As communication services applications continue to grow in number (e.g. Twitter, YouTube, Facebook, etc.) and in bandwidth (e.g. HDTV, 3D,...), all parts of the communication systems carrying this traffic must be able to operate at higher and higher speeds. This ever-growing capacity demand can only be handled by continually upgrading the capacity of all parts of the network, including long-haul links between major cities, as well as the 'last mile' distribution networks ending at or near the customer premises. As part of this upgrading of the physical layer of optical communication systems, there is increasing pressure to provide more optical bandwidth to accommodate more individual wavelength carriers in high-capacity wavelength-division multiplexed (WDM) systems. The current type of optical amplifiers, namely Erbium-doped fibre amplifiers (EDFAs) were introduced in the early 1990s, but have a fixed bandwidth of the order of 35 nm, covering the so-called C-band (1530-1565 nm). This bandwidth is being rapidly exhausted, and so there is a need for introducing novel optical amplifiers with substantially larger bandwidth. In addition, there is also a trend toward using high-spectral efficiency modulation formats (e.g. quadrature amplitude modulation, or QAM). However, such formats require high optical signal to noise ratios (OSNRs). We propose to investigate the suitability of fibre optical parametric amplifiers (OPAs) to amplify WDM optical communication signals in wideband optical communication systems. We will investigate both phase-insensitive OPAs (PIAs) and phase-sensitive OPAs (PSAs).The latter are particularly attractive because of their potential for noiseless amplification, which cannot be achieved with EDFAs or phase-insensitive OPAs. The project will consist of three phases with the following objectives: Phase I (12 months). We will first demonstrate phase-insensitive OPAs (PIAs) with an optical bandwidth matching that of EDFAs. These will be tested in a recirculating loop, with a fully-populated WDM signal spectrum, simulating propagation in a long-haul system. Only the signals will be used; the idlers will be discarded after each OPA. It is expected that the reach of the system will be several thousand kilometres. Different modulation formats will be tested, with baud rates up to 43.7 Gb/s. Aggregate throughput will reach several terabits per second. Phase II (12 months). We will then use signals and idlers in an alternating manner in the recirculating loop. This will allow us to exploit the wavelength conversion/phase conjugation aspects of OPAs to combat dispersion as well as some nonlinear effects. Testing will be done with a wider fully populated CWDM spectrum, at a higher aggregate rate. Phase III (12 months). We will use phase-sensitive OPAs (PSAs), which have the potential for lossless amplification, leading to an increase in system reach. We will investigate the suitability of propagation along principal states of polarization, in order to maintain the states of polarization of signals, idlers, and pump, necessary for optimum PSA operation. If the project is successful, it will demonstrate that fibre OPAs are indeed a potential contender for providing optical amplification over wavelength ranges exceeding that of EDFAs, and in a nearly noiseless manner, which is compatible with use in either long-haul or distribution optical communication networks. Hence they could in principle provide the next generation of optical amplifiers for future high-capacity optical networks.

As communication services applications continue to grow in number (e.g. Twitter, YouTube, Facebook, etc.) and in bandwidth (e.g. HDTV, 3D,...), all parts of the communication systems carrying this traffic must be able to operate at higher and higher speeds. This ever-growing capacity demand can only be handled by continually upgrading the capacity of all parts of the network, including long-haul links between major cities, as well as the 'last mile' distribution networks ending at or near the customer premises. As part of this upgrading of the physical layer of optical communication systems, there is increasing pressure to provide more optical bandwidth to accommodate more individual wavelength carriers in high-capacity wavelength-division multiplexed (WDM) systems. The current type of optical amplifiers, namely Erbium-doped fibre amplifiers (EDFAs) were introduced in the early 1990s, but have a fixed bandwidth of the order of 35 nm, covering the so-called C-band (1530-1565 nm). This bandwidth is being rapidly exhausted, and so there is a need for introducing novel optical amplifiers with substantially larger bandwidth. In addition, there is also a trend toward using high-spectral efficiency modulation formats (e.g. quadrature amplitude modulation, or QAM). However, such formats require high optical signal to noise ratios (OSNRs). We propose to investigate the suitability of fibre optical parametric amplifiers (OPAs) to amplify WDM optical communication signals in wideband optical communication systems. We will investigate both phase-insensitive OPAs (PIAs) and phase-sensitive OPAs (PSAs).The latter are particularly attractive because of their potential for noiseless amplification, which cannot be achieved with EDFAs or phase-insensitive OPAs. The project will consist of three phases with the following objectives: Phase I (12 months). We will first demonstrate phase-insensitive OPAs (PIAs) with an optical bandwidth matching that of EDFAs. These will be tested in a recirculating loop, with a fully-populated WDM signal spectrum, simulating propagation in a long-haul system. Only the signals will be used; the idlers will be discarded after each OPA. It is expected that the reach of the system will be several thousand kilometres. Different modulation formats will be tested, with baud rates up to 43.7 Gb/s. Aggregate throughput will reach several terabits per second. Phase II (12 months). We will then use signals and idlers in an alternating manner in the recirculating loop. This will allow us to exploit the wavelength conversion/phase conjugation aspects of OPAs to combat dispersion as well as some nonlinear effects. Testing will be done with a wider fully populated CWDM spectrum, at a higher aggregate rate. Phase III (12 months). We will use phase-sensitive OPAs (PSAs), which have the potential for lossless amplification, leading to an increase in system reach. We will investigate the suitability of propagation along principal states of polarization, in order to maintain the states of polarization of signals, idlers, and pump, necessary for optimum PSA operation. If the project is successful, it will demonstrate that fibre OPAs are indeed a potential contender for providing optical amplification over wavelength ranges exceeding that of EDFAs, and in a nearly noiseless manner, which is compatible with use in either long-haul or distribution optical communication networks. Hence they could in principle provide the next generation of optical amplifiers for future high-capacity optical networks.

The optical fibre core network underpins the internet and the digital economy, with the present capacity of today's core networks being limited to ~ 1Tbit/s per fibre. While in current networks, the limited broadband data rates afforded by the copper based access network prevents the optical core network from being stretched to capacity, as optical fibre permeates the access network, the bottleneck will move from the access network to the core network. To overcome these limitations and to maximise the opportunities afforded by a fibre optic access network will require the capacity of the installed core network to be increased, either by increasing the number of wavelengths used or by increasing the data rate per wavelength. The proposed research aims to combine both techniques simultaneously - transmitting 100 gigabit Ethernet (GbE) on each wavelength, while employing wavelength division multiplexing (WDM) to increase the capacity of the core network to beyond 10Tbit/s.Using conventional intensity modulation schemes, much of the installed fibre base is unable to support data rates faster than 10Gbit/s due to imperfections in the installed fibre which causes pulse spreading. Current research at UCL, led by the principal investigator (PI), has recently experimentally demonstrated the potential of digital signal processing (DSP) combined with coherent detection of spectrally efficient modulation formats to overcome these limitations for 40Gbit/s transmission systems, with the same principles being equally applicable to 100GbE systems. In a digital coherent receiver the four components of the optical field, the in-phase and quadrature components of the two polarisations, are mapped into the electrical domain. This allows digital compensation of transmission impairments and the use of spectrally efficient four-dimensional modulation formats. Given the huge investment which has been made into installing the fibre base infrastructure, the ultimate aim of the research is to determine how this four-dimensional modulation space can be used in conjunction with DSP to maximise the capacity of the installed fibre.The proposed research combines fundamental theoretical research with a determinedly experimental research program into the nonlinear transmission of four-dimensional modulation formats at 100Gbit/s+ and beyond. The initial workpackage will investigate both experimentally and theoretically quadrature amplitude modulation, in combination with polarisation division multiplexing as a four dimensional modulation scheme for 100GbE transmission systems. Within this first workpackage, the system under investigation will be receiver centric, such that all of the DSP, both linear and nonlinear, is based at the receiver. In the second workpackage this assumption will be relaxed and combined transmitter and receiver DSP will be investigated, both experimentally and through simulation. The third and final workpackage which is a theoretical study, will draw on the conclusions of the previous workpackages, and will aim to answer the question Given the optical fibre is dispersive and nonlinear, what is the optimal modulation scheme which enables the capacity of the core network to be maximised assuming we are able to employ appropriate digital signal processing?

To meet the demands of the internet to transmit large volumes of data over long distances, information is sent as short pulses of light. The photodetector which receives this information must have high sensitivity, a fast response, and low levels of 'noise' (random spurious signals). Photodetectors can even be made sensitive enough to detect single photons, and 'photon counting' is an important technique in many applications including sequencing the human genome and quantum computing. Most high-sensitivity photodetectors are semiconductor avalanche photodiodes (APDs): semiconductor materials are robust, cheap, compact, and efficient, while APDs make use of an effect where a very weak signal can trigger a very large current flow (like a single snowflake setting off a massive avalanche of snow).There are many different semiconducting materials, and each is sensitive to a different colour of light or wavelength. While silicon works really well as an APD, it doesn't detect infrared light at the wavelengths needed for optical communications and other applications. We can use combinations of material - one to absorb the light and one to do the avalanche multiplication - but it can be tricky getting the signal across from one material to the other. So APDs are hard to make and therefore expensive. We are going to make new types of APDs with the performance of silicon but sensitive to infrared light, which are also easier/cheaper to make than existing infrared detectors. Firstly, we are going to use a relatively new type of semiconductor (a 'dilute nitride') as the absorbing layer. Dilute nitrides are completely different from other materials: adding a small amount of nitrogen to a conventional semiconductor like gallium arsenide has a huge effect on the properties and can make it sensitive to infrared light. Dilute nitrides even seem to be less noisy than other absorbing layers, since their special properties suppress a source of noise which comes from quantum mechanical tunneling (electrons feel 'heavier' in dilute nitrides and find it harder to tunnel through barriers).Secondly, we are going to replace the conventional multiplication layer made of indium phosphide or gallium arsenide, which compared to multiplication layers made of silicon are rather noisy. The noise comes because multiplication is random: we know the probability that multiplication will occur within a certain time, but not exactly when it will occur. The particular electronic properties of dilute nitrides means that electrons in one energy band (the valance band) can easily trigger avalanches, while electrons in another band (the conduction band) should find it very hard. This situation should lead to very low multiplication noise, perhaps even as low as silicon, and has never been studied before.There is a lot of interesting physics in the movement of electrons in dilute nitride semiconductors, and in the statistics of avalanche multiplication in thin layers. We will use specialized techniques to study these, including squeezing the material under very high pressures to change its properties. This will give us the understanding we need to produce better high-sensitivity light detectors, which are useful for communications, medicine, pollution monitoring, and many other areas that affect our daily lives.

Information processing and communications enabled by advances in semiconductor technology are at the heart of the modern interconnected and application-driven world. Modern society has an enormous appetite for new platforms and services and meeting these demands places a considerable burden on device and systems development. Over the last 50 years, semiconductor manufacturing has met these demands through a scaling of device size to ever smaller dimensions. As a result, we now approach the true nanoscale regime and seek devices of size less than 10nm. The industry is however facing enormous technological and physical challenges to work at this precise scale, equivalent to only a few atomic layers. Yet with these challenges comes also enormous potential from emerging quantum device approaches which could dramatically increase in calculation capability, dramatically improve the security of data and to do this simultaneously with lower energy costs. Our well used semiconductor device production processes, based on epitaxy, patterning and etch will struggle to turn the promise of quantum technologies into manufacturable commercial devices. In contrast, we can grow naturally 'self assembled' structures with nanometer dimensions and from such materials we have extensively demonstrated quantum interactions. However self-assembly has an Achilles heel in that we cannot control the site or the dimensions because of random nucleation. As a result we cannot predict where the nanostructure is located nor its energy state. Unsurprisingly there has been very little development in terms of manufacturable devices utilising quantum technologies. What we need is an approach which combines the best aspects of patterning and self-assembly. The approach is directed (or site-controlled) self-assembly which uses lithography to define the site and then exploits self-assembly to produce the nanostructure. Structuring with light is the manufacturing technology of the 21st century. Many products now involve cutting, milling, surface processing, sealing etc processes using laser light. Our approach seeks to exploit the capabilities of light at much smaller dimensions, specifically its capability to create regular patterns on a very small stage through the optical interference process. We will design and build a system in which laser interference interacts with semiconductor growth to create a single step in-situ manufacturing route which is free of all major limitations of conventional high cost, low throughput nanostructuring approaches. We will build and demonstrate a custom instrument in which an interference pattern from laser interference interacts with the semiconductor growth surface to nucleate self-assembled growth on a regular grid pattern. Such an arrangement is a key requirement for developing electronic and photonic circuits based on arrays of single nanostructures. The method has the further advantage of precisely controlling assembly such that the array contains identical nanostructures in terms of size, shape and electronic properties. Using this approach we will create large area state of the art quantum dot and quantum wire arrays which are essential building blocks for the semiconductor devices of the future, enabling diverse applications including electronics, photonics, sensing and biomedicine.