A game theoretic framework for stochastic multipath routing in self-organized MANETs |
| |
Affiliation: | 1. Cyber Security-NTAMC, Power Grid Corporation of India Limited (POWERGRID), Gurgaon, Haryana, 122 001, India;2. Department of Electronics and Electrical Communication Engineering, Indian Institute of Technology, Kharagpur, 721 302, India;1. Institute for Information Industry, Taipei, Taiwan;2. Department of Computer Science and Information Engineering, National Taiwan University of Science and Technology, Taipei, Taiwan;1. Academic Unit of Garanhuns, Federal Rural University of Pernambuco, Brazil;2. Informatics Center, Federal University of Pernambuco, Brazil;3. Federal Institute of Education, Science, and Technology of Sergipe, Lagarto, Brazil;4. Department of Statistics and Informatics, Federal Rural University of Pernambuco, Recife, Brazil;1. Faculty of Engineering and Natural Sciences, Sabancı University, Istanbul, Turkey;2. Department of Pediatric Cardiology, İstanbul Medipol University, Istanbul, Turkey;1. Security Team, SKI Corp., Seoul, Republic of Korea;2. Graduate School of Human ICT Convergence, Sungkyunkwan University, Suwon, 440-746, Republic of Korea |
| |
Abstract: | In this paper we propose a game theoretic framework for stochastic multipath routing in mobile ad hoc networks (MANETs). In a MANET, intelligent and adaptive attackers may try to hijack, jam or intercept data packets traveling from source to destination. In our proposed game, at each stage the source node keeps track of the available multiple paths, the residual bandwidth of the paths and the strategy of the attackers from the information gathered during the previous stage. Based on these observations, the source node selects a path for data communication and switching strategy among the multiple established paths between the source node and the destination node. Accordingly, it selects an optimal routing strategy to send data packets to the destination at each stage of the game. Using minimax-Q learning, the selected routing strategy maximizes the expected sum of per stage discounted payoff, which is the utilization of residual bandwidth between a source–destination pair along with the probability that the path is safe. Performance analysis and numerical results show that our proposed scheme achieves significant performance gains in terms of residual bandwidth utilization, average end-to-end delay, packet delivery ratio, routing overhead and security. |
| |
Keywords: | MANET Stochastic game Security Zero-sum game Reinforcement learning |
本文献已被 ScienceDirect 等数据库收录! |
|