LSTM Recurrent Neural Network (RNN) for Anomaly Detection in Cellular Mobile Networks

Mamun, Shamim Al; Beyaz, Mehmet

doi:10.1007/978-3-030-19945-6_15

Cited by 7 publications

(7 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The specific threshold to consider each sample as degraded or not is typically established manually or automatically based on the general statistical distribution of such an error. Here, many different predictive models can be used, such as the AutoRegressive Integrated Moving Average (ARIMA) [ 18 ], forward linear predictors [ 24 ], or Long Short-Term Memory (LSTM) recurrent neural networks [ 25 ]. These methods, especially those using more advanced predictors, will typically require large datasets, and their performance for anomaly detection will be highly impacted by the proper configuration of the model and the variability and stationarity of the analyzed metric.…”

Section: Related Workmentioning

confidence: 99%

Transform-Based Multiresolution Decomposition for Degradation Detection in Cellular Networks

Fortes

Muñoz

Serrano

et al. 2020

Sensors

View full text Add to dashboard Cite

Anomaly detection in the performance of the huge number of elements that are part of cellular networks (base stations, core entities, and user equipment) is one of the most time consuming and key activities for supporting failure management procedures and ensuring the required performance of the telecommunication services. This activity originally relied on direct human inspection of cellular metrics (counters, key performance indicators, etc.). Currently, degradation detection procedures have experienced an evolution towards the use of automatic mechanisms of statistical analysis and machine learning. However, pre-existent solutions typically rely on the manual definition of the values to be considered abnormal or on large sets of labeled data, highly reducing their performance in the presence of long-term trends in the metrics or previously unknown patterns of degradation. In this field, the present work proposes a novel application of transform-based analysis, using wavelet transform, for the detection and study of network degradations. The proposed system is tested using cell-level metrics obtained from a real-world LTE cellular network, showing its capabilities to detect and characterize anomalies of different patterns and in the presence of varied temporal trends. This is performed without the need for manually establishing normality thresholds and taking advantage of wavelet transform capabilities to separate the metrics in multiple time-frequency components. Our results show how direct statistical analysis of these components allows for a successful detection of anomalies beyond the capabilities of detection of previous methods.

show abstract

Section: Related Workmentioning

confidence: 99%

Transform-Based Multiresolution Decomposition for Degradation Detection in Cellular Networks

Fortes

Muñoz

Serrano

et al. 2020

Sensors

View full text Add to dashboard Cite

show abstract

“…Once the training of NN is completed, the obtained NN can be easily used to detect whether anomaly exists for each input network measurement. Beyond using standard artificial NNs for AD, some works have used new types of NN for AD [1,[15][16][17][18]], which will be described in detail below.…”

Section: Neural Network (Nn)mentioning

confidence: 99%

“…Therefore, this method can detect global anomalies and identify which key performance indicator (KPI) anomalies are specific to the core network. Compared with unsupervised NN methods in [15] and other works, such as [19], some semi-supervised [16], and supervised [1,17,18] NN methods with lower ambiguity of training samples, controllable prediction results and better model quality have been pervasive in cellular network anomaly detection.…”

Section: Neural Network (Nn)mentioning

confidence: 99%

“…Based on the RNN (not the same NN as the one mentioned above), Mamun et al [18] also adopt an unsupervised method called long short-term memory (LSTM) to conduct anomaly detection. Different from the traditional RNN, LSTM adds a cell to determine whether the information is useful or not.…”

Section: Neural Network (Nn)mentioning

confidence: 99%

“…Under such repeated operation, LSTM can effectively solve the problem of long sequence dependence. Based on the LSTM algorithm, Mamun et al [18] model the time series of KPI values of cells in the LTE network and was able to detect long and short interval anomalies in the cellular network at the same time, achieving faster cell level anomaly detection.…”

Section: Neural Network (Nn)mentioning

confidence: 99%

See 2 more Smart Citations

Anomaly detection for cellular networks using big data analytics

Zhao

Zhang

et al. 2019

IET Communications

View full text Add to dashboard Cite

Large language models (LLMs) have shown their potential in long-context understanding and mathematical reasoning. In this paper, we study the problem of using LLMs to detect tabular anomalies and show that pre-trained LLMs are zeroshot batch-level anomaly detectors. That is, without extra distribution-specific model fitting, they can discover hidden outliers in a batch of data, demonstrating their ability to identify low-density data regions. For LLMs that are not well aligned with anomaly detection and frequently output factual errors, we apply simple yet effective datagenerating processes to simulate synthetic batchlevel anomaly detection datasets and propose an end-to-end fine-tuning strategy to bring out the potential of LLMs in detecting real anomalies. Experiments on a large anomaly detection benchmark (ODDS) showcase i) GPT-4 has on-par performance with the state-of-the-art transductive learning-based anomaly detection methods and ii) the efficacy of our synthetic dataset and fine-tuning strategy in aligning LLMs to this task.

show abstract