Loading...
Please wait, while we are loading the content...
Similar Documents
Distributed intelligent multi-agents for telecommunication network management.
| Content Provider | CiteSeerX |
|---|---|
| Author | Andrag, W. H. Omlin, C. W. |
| Abstract | Modern communication networks must cope with ever increasing demands on network resources. Intelligent agents which learn to manage network traffic from past experience hold the promise of providing the robustness necessary in order to meet these demands. Qlearning belongs to a class of machine learning algorithms which enable intelligent agents to change their behaviour based on feedback from their environment. This algorithm (`Q-routing') and its improved version (`Dual Reinforcement Q-routing') have been proposed for routing of data packets. Each node learns a routing policy based on feedback from its neighbours. Using the British Synchronous Digital Hierarchy Network as an example, we present empirical evidence which suggests that DRQ-routing does not always outperform simple Q-routing, particularly when the network has to recover from link failures. |
| File Format | |
| Access Restriction | Open |
| Subject Keyword | Telecommunication Network Management Intelligent Multi-agents Improved Version Simple Q-routing Empirical Evidence Machine Learning Algorithm Past Experience British Synchronous Digital Hierarchy Network Dual Reinforcement Q-routing Link Failure Data Packet Modern Communication Network Enable Intelligent Agent Routing Policy Network Traffic Intelligent Agent Network Resource |
| Content Type | Text |
