With the rapid development of deep learning (DL), the recent trend of log-based anomaly detection focuses on extracting semantic information from log events (i.e., templates of log messages) and designing more advanced DL models for anomaly detection. Indeed, the effectiveness of log-based anomaly detection can be improved, but these DL-based techniques further suffer from the limitations of more heavy dependency on training data (such as data quality or data labels) and higher costs in time and resources due to the complexity and scale of DL models, which hinder their practical use. On the contrary, the techniques based on traditional machine learning or data mining algorithms are less dependent on training data and more efficient but produce worse effectiveness than DL-based techniques, which is mainly caused by the problem of unseen log events (some log events in incoming log messages are unseen in training data) confirmed by our motivating study. Intuitively, if we can improve the effectiveness of traditional techniques to be comparable with advanced DL-based techniques, then log-based anomaly detection can be more practical. Indeed, an existing study in the other area (i.e., linking questions posted on Stack Overflow) has pointed out that traditional techniques with some optimizations can indeed achieve comparable effectiveness with the state-of-the-art DL-based technique, indicating the feasibility of enhancing traditional log-based anomaly detection techniques to some degree. Inspired by the idea of “try-with-simpler,” we conducted the first empirical study to explore the potential of improving traditional techniques for more practical log-based anomaly detection. In this work, we optimized the traditional unsupervised PCA (Principal Component Analysis) technique by incorporating a lightweight semantic-based log representation in it, called SemPCA , and conducted an extensive study to investigate the potential of SemPCA for more practical log-based anomaly detection. By comparing seven log-based anomaly detection techniques (including four DL-based techniques, two traditional techniques, and SemPCA ) on both public and industrial datasets, our results show that SemPCA achieves comparable effectiveness as advanced supervised/semi-supervised DL-based techniques while being much more stable under insufficient training data and more efficient, demonstrating that the traditional technique can still excel after small but useful adaptation.

In this paper, we concentrate on one algebraic-based quasi-cyclic low-density parity-check (QC-LDPC) code constructed from two subsets of a finite field and generalize it to propose a class of variable-rate QC-LDPC (VR-QC-LDPC) codes, whose parity-check matrices are nested horizontally and have constant number of rows. Thus the proposed codes are significant at least in terms of storage complexity and can be simply implemented. The constructed codes also inherit the original algebraic-based QC-LDPC codes and their exponent matrices can be obtained from two subsets of the given finite field. We hereby analyze the structural properties from the isomorphism perspective, and present some rules to significantly prune the size of search space and determine the non-isomorphic exponent matrices. By distinguishing the smaller quantities of non-isomorphic matrices with cycle property metric, we can easily construct a series of nested exponent matrices with better cycle distributions and obtain the VR-QC-LDPC codes. Numerical results demonstrate that the constructed codes have better iterative decoding performance within a range of code rates and decoding iterations.

To provide reliable transmissions with flexible rates for satellite communications, this paper presents a novel method of constructing multi-rate quasi-cyclic low-density parity-check (LDPC) codes. The basic idea is to generate low-rate codes from a high-rate mother code by combining shortening and extending, which ensures that the generated code family owns the same code length, in order to maintain the same frame structure. The code construction involves the design of base matrices and exponent matrices for the designed codes. A progressive row elimination and addition algorithm is proposed for designing the code base matrices from a high rate to low rates. This algorithm leads to the nested and systematic structure of the parity-check matrices, which are desirable for practical implementations of their encoders and decoders, while ensuring the optimal decoding thresholds. In addition, we construct a circulation coefficient matrix based on finite fields and select the optimal rows in this matrix to construct exponent matrices while considering of cycle structures. We demonstrate that the designed codes achieve better performance for all the code rates than the LDPC codes in DVB-S2X standards. In addition, the proposed codes do not exhibit error floors for their block error rates down to 10−5.

This paper presents a new type of globally-coupled low-density parity-check (GC-LDPC) codes whose base matrix has a cyclic structure in the global part. Therefore, the resulting codes are referred to as tail-biting GC-LDPC (TB-GC-LDPC) codes. We propose two methods to construct TB GC quasi-cyclic LDPC (TB-GC-QC-LDPC) codes. For the first method, we extract a replicated version of a constructed base matrix and mask it with a designed masking matrix. Compared to the conventional construction methods, this method provides more flexibility in code length for TB-GC-QC-LDPC codes. The second method is based on designing the incidence matrix of a special type of packings. Examples show that the constructed TB-GC-QC-LDPC codes perform well over the additive white Gaussian noise channel (AWGNC) and the binary erasure channel (BEC). The asymptotic performance of TB-GC-LDPC ensembles over BECs are also analyzed by resorting to density evolution. Moreover, numerical results show that TB-GC-LDPC ensembles can achieve better flooding-schedule decoding (FSD) thresholds than the corresponding GC-LDPC ensembles with a similar structure. With sufficient decoding iterations in the global phase, the local/global two-phase iterative decoding (TPD) thresholds of the TB-GC-LDPC ensembles significantly outperform those of the corresponding GC-LDPC ensembles as well.

This paper presents a family of rate‐compatible (RC) globally coupled low‐density parity‐check (GC‐LDPC) codes, which is constructed by combining algebraic construction method and graph extension. Specifically, the highest rate code is constructed using the algebraic method and the codes of lower rates are formed by successively extending the graph of the higher rate codes. The proposed rate‐compatible codes provide more flexibility in code rate and guarantee the structural property of algebraic construction. It is confirmed, by numerical simulations over the AWGN channel, that the proposed codes have better performances than their counterpart GC‐LDPC codes formed by the classical method and exhibit an approximately uniform gap to the capacity over a wide range of rates. Furthermore, a modified two-phase local/global iterative decoding scheme for GC‐LDPC codes is proposed. Numerical results show that the proposed decoding scheme can reduce the unnecessary cost of local decoder at low and moderate SNRs, without any increase in the number of decoding iterations in the global decoder at high SNRs.

In this paper, we investigate feedback in EWF codes. The considered type of feedback is an acknowledgment, which is a single bit per source block to inform the most important bits have been decoded completely. Moreover, we propose a practical prediction model to calculate the average overheads of successful decoding for different importance classes of the source block. Asymptotic analysis and simulation results demonstrate that, compared with standard EWF codes, the proposed EWF codes with feedback can obtain a lower average overhead of successful decoding in lower prioritized data, and the proposed prediction model is effective.

Novel precoded Expanding Window Fountain (EWF) codes, based on low rate low-density parity-check (LDPC) codes, are investigated for finite message length over binary erasure channels (BECs). The proposed codes are obtained by utilizing low rate LDPC codes as precodes and degree distributions with high intermediate symbol recovery rate for EWF codes. Simulation results show that, compared with conventional precoded EWF codes, our proposed codes have better performance in the scenarios of small and moderate message lengths over the BEC.

In this paper, we propose an efficient scheme to construct fountain codes that can provide unequal error protection (UEP) property. To reduce the coding complexity and improve the performance of previous duplicate UEP schemes in the literature, we implement different rate low-density parity-check codes instead of the duplication process and replace the high-complexity classical robust soliton distribution by low constant average degree distributions with high intermediate symbol recovery rates. Asymptotic analysis of both the duplicate scheme and our scheme is obtained by deriving unequal density evolution formulas over the binary erasure channel. Compared to previous UEP schemes, our scheme has lower complexity. Experimental results show that our scheme can obtain better decoding performance, especially for smaller input information length. Comparison of peak signal-noise ratio performance for the scalable video coding shows that, for moderate input information length, our scheme can provide a better basic video quality at lower overhead, but needs larger overhead to achieve a high video quality.

In this paper, we propose an efficient scheme to construct LT codes that can provide unequal error protection (UEP) property. We implement low rate LDPC codes instead of the duplication and utilize a low constant average degree distribution with high intermediate symbol recovery rates (ISRR) to improve the performance of more important bits (MIB). Asymptotic analysis of the proposed scheme is obtained by deriving unequal density evolution (DE) formulas over the binary erasure channel (BEC). Utilizing LDPC codes and the degree distribution with high ISRR in the proposed scheme, the UEP property is flexible and efficient. Compared with previous duplicate UEP-LT codes, mathematical analysis and simulation results demonstrate that the proposed UEP-LT codes have better performance.