A new Multi-Agents System based on Blockchain for Prediction Anomaly from System Logs

Abstract

The execution traces generated by an application contain information that the developers believed would be useful in debugging or monitoring the application, it contains application states and significant events at various critical points that help them gain insight into failures and identify and predict potential problems before they occur. Despite the ubiquity of these traces universally in almost all computer systems, they are rarely exploited because they are not readily machine-parsable. In this paper, we propose a Multi-Agents approach for prediction process using Blockchain technology, which allows automatically analysis of execution traces and detects early warning signals for system failure prediction during executing. The proposed prediction approach is constructed using a four-layer Multi-Agents system architecture. The proposed prediction approach performance is based on data prepossessing and supervised learning algorithms for prediction. Blockchain was used to coordinate collaboration between agents, and to synchronize prediction between agents and the administrators. We validated our approach by applying it to real-world distributed systems, where we predicted problems before they occurred with high accuracy. In this paper we will focus on the Architecture of our prediction approach.