ADApt: Edge Device Anomaly Detection and Microservice Replica Prediction

TL;DR

ADApt addresses the challenge of orchestrating microservices on resource-constrained edge devices by combining online edge-device anomaly detection with replica-based scaling. It extends ADA-PIPE with a k-means-based anomaly detector on CORE/MEM utilization and explores gradient boosting regression, bagging, and MLP models to predict required replicas, enabling adaptive scheduling. The approach achieves best-in-class replication accuracy (e.g., MAE $=0.038$, MAPE $=0.002$, RMSE $=0.196$) and reduces device CPU utilization by 14–28% in experiments, highlighting practical gains for edge microservice management. Overall, ADApt demonstrates how online monitoring, clustering, and ensemble regression can enable proactive, efficient edge orchestration with potential privacy-preserving extensions.

Abstract

The increased usage of Internet of Things devices at the network edge and the proliferation of microservice-based applications create new orchestration challenges in Edge computing. These include detecting overutilized resources and scaling out overloaded microservices in response to surging requests. This work presents ADApt, an extension of the ADA-PIPE tool developed in the DataCloud project, by monitoring Edge devices, detecting the utilization-based anomalies of processor or memory, investigating the scalability in microservices, and adapting the application executions. To reduce the overutilization bottleneck, we first explore monitored devices executing microservices over various time slots, detecting overutilization-based processing events, and scoring them. Thereafter, based on the memory requirements, ADApt predicts the processing requirements of the microservices and estimates the number of replicas running on the overutilized devices. The prediction results show that the gradient boosting regression-based replica prediction reduces the MAE, MAPE, and RMSE compared to others. Moreover, ADApt can estimate the number of replicas close to the actual data and reduce the CPU utilization of the device by 14%-28%.

Figures (4)

• Figure 1: Device $d_0$ overutilization anomaly in ten time slots.
• Figure 2: ADApt architecture.
• Figure 3: Over- and full/under-utilization clusters in k-means-based anomaly detection model.
• Figure 4: Loss during the training iterations.