Identifier

etd-07072011-183310

Degree

Master of Science (MS)

Department

Electrical and Computer Engineering

Document Type

Thesis

Abstract

Utilizing wireless sensor network (WSN) is a novel idea in a variety of applications. However, the limited resources allocated to the sensor nodes make the design of WSNs a challenging problem. We consider the problem of hypothesis testing in a bandwidth-constrained, low-power wireless sensor network operating over insecure links. Sensors quantize their observations and transmit their decisions to an intended (ally) fusion center (AFC) which combines the received messages to detect the state of an unknown hypothesis.

In many applications the sensor messages are vulnerable to unauthorized eavesdropping. The scarce bandwidth and processing power for the sensors rule out the utilization of advanced encryption techniques. To protect their transmissions from an unauthorized (third party) fusion center (TPFC), the sensors use a simple encryption whereby they randomly flip their quantization outcomes, similarly to what happens in a discrete memoryless channel. It is assumed that AFC is aware of the encryption probabilities (keys) but TPFC is not. For the AFC the decision rule is formulated as a constrained optimization problem where one constraint is a lower bound on the error probability of TPFC. The optimal decision rules for the two fusion centers are then derived. It is shown that by appropriate design of the encryption probabilities and the AFC decision rule, it is possible to degrade the error probability of the TPFC significantly and still achieve very low probability of error for the AFC. Numerical results are presented to show that it is possible to ensure that TPFC does not gain any information from the observation of sensors transmissions.

Date

2011

Document Availability at the Time of Submission

Release the entire work immediately for access worldwide.

Committee Chair

Naraghi-Pour, Morteza

Share

COinS