Master of Science in Electrical Engineering (MSEE)
Electrical and Computer Engineering
We consider the estimation of a deterministic unknown parameter in an encrypted wireless sensor networks. Adaptive quantization is used on the sensor's observation and the outputs of the sensors are then encrypted using a probabilistic cipher. In a conventional fixed quantization scheme, estimation error grows exponentially with the difference between the threshold and the unknown parameter to be estimated. Hence, to avoid this, we used and adaptive quantization scheme where each sensor adaptively adjusts its quantization threshold. We find the Cramer-Rao Lower Bound for the Ally Fusion Center (AFC) and then find the optimal estimate of the unknown parameter for the AFC. To find this, we first prove that the sequence of thresholds used for the quantization process forms a markov chain and that this chain is recurrent non-null and thus has a stationary distribution. This distribution is then obtained analytically in closed form as well as through numerical methods. The optimal estimate of the unknown parameter for the AFC is obtained asymptotically in the number of sensors. The performance of the Third Party Fusion Center (TPFC) is only computed through simulation and compared to that of AFC.
Document Availability at the Time of Submission
Release the entire work immediately for access worldwide.
Konakalla, Niharika, "Adaptive quantization in wireless sensor networks with encryption" (2010). LSU Master's Theses. 2530.