Semi-Supervised Few-Shot Learning (SSFSL) aims to address the data scarcity in few-shot learning by leveraging both a few labeled support data and abundant unlabeled data. In SSFSL, a classifier trained on scarce support data is often biased and thus assigns inaccurate pseudo-labels to the unlabeled data, which will mislead downstream learning tasks. To combat this issue, we introduce a novel method called Certainty-Aware Recursive Confidence Training (CARCT). CARCT hinges on the insight that selecting pseudo-labeled data based on confidence levels can yield more informative support data, which is crucial for retraining an unbiased classifier to achieve accurate pseudo-labeling-a process we term pseudo-labeling calibration. We observe that accurate pseudo-labels typically exhibit smaller certainty entropy, indicating high-confidence pseudo-labeling compared to those of inaccurate pseudo-labels. Accordingly, CARCT constructs a joint double-Gaussian model to fit the certainty entropies collected across numerous SSFSL tasks. Thereby, A semi-supervised Prior Confidence Distribution (ssPCD) is learned to aid in distinguishing between high-confidence and low-confidence pseudo-labels. During an SSFSL task, ssPCD guides the selection of both high-confidence and low-confidence pseudo-labeled data to retrain the classifier that then assigns more accurate pseudo-labels to the low-confidence pseudo-labeled data. Such recursive confidence training continues until the low-confidence ones are exhausted, terminating the pseudo-labeling calibration. The unlabeled data all receive accurate pseudo-labels to expand the few support data to generalize the downstream learning task, which in return meta-refines the classifier, named self-training, to boost the pseudo-labeling in subsequent tasks. Extensive experiments on basic and extended SSFSL setups showcase the superiority of CARCT versus state-of-the-art methods, and comprehensive ablation studies and visualizations justify our insight. The source code is available at https://github.com/Klein-JING/CARCT.

This work presents a novel K-Repetition based Hybrid Automatic Repeat reQuest (HARQ) scheme for uplink sparse code multiple access (SCMA) systems. Our core idea is to apply network coding (NC) principle to re-encode different packets (after channel coding and interleaving) or their fragments, where K-Repetition is an emerging HARQ technique (recommended in 3GPP Release 15) for enhanced reception in future massive machine-type communications. Such a proposed scheme is referred to as the NC aided K-repetition SCMA (NCK-SCMA) in this paper. We aim to understand the optimal NCK-SCMA design criteria for maximizing the channel diversity as well as the efficient receiver processing for superior error rate performances. It is found that NC can enable a larger diversity order for NCK-SCMA with fewer resources (i.e., higher spectrum efficiency). Toward this objective, some novel design criteria are developed for the efficient configuration of NCK-SCMA. Moreover, we propose an iterative network decoding and SCMA detection (INDSD) algorithm for robust and low-complexity recovery of the transmit data from a low-density parity-check (LDPC) coded uplink NCK-SCMA system. Simulation results demonstrate that the proposed NCK-SCMA lead to higher throughput and improved reliability over the conventional K-SCMA.

Grant-free sparse code multiple access (GF-SCMA) is considered to be a promising multiple access candidate for future wireless networks. In this article, we focus on characterizing the performance of uplink GF-SCMA schemes in a network with ubiquitous connections, such as the Internet-of-Things (IoT) networks. To provide a tractable approach to evaluate the performance of GF-SCMA, we first develop a theoretical model taking into account the property of multiuser detection (MUD) in the SCMA system. Then, the error rate performance of GF-SCMA in the case of codebook collision is analyzed to investigate the reliability of GF-SCMA when reusing codebook in massive IoT networks. For performance evaluation, accurate approximations for both success probability and average symbol error probability (ASEP) are derived. To elaborate further, the analytical results are utilized to discuss the impact of codeword sparse degree in GF-SCMA. After that, we conduct a comparative study between SCMA and its variant, dense code multiple access (DCMA), with GF transmission to offer insights into the effectiveness of these two schemes. This facilitates the GF-SCMA system design in practical implementation. Simulation results show that denser codebooks can help to support more user equipments (UEs) and increase the reliability of data transmission in a GF-SCMA network. Moreover, a higher success probability can be achieved by GF-SCMA with denser UE deployment at low detection thresholds since SCMA can achieve overloading gain.

Abstract Multiple‐input multiple‐output‐sparse code multiple access, a non‐trivial integration of sparse code multiple access and multiple‐input multiple‐output techniques, is able to achieve high spectrum efficiency and massive user connections. However, this integration also increases the complexity of signal detection. Here, the signal detection problem of multiple‐input multiple‐output sparse code multiple access is transformed into a tree search problem and use sphere decoding to detect the signal. By setting the initial radius to positive infinity, sphere decoding can achieve optimal maximum likelihood performance while the complexity is high. In order to further reduce the complexity of sphere decoding, a block‐wise sorted QR decomposition algorithm is proposed. Based on block‐wise sorted QR decomposition, the improved sphere decoding, namely block‐wise sorted QR decomposition‐sphere decoding, is able to make the tree search more efficient. Since only the detection order of each user's signal has been changed, block‐wise sorted QR decomposition‐sphere decoding can maintain the optimal maximum likelihood performance. Simulation results and complexity analysis show that block‐wise sorted QR decomposition‐sphere decoding can achieve optimal performance and both hard‐output and soft‐output block‐wise sorted QR decomposition‐sphere decoding have much lower complexity than joint message passing algorithm. Furthermore, given the same signal‐to‐noise ratio, the complexity of block‐wise sorted QR decomposition‐sphere decoding decreases with the increase of receiving antennas, while the complexity of joint message passing algorithm increases linearly.

Sparse code multiple access (SCMA) is one of the non‐orthogonal multiple access schemes to meet the demands of massive connection and high spectral efficiency in the 5G communications. Since the sparsity of the codebook, the message passing algorithm (MPA) can be utilised for multi‐user detection. However, the high computational complexity of MPA hinders the practical application of SCMA. The extended Max‐Log MPA (EML‐MPA) and the channel‐adaptive extended Max‐Log MPA (CAD‐EML‐MPA) help to reduce the complexity of conventional Log‐MPA with trellis representation. On this basis, the authors propose a dynamic trellis based MPA (DT‐MPA) in this study. DT‐MPA selects symbols dynamically according to a specific threshold in each iteration of MPA. The impact of the threshold on the complexity and convergence behaviour is analysed. Simulation results show that the DT‐MPA with variable which is related to bit signal‐to‐noise can achieve better bit error rate performance and lower complexity than EML‐MPA and CAD‐EML‐MPA.

In this study, a multiple access scheme called low density signature isotropic orthogonal transfer algorithm (LDS‐IOTA) is proposed for overloading transmissions, and an analytical model for the performance limitations of this system is developed. LDS‐IOTA utilises a low density signature structure for spreading the symbols in frequency domain, and applies such technique to IOTA‐based filter bank multicarrier system. The input–output mutual information of LDS‐IOTA is analysed for the case of finite size modulation, and LDS‐IOTA is compared with OFDM as well as LDS‐OFDM over multipath fadding channels.

Sparse Code Multiple Access (SCMA) is a novel non- orthogonal multiple access technique for 5G systems. In this paper, we propose a suboptimal algorithm for SCMA codebook design over uplink Rayleigh fading channels. In uplink channels, the pairwise error probability (PEP) is derived firstly, based on which the diversity gain and the coding gain of SCMA are analyzed. According to the analysis, the multi-dimensional mother constellation design is formulated as a non-convex optimization problem, aiming to enlarge two factors to achieve a better coding gain. Furthermore, the optimization problem is simplified as a convex second-order cone programming problem to obtain a suboptimal solution. Simulation results show that the bit error ratio (BER) performance is improved significantly on account of the enlarged factors.

Sparse code multiple access (SCMA) is a promising candidate air interface of the next generation mobile networks. However, the decoding complexity of current message passing algorithm for SCMA is very high. In this study, the authors map SCMA constellation to q ‐order Galois field ( ) and introduce a trellis representation to SCMA. Based on the trellis representation, they propose low‐complexity decoding algorithms for SCMA by using truncated messages, which is referred as extended max‐log (EML) algorithm. As the truncated length of EML is unitary for each user, they further propose a channel‐adaptive EMLalgorithm to truncate the messages with a rule that can be adaptive to the channel state. Simulation results show that the proposed schemes obtain a low computational complexity with only a slight performance degradation when the truncated length is selected appropriately.

Sparse code multiple access (SCMA) is a novel non‐orthogonal multiple access scheme for 5G systems, in which the logarithm domain message passing algorithm (Log‐MPA) is applied at the receiver to achieve near‐optimum performance. However, the computational complexity of Log‐MPA detector is still a big challenge for practical implementation, especially for energy‐sensitive user equipments in the downlink scenario. A region‐restricted detector with an improved Log‐MPA (RRL detector) is proposed for downlink SCMA systems, in which the complexity is reduced from two perspectives. To avoid unnecessary calculations when searching the superposition constellation exhaustively, the proposed RRL detector updates the function nodes only within a restricted search region. While constellation points outside the search region are neglected, the performance is well maintained which is verified by simulations. Besides, the original Log‐MPA heavily relies on exponential operations, resulting in high computational complexity. To solve this problem, an improved Log‐MPA is also put forward in this study to make a better compromise between complexity and performance. Simulation results show that the complexity of the RRL detector is reduced considerably while the bit error rate performance degrades unnoticeably.

In the Low Density Signature (LDS) based multiple access systems, the constellation of chip symbols superposed by several users' symbols is named as superposition constellation. When LDS is applied into the downlink multiple access scenario, maximizing the minimum Euclidean distance of the superposition constellation will be a big challenge. In this paper, by optimizing the superposition constellation as square QAM, we propose a novel LDS matrix construction algorithm. Besides, the original Message Passing Algorithm (MPA) detector for LDS system leads to high complexity which is a big problem for energy-sensitive user equipment. We propose a low complexity region-restricted MPA detector by restricting the search region to a quadrant-like part of the superposition constellation. The complexity of the new detector is reduced to 25%~50% while performance in terms of bit error rate (BER) maintains almost the same. The theoretical analysis and simulation results further verify the effectiveness of our proposed algorithms.

Radar‐embedded communication (REC) provides a reliable and low probability of intercept (LPI) communication link to the intended receivers while masking its presence from intercept receivers. This study summarises all the waveform design schemes as well as receiver processing strategies for intra‐pulse REC, which forms the communication symbol waveform during one pulse interval and utilises de‐correlation processing on the receiver side. Based on the set of eigenvectors after eigen‐decomposition of sampled radar waveform, three design approaches are considered: eigenvectors‐as‐waveforms, weighted‐combining and dominant‐projection (DP). Specifically, DP performs better than the other two approaches even in multipath environment by using one projection matrix masked random vector to represent a communication symbol. Aimed by the drawbacks of DP, the authors propose a novel waveform generation approach called modified DP (MDP) in this study. MDP uses a common digitally‐modulated communication symbol spread by a pseudorandom sequence, so that it can enlarge the Euclidean distances between different data symbols and generate soft information for soft decision decoding, thus improving symbol error ratio and bit error ratio. Simulation results show that MDP significantly improves the reliability performance especially when channel coding is used, while also maintaining a relatively equal anti‐intercept performance.

Joint sparse graph for code division multiple access systems (JSG‐CDMA) combines multiple accessing (low‐density signature) and forward error correction (low‐density parity check code) techniques, and it achieves satisfactory performance under overloaded conditions. In this study, the authors carry on the research on the joint detection and decoding for JSG‐CDMA, and analyse the syndrome effect of message passing on the JSG by the extrinsic information transfer (EXIT) chart. An improved algorithm for joint detection and decoding on the JSG‐CDMA receiver is proposed, and the convergence behaviour of the algorithm is analysed by EXIT chart. Simulation results show that improved algorithm enhance the system performance significantly with a marginal increase of the computational complexity.

Abstract Low density signature for orthogonal frequency division multiplexing (LDS‐OFDM) is a promising technique for multiple accesses in wireless communications, but its performance is not optimal. In this paper, we study a non‐binary joint sparse graph which combines LDS‐OFDM and non‐binary low density parity check (LDPC) codes, namely non‐binary joint sparse graph for OFDM (NJSG–OFDM). Considering the fact that syndrome computing can be utilized to improve the convergence rate of the joint detection and decoding, syndrome nodes are added to the non‐binary joint sparse graph; consequently, an optimized receiver for the NJSG–OFDM is proposed and analyzed by extrinsic information transfer (EXIT) chart. Simulation results show that the optimized receiver has several advantages such as accelerating the convergence rate of the message passing, improving system performance and alleviating near‐far effect. Copyright © 2017 John Wiley & Sons, Ltd.

Low‐density signature code division multiple access (LDS‐CDMA) and low‐density parity‐check (LDPC) code can both be represented by a single sparse graph. In this study, the authors propose a joint sparse graph (JSG) over GF( q ) which combines LDS‐CDMA and non‐binary LDPC codes, namely JSG‐CDMA. On the JSG, multiple accessing and channel coding are well‐linked together, in addition, joint detection and decoding are performed by message passing algorithm. Two schedules for message updating on the JSG, that is, flooding and serial schedules, are, respectively, presented. To predict the convergence behaviour of the joint detection and decoding, they depict the iterative structure of the JSG‐CDMA receiver and analyse its extrinsic information transfer chart. Key factors of JSG‐CDMA include message passing schedule, maximum iteration number and Galois field order. Simulation results show that JSG‐CDMA performs much better than conventional CDMA, in addition, compared with LDS‐CDMA and turbo structured LDS‐CDMA, at a bit error rate of 3 × 10 −4 , JSG‐CDMA brings about 1.9 and 1.1 dB gain, respectively.

We present an automatic statistical intensity-based approach to extract the 3D cerebrovascular structure from time-of flight (TOF) magnetic resonance angiography (MRA) data. We use the finite mixture model (FMM) to fit the intensity histogram of the brain image sequence, where the cerebral vascular structure is modeled by a Gaussian distribution function and the other low intensity tissues are modeled by Gaussian and Rayleigh distribution functions. To estimate the parameters of the FMM, we propose an improved particle swarm optimization (PSO) algorithm, which has a disturbing term in speeding updating the formula of PSO to ensure its convergence. We also use the ring shape topology of the particles neighborhood to improve the performance of the algorithm. Computational results on 34 test data show that the proposed method provides accurate segmentation, especially for those blood vessels of small sizes.

Low density signature orthogonal frequency division multiplexing (LDS-OFDM) and low density parity-check (LDPC) codes are multiple access and forward error correction (FEC) techniques, respectively. Both of them can be expressed by a bipartite graph. In this paper, we construct a joint sparse graph combining the single graphs of LDS-OFDM and LDPC codes, namely joint sparse graph for OFDM (JSG-OFDM). Based on the graph model, a low complexity approach for joint multiuser detection and FEC decoding (JMUDD) is presented. The iterative structure of JSG-OFDM receiver is illustrated and its extrinsic information transfer (EXIT) chart is researched. Furthermore, design guidelines for the joint sparse graph are derived through the EXIT chart analysis. By offline optimization of the joint sparse graph, numerical results show that the JSG-OFDM brings about 1.5-1.8 dB performance improvement at bit error rate (BER) of 10-5over similar well-known systems such as group-orthogonal multi-carrier code division multiple access (GO-MC-CDMA), LDS-OFDM, and turbo structured LDS-OFDM.

Low density signature for OFDM (LDS-OFDM) is able to achieve satisfactory performance in overloaded conditions, but the existing LDS-OFDM has the drawback of slow convergence rate for multiuser detection (MUD) and high receiver complexity. To tackle these problems, we propose a serial schedule for the iterative MUD. By doing so, the convergence rate of MUD is accelerated and the detection iterations can be decreased. Furthermore, in order to exploit the similar sparse structure of LDS-OFDM and LDPC code, we utilize LDPC codes for LDS-OFDM system. Simulations show that compared with existing LDS-OFDM, the LDPC code improves the system performance.

UNLABELLED: BlastGraph is an interactive Java program for comparative genome analysis based on Basic Local Alignment Search Tool (BLAST), graph clustering and data visualization. The software generates clusters of sequences of multiple genomes from all-to-all BLAST results and visualizes the results in graph plots together with related information such as sequence features, gene conservation and similarity relationships. Pruning algorithms are used to reduce results to more meaningful subclusters. Subsequent analyses can then be conducted based on the predicted clusters, including gene content, genome phylogenetics and gene gain and loss. AVAILABILITY AND IMPLEMENTATION: https://github.com/bigwiv/BlastGraph.