| Summary: | Thesis (M.Eng.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 2003. === Includes bibliographical references (p. 81). === One of the most significant applications of Micro-Electromechanical Systems (MEMS) technology in optical communications today is in building large non-blocking optical crossconnects based on arrays of tiltable micro-mirrors. The complexity for these crossconnects to make all possible connections lies in the calibration or fine-tuning of the mirror tilt angles to optimize the transmissivity through each possible input/output pair. The result from the fine-tuning process that produces optimization at one point in time, however, does not guarantee optimization for future attempts. This thesis models the transmissivity as a function of control variables in the vicinity of an optimal point and uses this model to re-optimize the connections quickly when a connection is reestablished. The re-optimization algorithm achieves the goal of optimizing quickly by requiring that some prior knowledge about each connection is already known. Scalable methods for representing the per-connection transmissivity model are also studied. Experimental results of the algorithm performance on real crossconnect systems are reported, including connection setup in under 50 milliseconds. === by Piyajit Phanaphat. === M.Eng.
|
|---|
