Wireless Sensor Missing Value Estimation Algorithm Based On Multi-Attribute

Abstract

Because the wireless sensor is arranged in the environment of unmanned management and complex, in the process of collecting data and transmitting data, it often leads to data loss due to the influence of itself or external environment. The best way to reduce the impact of missing value is to estimate the missing value. In this paper, we propose a missing value estimation algorithm based on time attribute and trust mechanism. We use Arima to predict the time attribute, and use the relationship between current value, historical value and error to predict the future data. We use subjective logic to convert the interaction information between nodes into trust value, and then Linear Regression prediction value by selecting the number of trust nodes. Finally, according to the optimal fit degree, weight distribution is carried out to form the final prediction value. Because the algorithm not only considers the node data of the trusted neighbor, but also predicts the future data changes through the changes of its own historical data, it has higher accuracy and lower error when compared with other algorithms.