Loading...
| Friend's email | |
| Your name | |
| Your email | |
| enter code | |
This page was sent successfuly
- Type of Document: M.Sc. Thesis
- Language: Farsi
- Document No: 50227 (05)
- University: Sharif University of Technology
- Department: Electrical Engineering
- Advisor(s): Nasiri Kenari, Masoumeh; Aminzadeh Gohari, Amin
- Abstract:
- In recent years, molecular communications have received considerable attentions for communication between nano-machines. In this communication scheme, which uses molecules as information carriers, nano transmitter and receiver are in distance apart of nano meter to meter and information is encoded into type, concentration or releasing time of molecules. Since nano transmitter and receiver can do very simple operations, such as adding, sensing, and etc., for more complicated operations, a network of nano transceivers is required. In this thesis, a small scale imaging of an abnormality in the environment is investigated. To this end, a network of sensors is assumed to be placed in the environment, each measuring the intensity of the abnormality's marker in its own place. Then, based on the measured value, the sensor releases a number of molecules, toward the central receiver (or the fusion center-FC). To distinguish the transmitted information from different sensors at FC, each sensor utilizes a distinct type of molecules. Based on the received information from the sensors, the FC provides an image of the covered environment using different methods of estimations. Two parameter estimation schemes, namely Classic estimation and Bayesian estimation, are considered. In the Classic estimation, it is assumed that abnormality is distributed into the environment and so by deploying the best (Maximum Likelihood Estimation-MLE) and the simplest (Least square-LS) methods, the intensity of the produced marker over the covered environment is determined. Due to limited number of sensors and the lack of knowledge about the number of abnormality's sources, the covered environment should be quantized into small areas with one or a few sensors placed in each area. After estimating the produced marker's intensity in each area, the interpolation technique is applied to determine the produced marker's intensity at the remained places. In the Bayesian estimation, because of the complexity of the problem in general case, it is assumed that only one abnormality source exists in the environment. After estimating the place and the intensity of the abnormality source, the image of the marker distribution in the environment can be obtained. For performance evaluation, an example of abnormality occurrence is assumed. Then, using different proposed estimation schemes, we obtain images of the marker distribution and compare the results with original image. We also compare the performance of these methods using mont-carlo simulation and investigate the impacts of the number of available sensors and the number of transmission slots of each sensor. Numerical results shows that increasing the number of sensors and also the number of transmission slots, improve the performance of the proposed schemes. In the Classic estimation, the performance of MLE is not much better than that of the optimum LS estimator, while former is more complicated. Also in the Bayesian estimation, Maximum A Posteriori (MAP) method has better performance, while more complexity in comparison with Linear Minimum Mean Square Error (LMMSE)
- Keywords:
- Sensors ; Bayesian Estimation ; Image ; Molecular Communication ; Anomaly ; Classic Estimation ; Marker
Digital Object List
محتواي کتاب
Bookmark