Research On Data Quality In Wireless Sensor Networks

In this century,wireless sensor networks show their broad application prospect in many fields and various industries.Many applications deploy a large number of heterogeneous sensor nodes,they can monitor various environmental information of interest to users in a designated area in real time and process the data according to user requirements.Extensive application prospects and diverse needs make its data-centric characteristics more important.The data in the wireless sensor networks has many data problems due to node softness,hardware failure and so on,such as data anomaly,data loss and data failure.Low quality data will reduce the application's performance and may cause the application to make erroneous business decisions.Anomaly detection technology can filter the wrong data to avoid its adverse effects on the sensor network applications.At the same time,executing data cleansing at the central node to improve data quality is an effective way to avoid low-quality data from making false decisions.This paper studies real-time and efficient anomaly detection methods and reasonable data cleaning strategies in sensor networks.The main work of the paper includes the following aspects.Firstly,the problem of anomaly detection of wireless sensor networks based on pattern frequency is studied.Aiming at the problem of abnormal data,this paper proposes a distributed anomaly data detection method for wireless sensor networks based on data change pattern and data spatial correlation.The method first creates a distribution model for the data,that is,the data change pattern is mapped to the partition of a feature space,determines the infrequent partition,and then determines whether the data falls on the infrequent partition to filter out the potential anomaly value.Finally,according to the spatial correlation of data between neighborhood nodes to determines the cause of the outliers.The experimental results show that the proposed method can effectively detect and distinguish the wrong data and abnormal events in the network while avoiding the high computational complexity.Secondly,the data quality analysis and the cleaning strategy of wireless sensor networks are studied.Unreasonable data cleaning strategies can't achieve the desired cleaning effect,while increasing the system cleaning costs.To solve the above problem,this paper first gives the data quality evaluation indicators and its measurement methods of wireless sensor networks.Secondly,we researched the impact of the relationship between different indicators on the quality assessment during the data cleaning process.Finally,by comparing and analyzing data cleaning schemes with different orders,a wireless sensor networks data cleaning strategy based on the relationship between data quality indicators is proposed.The experimental results show that under the same cleaning cost,the data cleaning strategy can effectively improve the data availability of wireless sensor networks and have better cleaning effect.