Exploiting data correlation for multi-scale processing in sensor networks

With the emergence of large and multi-scale sensor networks, the technologies of multi-scale processing among various sensors become an essential issue. In this paper, the problem of exploiting data correlation for multi-scale sensor networks is considered, and an architecture exploiting correlation is designed for both intra- and inter-data processing. Our correlation-adaptive scheme follows the characteristics of real sensor data, and fills the gap of the correlation models addressed by most of previous research with inherent support for related data gathering algorithms. A core solution module of this architecture is devised, and theoretical analysis and simulation studies are conducted on real-world datasets. Through the real-world data experiments in terms of accuracy and energy-consumption evaluation, the correlation-adaptive scheme is shown to work well in multi-scale processing sensor networks.