Towards trusted node selection using blockchain for crowdsourced abnormal data detection

Abstract

Node selection plays an important role to design and implement the crowdsourced abnormal data detection system with the purpose of completing complex tasks to meet the requirements of computing performance. Even though the blockchain-based trusted node selection approaches improve the reliability of the crowdsourcing task results, they still need to consider the crowdsourcing cost during the node selection process so as to embody a tradeoff between reliability and cost. In this paper, we propose to study the node selection problem for the crowdsourced abnormal data detection under both reliability and cost requirements. First, the working node selection is modeled as an inverse 0−1 knapsack problem in order to minimize the crowdsourcing cost under the budget constraints of the trustworthiness of the selected working nodes, where blockchain is used to calculate the trustworthiness of the working nodes. Then, a trusted working node selection (TWNS) algorithm is developed to select trusted working nodes with the minimum crowdsourcing cost for crowdsourced abnormal data detection, where the branch and bound method is utilized to efficiently solve the inverse 0−1 knapsack problem. Finally, extensive simulations are conducted based on three groups of real-world datasets. The results show that the trust value evaluation is accurate by using blockchain and the TWNS algorithm can ensure the reliability of the detection result. The crowdsourcing cost is minimized in trusted working node selection process. Compared to existing approaches without considering the cost, the TWNS algorithm reduces the crowdsourcing cost by 64.6%.