A Collusion Avoidance Node Selection Scheme for Social Network-Based Distributed Data Storage

Motivated by the privacy protection requirement of a traffic accident emergency rescue system, we propose to select trustworthy nodes in one's social network as the data keepers and store pieces of encoded data in a distributed way. Specifically, we utilize Fountain Code to encode data and generate sufficient amount of coding symbols meanwhile select sufficient number of data keepers to ensure the success of recovery of data in such an unstable environment. Besides, we further consider the concept of gaming theory to prevent the collusion between data keepers, we derive a trust score function to evaluate nodes in one's social network and propose a Collusion Avoidance Node Selection Scheme to improve the reliability and avoid collusion. By this approach, we solve the problems of privacy preserving and the instability of nodes in the social network. In the experiments, we use real datasets to study the performance of our proposed approach. The experimental results show that our approach outperforms the Best Score Selection and Random Selection methods and our approach can avoid collusion effectively and improve the data privacy.