Cancan Hua

Cancan Hua

Auto-generated profile, based on publicly available information

Cancan Hua


Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
3 March 2026 in Preprints.org
Verified
Preprint
Open access
PDF file
CHCancan Hua
Cancan Hua
This study proposes an anomaly perception and prediction model that integrates graph structure and temporal dependence to address the complexity of anomaly detection and fault prediction in cloud service systems. The model first constructs a service dependency graph to capture the structural topology of the system and then employs a temporal encoding module to model the dynamic evolution of service metrics. In the graph-temporal fusion layer, structural and temporal features are deeply integrated to enhance the model's ability to capture anomaly propagation paths and temporal patterns. Using multidimensional monitoring data as input, the model achieves a global representation of cloud service states through spatial aggregation and temporal dependency modeling, leading to stable performance in anomaly detection and root cause localization. Experimental results show that the proposed model significantly outperforms traditional methods in accuracy, recall, and F1-Score, maintaining robustness and generalization under complex topologies and highly dynamic environments. Further experiments on hyperparameter, environment, and data sensitivity verify the model's adaptability to key factors such as learning rate, communication delay, data missing rate, and anomaly ratio. This method provides an interpretable and structured solution for intelligent cloud operations and offers transferable theoretical and technical support for anomaly detection and risk awareness in large-scale distributed systems.

16 December 2025 in Preprints.org
Verified
Preprint
Open access
PDF file
CHCancan Hua
Cancan Hua
NLNing Lyu
Ning Lyu
CWChen Wang
Chen Wang
TYTingzhou Yuan
Tingzhou Yuan
This study proposes a Transformer-based change-point detection method for modeling and anomaly detection of multidimensional time-series metrics in Kubernetes nodes. The research first analyzes the complexity and dynamics of node operating states in cloud-native environments and points out the limitations of traditional single-threshold or statistical methods when dealing with high-dimensional and non-stationary data. To address this, an input representation mechanism combining linear embedding and positional encoding is designed to preserve both multidimensional metric features and temporal order information. In the modeling stage, a multi-head self-attention mechanism is introduced to effectively capture global dependencies and cross-dimensional interactions. This enhances the model's sensitivity to complex patterns and potential change points. In the output stage, a differentiated scoring function and a normalized smoothing method are applied to evaluate the time series step by step. A change-point decision function based on intensity scores is then constructed, which significantly improves the ability to identify abnormal state transitions. Through validation on large-scale distributed system metric data, the proposed method outperforms existing approaches in AUC, ACC, F1-Score, and Recall. It demonstrates higher accuracy, robustness, and stability. Overall, the framework not only extends attention-based time-series modeling at the theoretical level but also provides strong support for intelligent monitoring and resource optimization in cloud-native environments at the practical level.