Quality assurance for data acquisition in error prone WSNs

This paper proposes a data acquisition scheme which supports probabilistic data quality assurance in an error-prone wireless sensor network (WSN). Given a query and a statistical model of real-world data which is highly correlated, the aim of the scheme is to find a sensor selection scheme which is used to deal with inaccurate data and probabilistic guarantee on the query result. Since most sensor readings are real-valued, we formulate the data acquisition problem as a continuous-state partially observable Markov decision process (POMDP). To solve the continuous-state POMDP, the fitted value iteration (FVI) is applied to find a sensor selection scheme. Numerical results show that FVI can achieve high average long-term reward and provide probabilistic guarantees on the query result more often when compared to other algorithms.