The predicted growth in mobile phone traffic and the move towards enhanced mobility will lead to a need for a wireless infrastructure that provides increasing bandwidth per user. It is envisaged that our world will become increasingly interconnected, with mobile communications enabling us to perform an increasing range of tasks. / Future wireless networks will require an optical network to provide antenna Base Stations with sufficient bandwidth to provide individual users with a larger bandwidth. The combined optical and wireless network is referred to as a “fibre-radio” or “radio-over-fibre” or “fibre-wireless”; network. It is expected that such high-capacity networks will use Wavelength Division Multiplexing (WDM) to increase the total bandwidth transmitted over the optical access network. Such a high-capacity network would not be achievable using a single wavelength or using a copper or coax network. Optical crosstalk is present in WDM optical networks and degrades the received signal quality, increasing the bit-error-rate. Two types of crosstalk occur, depending on whether the crosstalk channel is a different wavelength to the signal or at the same wavelength (out-of-band and in-band crosstalk, respectively). An important consideration for fibre-radio networks is whether or not the optical network transports data at baseband, using standard intensity modulation, or at an RF frequency, using subcarrier modulation. The nature of the optical modulation scheme has implications for the design of the Central Office and the Base Stations, and potentially for optical crosstalk. (For complete abstract open document)

Performance requirements of short-distance digital communication links have increased considerably with the escalating demand for high speed and high density data links. The high aggregate bandwidth and channel density achievable by free-space optical interconnects (FSOIs) make them ideal replacement for electrical interconnection schemes. Optical interconnects potentially have low power consumption, and can facilitate the development of radically novel designs for VLSI architectures including heterogeneous multiprocessor systems, and highly parallel computing systems. Recent developments in the integration of Vertical-Cavity Surface-Emitting Laser (VCSEL) arrays and photodetector arrays with CMOS electronic circuitry have increased the practical potential of optical interconnects. However, VCSELs tend to operate in several transverse modes simultaneously, which will degrade the performance of FSOIs. Experimental investigation was performed to evaluate the operation characteristics and the intensity noise in VCSELs. The measurement results were later combined with optical simulations to analyse the effect of optical crosstalk in free-space optical interconnects. The VCSEL characterization included light-current-voltage relationships, relative intensity noise, modal spectral composition and modal beam profiles. The optical system simulation software (Code V) was used to simulate optical crosstalks in the FSOI system. Experimentally measured spectrally-resolved near-field images of VCSEL higher order modes were used as extended sources in the proposed simulation model. The simulation was performed using a combination of exact ray-tracing and the beam propagation method. A new type of crosstalk referred to as the stray-light crosstalk (SLC) was introduced. This type of crosstalk is caused by the overfill of the transmitter microlens by the VCSEL beam. It was discovered that part of the signal was imaged by the adjacent microlens to another channel, possibly far from the intended one. The simulation showed that the SLC is strongly dependent on the fill factor of the microlens, array pitch, and the channel density of the system. When comparing the diffraction-caused crosstalk (DCC) to SLC, an increase in the interconnection distance has little influence on the SLC. A simple behavioural model was developed which accurately approximates the crosstalk noise for a range of optical sources and interconnect configurations. The effect of transmitter and receiver array configurations on the performance of FSOIs was investigated. Our results demonstrate the importance of SLC in both square and hexagonal configuration. By changing the array lattice geometry from square to a hexagonal, we obtained an overall optical signal-to-noise ratio (SNR) improvement of 3 dB. The optical SNR is optimal for the hexagonal channel arrangement regardless of the transverse mode structure of the VCSEL beam was shown. Furthermore, the VCSEL drive current required for the best performance of the FSOI system was determined. The optimal focal length of the transmitter microlens array which maximises the SNR by minimising the combined effects of DCC and SLC was determined. Our results show that shorter focal length needs to be used for higher order modes to obtain optimal SNR in an FSOI system.

Performance requirements of short-distance digital communication links have increased considerably with the escalating demand for high speed and high density data links. The high aggregate bandwidth and channel density achievable by free-space optical interconnects (FSOIs) make them ideal replacement for electrical interconnection schemes. Optical interconnects potentially have low power consumption, and can facilitate the development of radically novel designs for VLSI architectures including heterogeneous multiprocessor systems, and highly parallel computing systems. Recent developments in the integration of Vertical-Cavity Surface-Emitting Laser (VCSEL) arrays and photodetector arrays with CMOS electronic circuitry have increased the practical potential of optical interconnects. However, VCSELs tend to operate in several transverse modes simultaneously, which will degrade the performance of FSOIs. Experimental investigation was performed to evaluate the operation characteristics and the intensity noise in VCSELs. The measurement results were later combined with optical simulations to analyse the effect of optical crosstalk in free-space optical interconnects. The VCSEL characterization included light-current-voltage relationships, relative intensity noise, modal spectral composition and modal beam profiles. The optical system simulation software (Code V) was used to simulate optical crosstalks in the FSOI system. Experimentally measured spectrally-resolved near-field images of VCSEL higher order modes were used as extended sources in the proposed simulation model. The simulation was performed using a combination of exact ray-tracing and the beam propagation method. A new type of crosstalk referred to as the stray-light crosstalk (SLC) was introduced. This type of crosstalk is caused by the overfill of the transmitter microlens by the VCSEL beam. It was discovered that part of the signal was imaged by the adjacent microlens to another channel, possibly far from the intended one. The simulation showed that the SLC is strongly dependent on the fill factor of the microlens, array pitch, and the channel density of the system. When comparing the diffraction-caused crosstalk (DCC) to SLC, an increase in the interconnection distance has little influence on the SLC. A simple behavioural model was developed which accurately approximates the crosstalk noise for a range of optical sources and interconnect configurations. The effect of transmitter and receiver array configurations on the performance of FSOIs was investigated. Our results demonstrate the importance of SLC in both square and hexagonal configuration. By changing the array lattice geometry from square to a hexagonal, we obtained an overall optical signal-to-noise ratio (SNR) improvement of 3 dB. The optical SNR is optimal for the hexagonal channel arrangement regardless of the transverse mode structure of the VCSEL beam was shown. Furthermore, the VCSEL drive current required for the best performance of the FSOI system was determined. The optimal focal length of the transmitter microlens array which maximises the SNR by minimising the combined effects of DCC and SLC was determined. Our results show that shorter focal length needs to be used for higher order modes to obtain optimal SNR in an FSOI system.

How can we evaluate and design integrated photonic circuit performance systematically? Can active photonic circuits be controlled for optimized performance? This work uses control theory to analyze, design, and optimize photonic integrated circuits based on versatile semiconductor optical amplifiers (SOAs). Control theory provides a mathematically robust set of tools for system analysis, design, and control. Although control theory is a rich and well-developed field, its application to the analysis and design of photonic circuits is not widespread. Following control theoretic methods already used for fibreline systems we derive three interrelated state-space models: a core photonic model, a photonic model with gain compression, and a equivalent circuit optoelectronic model. We validate each model and calibrate the gain compression model by pump/probe experiments. We then linearize the state-space models to design and analyze SOA controllers. We apply each linearized model to proof-of-concept SOA control applications such as suppressing interchannel crosstalk and regulating output power. We demonstrate the power of linearized state-space models in controller design and stability analysis. To illustrate the importance of using the complete equivalent circuit model in controller design, we demonstrate an intuitive bias-current controller that fails due to the dynamics of the intervening parasitic circuitry of the SOA. We use the linearized state-space models to map a relationship between feedback delay and controller strength for stable operation, and demonstrate that SOAs pose unusual control difficulties due to their ultrafast dynamics. Finally, we leverage the linearized models to design a novel and successful hybrid controller that uses one SOA to control another via feedback (for reliability) and feedforward (for speed) control. The feedback controller takes full advantage of the equivalent circuit modelling by sampling the voltage of the controlled SOA and using the error to drive the bias current of the controller SOA. Filtering in the feedback path is specified by transfer function analysis. The feedforward design uses a novel application of the linearized models to set the controller bias points correctly. The modelling and design framework we develop is entirely general and opens the way to the robust optoelectronic control of integrated photonic circuits.

How can we evaluate and design integrated photonic circuit performance systematically? Can active photonic circuits be controlled for optimized performance? This work uses control theory to analyze, design, and optimize photonic integrated circuits based on versatile semiconductor optical amplifiers (SOAs). Control theory provides a mathematically robust set of tools for system analysis, design, and control. Although control theory is a rich and well-developed field, its application to the analysis and design of photonic circuits is not widespread. Following control theoretic methods already used for fibreline systems we derive three interrelated state-space models: a core photonic model, a photonic model with gain compression, and a equivalent circuit optoelectronic model. We validate each model and calibrate the gain compression model by pump/probe experiments. We then linearize the state-space models to design and analyze SOA controllers. We apply each linearized model to proof-of-concept SOA control applications such as suppressing interchannel crosstalk and regulating output power. We demonstrate the power of linearized state-space models in controller design and stability analysis. To illustrate the importance of using the complete equivalent circuit model in controller design, we demonstrate an intuitive bias-current controller that fails due to the dynamics of the intervening parasitic circuitry of the SOA. We use the linearized state-space models to map a relationship between feedback delay and controller strength for stable operation, and demonstrate that SOAs pose unusual control difficulties due to their ultrafast dynamics. Finally, we leverage the linearized models to design a novel and successful hybrid controller that uses one SOA to control another via feedback (for reliability) and feedforward (for speed) control. The feedback controller takes full advantage of the equivalent circuit modelling by sampling the voltage of the controlled SOA and using the error to drive the bias current of the controller SOA. Filtering in the feedback path is specified by transfer function analysis. The feedforward design uses a novel application of the linearized models to set the controller bias points correctly. The modelling and design framework we develop is entirely general and opens the way to the robust optoelectronic control of integrated photonic circuits.