Estados financieros consolidados por los años terminados el 31 de diciembre de 2012 y 2011 e informe de los
CORPBANCA ADMINISTRADORA GENERAL DE FONDOS S.A a Directorio
This dissertation explores the notions of connected-k-coverage, information coverage to the area or barrier, and multi-round sensor deployment in wireless sensor networks, and studies the theoretical foundations about them. Main contributions of the dissertation are summarized as follows.
7.1.1 Connected-k-Coverage
We explore the notion of connected-k-coverage. A region is said to be connected-k-covered if it is k-covered by a connected component of active sensors. Clearly, such a requirement is less demanding than requiring k-coverage of the sensing region and connectivity among all active sensors at the same time, but can still guarantee proper functioning of the sensor network. As a result, the number of required active sensors may be reduced, thus prolonging the sensor network lifetime. As one of the first efforts in understanding the fundamental basics about connected-k-coverage, we analyze the critical conditions for connected-k-coverage using the percolation theorem and demonstrate their effectiveness through simulation-based validation. In addition, we derive an effective asymptotic lower bound on the probability of connected-k-coverage, and propose a simple randomized scheduling algorithm and select proper operational parameters to prolong the lifetime of a large-scale sensor network.
7.1.2 On-Demand Object Detection Framework based on Decision Fusion
Based on the probabilistic sensing model, we define ρ-coverage as a measure of the coverage per- formance for a randomly-deployed wireless sensor network. Upon detailed analytical study, we find that simple decision fusion-based collaborations among active sensors indeed degrade the coverage performance due to the requirement of maintaining the target false detection probability. This counter-
intuitive result motivates us to develop an on-demand collaborative framework for object detection. The idea of our framework is that it no longer mandates active sensors to collaborate only with each other; instead, upon sensing a measurement higher than the decision threshold, an active sensor triggers its neighboring inactive sensors to collaboratively sense the environment. This way, by leveraging on the inactive sensors we could achieve the same low false detection probability while increasing the prob- ability of detection because the density of inactive sensors is usually much higher than that of active sensors. The effectiveness of the proposed framework is supported by detailed theoretical analysis as well as simulation-based validation. Moreover, since decision fusion incurs extra energy consumption in aggregating collaborative messages, we further investigate the energy efficiency performance of the proposed framework, and offer some interesting observations and insights on how to select proper sys- tem parameters to maximize the network lifetime while maintaining the target coverage performance.
7.1.3 Information Barrier Coverage
We study the barrier coverage problem under the assumption that neighboring sensors may collab- orate with each other to form a virtual sensor which makes the detection decision based on combined sensed readings. This is also known as the barrier information coverage problem. Our goal is to information-cover a barrier with as few active sensors as possible. This is different from the barrier coverage problem under the conventional 0/1 disc sensing model because the size and shape of the coverage region of a virtual sensor vary with the decision threshold and the positions of collaborat- ing sensors. Note that the size and shape of sensors’ coverage regions affect the number of sensors needed to provide the coverage to the barrier, which in turn determines the decision threshold. Such mutual correlation makes this problem non-trivial and challenging. To address this problem, we pro- pose to use coverage projection to approximate the coverage region of a virtual sensor, and design a centralized algorithm to find a set of active sensors to information-cover the barrier. Moreover, when sensors’ location information is only available to their neighbors, we propose a simple distributed algo- rithm to help sensors determine their collaborating partners, in order to maximize the barrier coverage percentage.
7.1.4 Multi-Round Sensor Deployment for Guaranteed Barrier Coverage
We propose a line-based multi-round sensor deployment strategy for guaranteed barrier coverage, and conduct extensive analytical and simulation studies on reducing the number of sensors needed. We study the performance of multi-round sensor deployment and derive the optimal strategies that use the fewest sensors to cover a barrier. We find that the optimal two-round sensor deployment strategy yields the same barrier coverage performance as other optimal strategies with more than two rounds. This result is particularly encouraging as it implies that the best barrier coverage performance can be achieved with low extra deployment cost by deploying sensors in two rounds.
Furthermore, we propose two practical solutions, the two-round minimax solution and the pilot deployment solution, to deal with realistic situations when the knowledge about the deviation of sen- sors’ residence points with respect to their intended deployment points is not fully available. The pilot deployment solution performs particularly well and the idea is to introduce an additional pilot round which deploys a small number of sensors to estimate the distribution of sensors’ residence points and then use this information to aid the following rounds of sensor deployment.