation. First, we investigate the CS framework to fully extract the underlying side information in the temporal correlation and propose a novel CS-MUD algorithm. Then, to mitigate the performance degradation due to the imperfect channel estimation in practice, the proposed algorithm is further extended by utilizing the perturbed CS, where the impact of channel estimation errors is modeled as certain perturbation in the measurement matrix. Different from most of the state-of-the-art CS-MUD algorithms, both proposed algorithms can apply even in the absence of prior knowledge on the number of active users. Simulation results indicate that the proposed algorithms achieve better performance than the existing CS-MUD methods. Their convergence and complexity issues are also discussed theoretically and numerically.

Spectrum sensing is an important topic in cognitive radio networks. The traditional schemes suffer from high time consumption, hardware loss and computational complexity. To overcome the above shortcomings, compressed sensing is integrated into cognitive radio networks. In this paper, we propose two compressed spectrum sensing algorithms called Logit Weighted Block Orthogonal Matching Pursuit (LW-BOMP) and Logit Weighted Block Orthogonal Matching Pursuit with Joint Judgement (LW-BOMP-J), which exploit the block sparsity and prior information. The first algorithm integrates support probabilities into the matching pursuit procedure, and the second algorithm extends the first one to the scenario with inaccurate block support probabilities. The performance of the algorithms is simulated and compared with conventional algorithm. The results show that the proposed algorithms are more promising in reconstructing original spectrum and LW-BOMP-J is better than LW-BOMP under the inaccurate prior information condition.

Compressed sensing (CS) has attracted a lot of research interests in data transmission, since it can significantly reduce the redundancy of the data while still holding the information completely. Generally, in order to deal with the experienced time-varying channel, the data transmission scheme based on CS needs to estimate the channel frequently. In this paper, we propose a novel CS-based data transmission scheme for slowly time-varying channel, which estimates channel and reconstructs data jointly. Specifically, the scheme consists of a data reconstruction part and a channel state information (CSI) update part. The former reconstructs the data with inaccurate CSI by modeling the problem as a perturbed CS reconstruction problem. The latter updates the CSI with the reconstruction result as a semi-blind channel estimation problem. Comparing with the classical scheme, the proposed one enjoys a reduced cost by avoiding inserting pilots into data frame frequently. Simulation results demonstrate that our proposed scheme works robustly in slowly time-varying channel and the performance is comparable to that of the scheme reconstructing the data with perfect CSI.

Analog-to-information conversion (AIC) is an efficient way to obtain the compressed samples directly from an analog signal. Modulated wideband converter (MWC) proposed by M. Mishali and Y. C. Elda is a successful hardware architecture of AIC that measures the analog signal by K branches of mixers and integrators (BMI). However, MWC has high implementation complexity owing to the fact that its measurement matrix is dense. To reduce the complexity of MWC, in this paper we propose the diagonal modulated wideband converter (D-MWC), where each BMI only works within a partial time period that is non-overlapping instead of each BMI working all the time in MWC. The measurement matrix of D-MWC is sparse that can significantly simplify the sampling structure. Simulations verify the effectiveness of D-MWC. The proposed D-MWC offers a good tradeoff between complexity and sampling performance.