A Statistical Integrity Authentication Scheme without Grouping for Streaming Data

The development of the sensor networks has enabled increasing commercial or scientific applications in which data are available as streams. And sometimes the core security requirement of streaming data in WSNs is integrity rather than confidentiality. Former stream authentication schemes based on watermarking usually suffer from the vulnerability of grouping. This paper proposed a novel integrity authentication scheme for streaming data. Copies of data elements are buffered in the queue instead of being divided into groups, and parts of the queue are selected randomly and uniformly according to probability to compute one-bit watermark. So, any data element will be verified by the statistical judging of its related one-bit watermarks. Low false negative rate results from several watermarks, and the watermarks related with certain data element will be distributed uniformly in the queue, which results in low false positive rate. The proposed method resolves a series of vulnerability issues from grouping, and the cost of integrity is composed of a small amount of computational overhead and some storage space. Our scheme is ideal for applications with real-time needs and limited resources in WSNs.