A novel consensus based prediction strategy for data sensing

Financial contagion problems have been extensively studied in area of financial research. Most of the works focus on studying the contagion effect on the financial system. Contagion prediction is considered as one of the most important strategies to prevent contagion. But the prediction issue is seldom researched in the financial area. Traditional financial management uses a centralized method to predict contagion risk. But the central management cannot instantly acquire complete information of entire network. A decentralized method is needed to achieve a prediction in real time. This paper introduces a distributed risk contagion prediction strategy of the financial network. Firstly, consensus algorithm is used to distributively acquire contagion risk information of the entire financial network. This distributed strategy enables the system to instantly predict the risk of contagion. Secondly, the impact of the financial crisis could enormously influence the convergency of consensus algorithm. So a consensus based Kalman filter (CAF) is proposed to maintain the convergency of consensus and ensure the accuracy of the prediction. Finally, the strategy is tested in different kinds of financial systems which are impacted by different levels of the financial crisis. The simulation result shows that the strategy is robust, flexible and feasible for practical use. It also proves that the proposed strategy can provide an accurate prediction in any condition.