A proposed method of implementing cross connections in an optical communication network is based on the use of a spatial light modulator (SLM) to form controlled diffraction patterns that connect inputs (light sources) and outputs (light sinks). Sources would typically include optical fibers and/or light-emitting diodes; sinks would typically include optical fibers and/or photodetectors. The sources and/or sinks could be distributed in two dimensions; that is, on planes. Alternatively or in addition, sources and/or sinks could be distributed in three dimensions - for example, on curved surfaces or in more complex (including random) three-dimensional patterns.

This Optical Assembly would convert input light of any polarization (S0) to two mutually independent polarizations (S1 and S2). S2 would then be further converted to S1. The output beam would be pure S1, attenuated from S0 by a factor of 2.

The proposed method offers the following advantages over prior methods:

  • Invariance to polarization of incoming light;
  • Minimization of crosstalk;
  • A full connectivity matrix (that is, the possibility of connecting or disconnecting between any input and any output terminal) in a given optical crossbar switch;
  • Retention of switched information in light-borne form (in contradistinction to absorption of light, intermediate processing in electronic form, and re-emission of light);
  • Accommodation of the undesired but unavoidable coupling of phase and amplitude modulation in a realistic spatial light modulator;
  • Automated dynamic alignment of the components of a newly assembled optical crossbar switch;
  • Switching in a single stage rather than multiple "butterfly" stages;
  • Computational tradeoff among desired but at least partly mutually exclusive switch characteristics (for example, among diffraction efficiency, uniformity of connection strengths, and crosstalk);
  • Design for operation in the Fresnel (near-field) diffraction regime rather than in the Fourier (far-field approximation) regime;
  • Ability to utilize inexpensive lenses and other less-than-ideal fixed optical elements; and
  • Direct (in contradistinction to indirect) optimization of switch properties.

The method incorporates a combination of synergistic techniques and concepts developed to solve problems encountered in prior research on crossbar optical switches. The combination of techniques and concepts is so extremely complex that only a highly abbreviated summary of a few salient features, addressing some of the aforementioned advantages, can be given below.

The issue of polarization arises because the performances of many SLMs affect the polarization of output light and are affected by the polarization of input light. Because it is impractical to guarantee the polarization of input light from disparate sources, it would be better to render a crossbar switch insensitive to input polarization. In a crossbar switch according to the proposal, all of the input light would be converted to a single input polarization desired for the SLM by use of an optical assembly like that shown in the figure. The light would be attenuated by a factor of 2 by passage through this assembly, but in a typical case, the disadvantage of this attenuation would be offset by the advantage of obtaining the polarization needed to optimize the performance of the SLM.

The method provides for the choice of input and output locations in a computational optimization process to minimize crosstalk, while making it possible to connect from any desired input terminal(s) to any desired output terminal(s). In this process, the far-field approximation would be irrelevant and unnecessary because one would utilize realistic near-field measured and/or computed diffraction patterns generated by the SLM. The inherently comprehensive nature of the optimization calculations is such that realistic modulation characteristics of the SLM, realistic optical characteristics of all components, alignment or misalignment of components, diffraction efficiency, uniformity or nonuniformity of signal strengths, and crosstalk would all be automatically taken into account.

This work was done by Richard Juday of Johnson Space Center.

This invention is owned by NASA, and a patent application has been filed. Inquiries concerning nonexclusive or exclusive license for its commercial development should be addressed to

the Patent Counsel
Johnson Space Center
(281) 483-0837

Refer to MSC-23320.