DOI: 10.17587/prin.9.11-21

The Distributed Multi-agent System for Monitoring and Forecasting of the Networking Transport Layer Performance

O. V. Bogoiavlenskaia, e-mail: olbgvl@cs.karelia.ru, Petrozavodsk State University, Petrozavodsk, 185910, Russian Federation

Corresponding author: Bogoiavlenskaia Olga V., Associate Professor, Petrozavodsk State University, Petrozavodsk, 185910, Russian Federation, E-mail: olbgvl@cs.karelia.ru

The paper describes the architecture of the distributed multi-agent system for monitoring and forecasting of the networking transport layer performance. The system consists of the set of the agents which monitor, analyze captured data and forecast the performance of most intensively used end-to-end paths. The forecast allows users and network application to choose most efficient strategies to utilize the capacity of network resources. Therefore the end-users, mobile devices, LAN administrations, and network applications can choose content provider, end-to-end-path or data source to avoid drawbacks of congested and/or unreliable end-to-end paths. Content providers can use the forecast to monitor QoS and/or QoE of their users and subscribers and thus adjust marketing policies and investment plans. The high level architecture consists of the monitoring module, forecasting module, list of the end-to-end paths under analysis, compressed data store and interface module. We define two levels of detailing for the agents architecture to reduce monitoring and processing overhead. The agents exchange data they possess about end-to-end paths performance and maintain list of agents they can communicate. The set of use cases is presented. The example of the experimental analysis for choosing video-content provider is presented as well.

Keywords: monitoring, Linux kernel, network performance, transport layer, open architecture, distributed systems, probalistic modeling, time-series analysis, video-streaming, multi-agent systems

pp. 11–21

For citation:
Bogoiavlenskaia O. V. The Distributed Multi-agent System for Monitoring and Forecasting of the Networking Transport Layer Performance, Programmnaya Ingeneria, 2018, vol. 9, no. 1, pp. 11—21.