Quality, Reliability, Security and Robustness in Heterogeneous Networks. 7th International Conference on Heterogeneous Networking for Quality, Reliability, Security and Robustness, QShine 2010, and Dedicated Short Range Communications Workshop, DSRC 2010, Houston, TX, USA, November 17-19, 2010, Revised Selected Papers

Research Article

Self-optimizing Mechanism for Prediction-Based Decentralized Routing

Download
459 downloads
  • @INPROCEEDINGS{10.1007/978-3-642-29222-4_32,
        author={Abutaleb Turky and Florian Liers and Andreas Mitschele-Thiel},
        title={Self-optimizing Mechanism for Prediction-Based Decentralized Routing},
        proceedings={Quality, Reliability, Security and Robustness in Heterogeneous Networks. 7th International Conference on Heterogeneous Networking for Quality, Reliability, Security and Robustness, QShine 2010, and Dedicated Short Range Communications Workshop, DSRC 2010, Houston, TX, USA, November 17-19, 2010, Revised Selected Papers},
        proceedings_a={QSHINE},
        year={2012},
        month={10},
        keywords={Traffic engineering self-organization ant-based routing quality of service artificial neural network},
        doi={10.1007/978-3-642-29222-4_32}
    }
    
  • Abutaleb Turky
    Florian Liers
    Andreas Mitschele-Thiel
    Year: 2012
    Self-optimizing Mechanism for Prediction-Based Decentralized Routing
    QSHINE
    Springer
    DOI: 10.1007/978-3-642-29222-4_32
Abutaleb Turky1,*, Florian Liers1,*, Andreas Mitschele-Thiel1,*
  • 1: Ilmenau University of Technology
*Contact email: abutaleb-abdelmohdi.turky@tu-ilmenau.de, florian.liers@tu-ilmenau.de, mitsch@tu-ilmenau.de

Abstract

In this paper, we introduce an adaptive traffic prediction approach for self-optimizing the performance of a Prediction-based Decentralized Routing (PDR) algorithm. The PDR algorithm is based on the Ant Colony Optimization (ACO) meta-heuristics in order to compute the routes. In this approach, an ant uses a combination of the link state information and the predicted available bandwidth instead of the ant’s trip time to determine the amount of deposited pheromone. A Feed Forward Neural Network (FFNN) is used to build adaptive traffic predictors which capture the actual traffic behavior. Our contribution is a new self-optimizing mechanism which is able to locally adapt the prediction validity period depending on the prediction accuracy in order to efficiently predict the link traffic. We study three performance parameters: the rejection ratio, the percentage of accepted bandwidth and the effect of prediction use. In general, our new algorithm reduces the rejection ratio of requests, achieves higher throughput when compared to the AntNet and Trail Blazer algorithms.