Spectrally Efficient Signalling for Wireless Optical Intensity Channels

Steve Hranilovic

Degree: Ph.D.
Supervisor: Prof. Frank R. Kschischang
Department: Electrical and Computer Engineering, University of Toronto, Ontario, Canada
Date of Graduation: November 2003

Contact Information:
Prof. Steve Hranilovic
Department of Electrical and Computer Engineering
McMaster University
Hamilton, Ontario, L8S 4K1 Canada
Email: hranilovic@ece.mcmaster.ca

Abstract

pectrally efficient signalling is important in wireless optical channels due to their low-pass multipath distortion as well as optoelectronic response limitations. This thesis presents two approaches to improving the spectral efficiency of wireless optical channels: through the extension of traditional electrical modem design to optical intensity domain and through the use of spatial diversity.

This work extends the results of signalling on electrical channels by defining a signal space model which captures both the amplitude constraints and average optical power cost of all time-disjoint optical intensity signal sets geometrically. Lattice codes satisfying the channel constraints are designed subject to a bandwidth constraint which is imposed via an effective dimension parameter. Spectrally efficient modulation for short range point-to-point links is shown to provide significant rate gain over PPM. Asymptotically exact upper and lower bounds on the capacity of given signalling sets are derived and spectrally efficient signalling is shown to be necessary to maximize rates at high optical SNR. A novel channel topology, the pixelated wireless optical channel, is presented which exploits spatial diversity to improve spectral efficiency. A prototype point-to-point link is constructed and a simulation model is derived based on measurements. Spatial discrete multitone modulation is defined and shown to be a convenient method of mitigating channel impairments. Multi-level codes coupled with multi-stage decoders are applied to this channel and are shown to provide spectral efficiencies of approximately 1.82 kbits/s/Hz using 512x512 transmit pixels (0.24 dpi) and 154x154 receive pixels (10x10um) over a range of 2 m.