Chapter

Computational Collective Intelligence. Technologies and Applications

Volume 6423 of the series Lecture Notes in Computer Science pp 398-405

An Innovative Routing Algorithm with Reinforcement Learning and Pattern Tree Adjustment for Wireless Sensor Networks

  • Chia-Yu FanAffiliated withCarnegie Mellon UniversityDepartment of Computer Science and Information Engineering, Fu-Jen Catholic University
  • , Chien-Chang HsuAffiliated withCarnegie Mellon UniversityDepartment of Computer Science and Information Engineering, Fu-Jen Catholic University
  • , Wei-Yi WangAffiliated withCarnegie Mellon UniversityDepartment of Computer Science and Information Engineering, Fu-Jen Catholic University

* Final gross prices may vary according to local VAT.

Get Access

Abstract

This paper proposes a new routing algorithm for wireless sensor network. The algorithm uses reinforcement learning and pattern tree adjustment to select the routing path for data transmission. The former uses Q value of each sensor node to reward or punish the node in the transmission path. The factor of Q value includes past transmission path, energy consuming, transmission reword to make the node intelligent. The latter then uses the Q value to real-time change the structure of the pattern tree to increase successful times of data transmission. The pattern tree is constructed according to the fusion history transmission data and fusion benefit. We use frequent pattern mining to build the fusion benefit pattern tree. The experimental results show that the algorithm can improve the data transmission rate by dynamic adjustment the transmission path.

Keywords

Wireless Sensor Networks reinforcement learning routing path fusion benefit pattern tree