Date of Award

December 2016

Degree Type


Degree Name

Doctor of Philosophy (PhD)


Electrical Engineering and Computer Science


Pramod K. Varshney


Localization, Sensor management, Tracking, Wireless sensor networks

Subject Categories



Wireless sensor networks (WSNs) are very useful in many application areas including battlefield surveillance, environment monitoring and target tracking, industrial processes and health monitoring and control. The classical WSNs are composed of large number of densely deployed sensors, where sensors are battery-powered devices with limited signal processing capabilities. In the crowdsourcing based WSNs, users who carry devices with built-in sensors are recruited as sensors. In both WSNs, the sensors send their observations regarding the target to a central node called the fusion center for final inference. With limited resources, such as limited communication bandwidth among the WSNs and limited sensor battery power, it is important to investigate algorithms which consider the trade-off between system performance and energy cost in the WSNs. The goal of this thesis is to study the sensor management problems in resource limited WSNs while performing target localization or tracking tasks.

Most research on sensor management problems in classical WSNs assumes that the number of sensors to be selected is given a priori, which is often not true in practice. Moreover, sensor network design usually involves consideration of multiple conflicting objectives, such as maximization of the lifetime of the network or the inference performance, while minimizing the cost of resources such as energy, communication or deployment costs. Thus, in this thesis, we formulate the sensor management problem in a classical resource limited WSN as a multi-objective optimization problem (MOP), whose goal is to find a set of sensor selection strategies which re- veal the trade-off between the target tracking performance and the number of selected sensors to perform the task. In this part of the thesis, we propose a novel mutual information upper bound (MIUB) based sensor selection scheme, which has low computational complexity, same as the Fisher information (FI) based sensor selection scheme, and gives estimation performance similar to the mutual information (MI) based sensor selection scheme. Without knowing the number of sensors to be selected a priori, the MOP gives a set of sensor selection strategies that reveal different trade-offs between two conflicting objectives: minimization of the number of selected sensors and minimization of the gap between the performance metric (MIUB and FI) when all the sensors transmit measurements and when only the selected sensors transmit their measurements based on the sensor selection strategy.

Crowdsourcing has been applied to sensing applications recently where users carrying devices with built-in sensors are allowed or even encouraged to contribute toward the inference tasks. Crowdsourcing based WSNs provide cost effectiveness since a dedicated sensing infrastructure is no longer needed for different inference tasks, also, such architectures allow ubiquitous coverage. Most sensing applications and systems assume voluntary participation of users. However, users consume their resources while participating in a sensing task, and they may also have concerns regarding their privacy. At the same time, the limitation on communication bandwidth requires proper management of the participating users. Thus, there is a need to design optimal mechanisms which perform selection of the sensors in an efficient manner as well as providing appropriate incentives to the users to motivate their participation. In this thesis, optimal mechanisms are designed for sensor management problems in crowdsourcing based WSNs where the fusion center (FC) con- ducts auctions by soliciting bids from the selfish sensors, which reflect how much they value their energy cost. Furthermore, the rationality and truthfulness of the sensors are guaranteed in our model. Moreover, different considerations are included in the mechanism design approaches: 1) the sensors send analog bids to the FC, 2) the sensors are only allowed to send quantized bids to the FC because of communication limitations or some privacy issues, 3) the state of charge (SOC) of the sensors affects the energy consumption of the sensors in the mechanism, and, 4) the FC and the sensors communicate in a two-sided market.


Open Access

Included in

Engineering Commons