Missing Data Not At Random: Characterization of Targeted Interference in Wireless Networks

Communication systems include data collection and estimation during their operations. At the receiver, the data can be missed due to various reasons such as channel conditions, malicious attack, failure at the receiver. Some of these conditions occur at random, but sometimes, their occurrences are not random. These occurrences can be due to the precise placement of interference to impede communication between the devices. There are different mechanisms proposed in the literature to address this data loss, by requesting retransmission, spreading the signal, etc. A different approach is by using statistical analysis to mine data received to estimate the data points that are missed not at random. In the statistical study, data that are missed not at random manifest during data collection when one or more subjects involved in the survey skip their responses for one or more data fields due to social, economic, and health reasons. These missed responses are filled by imputation from the responses collected from other subjects. Similarly, in a wireless network, data lost from a particular node that is under attack can be considered as data missed not at random and can be estimated from the data collected from the surrounding nodes.