No description available.

M. S.

M.S.

M.Com. (Informatics) / Please refer to full text to view abstract

Thesis (M.S.)--Massachusetts Institute of Technology, Dept. of Mechanical Engineering, 1980. / MICHROFICHE COPY AVAILABLE IN ARCHIVES AND ENGINEERING. / Includes bibliographical references. / by Bradley James Deghuee. / M.S.

Thesis. 1976. B.S.--Massachusetts Institute of Technology. Dept. of Mechanical Engineering. / Microfiche copy available in Archives and Engineering. / Includes bibliographical references. / by Joseph B. Hamilton. / B.S.

In recent years there has been a growing trend in optical wireless convergence. One particular aspect of this is 60 GHz radio-over-fiber technology. It is intended for use in wireless personal area networks. However, we think that the same technology could be used for monitoring applications in the indoor environment. It could be used to detect emergency situations or to detect intruders. We shall examine reasons why this choice might be a suitable one. We shall then propose a MAC layer protocol to accomplish this task. Since in case of emergency we might require to obtain data from only one node for an extended duration, flexibility in implementation is required. We shall develop an adaptive MAC protocol where this would be possible. We accomplish this by including two protocol modes called the Icarus mode, which is to be used in case of an emergency and the Resync mode which is used when normality is restored. A significant problem at high frequencies is that the beam becomes increasingly narrow and behaves more in a ray like condition. This implies that particularly in an indoor environment it is possible that the beam may be accidentally blocked. In this case the node must be able shift the beam in order to enable communication. We demonstrate three such strategies and offer a comparative analysis.

No description available.

D.Ing. (Electrical and Electronic ) / This thesis describes a neural network based system for the classification of handwritten digits as found on real-life mail pieces. The proposed neural network uses a modular architecture which lends itself to parallel implementation. This modular architecture is shown to produce adequate performance levels while significantly reducing the required training time. The aim of the system is not only to achieve a high recognition performance, but also to gain more insight into the functioning of the neural networks. This is achieved by using separate feature extraction and classification stages. The output of the feature extraction stage gives a good indication of the final performance level of the classifier, even before training. The need for an optimal feature set is expressed to elevate the performance levels even further.

No description available.