This paper reviews the use of fiber as a medium for generating and distributing millimeter waves and outlines the results achieved using a recently published new method. Following this introduction the three basic methods of fiber based generation and distribution of millimeter waves are described. A discussion of possible system topologies and applications is given, followed by a more in-depth review of recent literature addressing this area. A detailed description of a novel method followed by a comparison of the key features of each of the methods described conclude the paper.

A tuned SCM optical receiver, integrated with a serial tuning element as a GaAs MMIC, is reported. The receiver is designed for a specific system application and optimized to operate over the range 1.8 to 2.2 GHz. Predicted results indicate low noise and high gain performance and verify the validity of the design technique adopted.

Within the RACE project 1051 `multigigabit transmission in the IBCN subscriber loop' the major technologies for 5 and 10 Gbit/s fiber-optic transmission systems have been developed. Components and building blocks for all system functions like multiplexer, laser driver, receiver preamplifier, main amplifier, regenerator and demultiplexer/channel selector have been realized in the form of silicon bipolar chips for both bitrates. Integrated transmitter and receiver modules have been built around advanced optoelectronic components. With a cascade of 4 optical amplifiers it was possible to bridge a budget of 100 dB at 10 Gbit/s without applying optical filters. This means that more than 8 million customers can be supplied by only one optical transmitter.

This paper commences with a brief review of wide dynamic range optical receiver design techniques by discussing their relative merits and drawbacks in relation to the other major receiver performance parameters. In particular, the review focuses on the use of the PIN photodiode detector incorporated within the receiver rather than the APD to provide for the most economic design solutions. This is followed by discussion of a novel PIN-FET optical receiver design which utilizes a degenerative feedback loop in order to extend the dynamic range performance. The implementation of this receiver design strategy, using thick film technology, is then described in order to optimize the device dynamic range. Finally, the major performance characteristics of the optical receiver are reported, which indicate a minimum dynamic range of 40 dB over a bandwidth of 3 MHz.

Much interest is currently being shown in the design of very low noise tuned front-end optical receivers for use in lightwave subcarrier multiplex (SCM) systems. This paper is concerned with the design and realization of such receivers using GaAs MMIC technology. The use of such technology, while desirable as it enables close control to be exercised over tuning element values, does complicate the design process, as well as imposing certain constraints upon noise performance of the amplifier. It is these areas that this paper addresses. Here a generalized design approach is presented which is then applied to the design and analysis of a 6 GHz parallel tuned front-end receiver to be realized using the GEC/Marconi F20 foundry process.

This paper describes a range of methods, developed at the University of Wales, that affect signal shaping for high bit rate optical communications. All the signal designs presented here aim to offer optimum performance in terms of receiver sensitivity for a broad range of system impairments and are also designed to be tolerant to variations in these impairments. This makes the filters ideally suited for use in system applications where the actual impairments cannot be accurately specified. The designs described here are particularly appropriate for systems dominated by signal dependent noise such as systems based on avalanche photodiodes, optically pre-amplified receivers and soliton systems. The circuits described here are all implemented as GaAs monolithic microwave integrated circuits (MMICs) to further enhance their applicability to systems that are planned for field deployment.

A balanced detector receiver using resonance optical front-end has been presented. The performance of five kinds of resonance optical front-ends has been analyzed and compared with each other. The receiver using mixed tuning resonance optical front-end has a bandwidth of 4.16 GHz and very low noise performance. A sensitivity of -34.2 dBm (for bit error ratio of 10 ^-9) has been achieved in a 1.1 Gbit/s FSK heterodyne optical communication system.

The principles of the transmission of high quality high definition (HDTV) signals in single- mode optical fiber systems are introduced. This is followed by a review of eight modulation systems with particular emphasis on their suitability for the transmission of wideband signals. Theoretical comparisons are drawn between the systems in terms of signal to noise ratio (SNR) performance, transmission bandwidth, receiver cost and complexity and optical source linearity requirements. Frequency modulation (FM) based systems are then proposed for practical implementations and subcarrier frequency modulation (SCFM) circuit, using a hyperabrupt varactor tuned oscillator (HTO) to achieve the required frequency deviation, is described. This system is then experimentally compared with a square wave frequency modulation (SWFM) circuit configuration in terms of receiver sensitivity and subjectively assessed picture quality. Although the SCFM system performance exhibited a slight improvement over the SWFM system the latter was not fully optimized being derived from the SCFM implementation.

A novel low-cost multi-channel optical transmission system for video signals is described. The system uses time-division multiplex pulse-position modulation (TDPPM). The complete transmitter and receiver are realized in a standard bipolar process while an external low-cost laser and PIN photodiode are used. A maximum of 16 channels can be multiplexed, and a semi-professional quality with a weighted signal-to-noise ratio (SNR) of 54 dB is achieved. For a launched power of 0 dBm (pulse amplitude), the optical budget for a 45 dB weighted SNR exceeds 20 dB. The feasibility of a four-channel system has been confirmed by measurements.

Application of classical control techniques to a wideband fiber data transmission system employing pulse time modulation (PTM) has resulted in a significant improvement to dc performance. The method is applicable to many other high-speed data systems requiring improved dc behavior.

In this paper we present experimental results for a digital pulse position modulation (PPM) system that uses a low-bandwidth receiver. It is shown that the performance of such a system is relatively independent of channel dispersion. Thus a simple low-bandwidth receiver can be used on low- or high-dispersion digital PPM links.

Recent years progress in video compression has made it possible to achieve HDTV-quality in bit-rates below 30 Mbit/s. This means that a few channels may be distributed over the ether but it will also affect the way of distributing video over fibers. A question of importance is whether some kind of subcarrier multiplexed scheme or a digital version is most appropriate for the distribution. A theoretical comparison has been done between these two different methods for the fiber transmission: quadrature amplitude modulation (QAM) with subcarrier multiplexing (SCM) and binary shift keying (BSK) with time division multiplexing (TDM). QAM over subcarrier may be more flexible and more compatible with the existing CATV- systems for PAL or NTSC distribution. However, TDM makes it more easy to interact with other network levels over SDH and/or ATM. The result shows that a 20 channel HDTV distribution gains about 9 dB in power budget using TDM compared to SCM when thermal noise is the limiting factor. It means that TDM can be distributed to approximately 10 times more end users.

Simplicity, low cost circuitry, and good performance makes pulse time modulation (PTM) transmission an attractive alternative to counterparts in many applications. PTM transmission presents the system designer with a broad selection of modulation options, each with different transmission properties and multiplexing capabilities. This paper reports an experimental hybrid (or combined) multiplexing technique over optical fiber cable at 1.3 micrometers .

Self synchronization of the optical fiber n-ary PPM format entails acquiring the slot, frame frequency, and frame phase information directly from the PPM stream. Processing the PPM power spectral density can result in extracting the slot and frame frequency leaving an n^th degree frame phase ambiguity to be resolved. We identify two classes of naturally occurring PPM sequences and present a novel technique for acquiring the frame phase directly from the incoming data stream. The technique is superior to previous methods since it does not involve added code redundancy, loss in sensitivity or bit rate. The frame phase synchronizers are shown to be upgradable and thus capable of coping with the required PPM word size. Original theoretical and practical results are reported for the synchronizers. Tracking the first two additional sequences is demonstrated and shown to offer 10.4 dB improvement in the discrete frame spectral component joined with 8.4 X 10^-3 rad² reduction in the timing variance.

n-ary pulse position modulation (PPM) has been shown to be an effective modulation format for optical fiber communications in that it can be used to exchange excess fiber bandwidth for receiver sensitivity. The combination of this modulation technique with optical preamplification should lead to further improvements in receiver sensitivity. Here we present a theoretical analysis for such a combination and demonstrate that the incorporation of optical preamplification leads to a sensitivity improvement of 12.5 dB. The sensitivity of an equivalent PCM system was calculated to be 110 photons/bit compared to 33 photons/bit for a PPM system employing moderate operating parameters. This represents an improvement of 5 dB and furthermore demonstrates that an optically preamplified PPM system can offer receiver sensitivities beyond the fundamental limit of an optically preamplified PCM system.

In this paper we report the design and implementation of bit synchronizers for multigigabit transmission systems. We present the main features of dielectric resonators (DRs) and considerations to take into account when using them as filters for clock recovery circuits. We address the problem of using balanced and unbalanced nonlinear circuits (NLs) for which we perform a theoretical analysis of the full-wave and truncated squarer non-linearities. Finally we describe the design of a prototype for a 10 Gbit/s system and report measurements obtained with it.

Bandwidth requirements for computer communications have increased dramatically in recent years. This growth in bandwidth requirements has caused many computer users to look toward fiber optic communication systems. This paper attempts to resolve confusion in the marketplace with regard to bandwidth requirements, transmission methodologies, and fiber optic cable selection.

The ever increasing consumer demands on the present communication infrastructure has led to an interest in multigigabit long haul transmission systems. Techniques such as wavelength division multiplexing (WDM) are employed to increase the present channel capacity while optical amplifiers are used to extend the maximum attainable fiber reach. We have been investigating the effectiveness of pseudo-lumped element group delay equalizer realizations on gallium arsenide (GaAs), drawing specifically on the system implications of such networks with various modulation formats. Although designed to correct for a given length of fiber, the equalizers are evaluated for larger fiber spans to illustrate a measure of their robustness.

An electrical compensation network has been developed which preshapes the injection current modulation signal waveform to suppress both the low-frequency enhancement and relaxation oscillation of DFB-LD chip. The DFB-LD modules with electrical compensation network and optical isolators have been achieved with 4.7 GHz FM bandwidth.

The performance and efficiency of a praseodymium-doped La-Ga-S chalcogenide glass fiber amplifier have been analyzed using a comprehensive large-signal numerical model. The four level laser model incorporates the LP₀₁ mode radial field distribution in step-index single- mode fiber and experimentally obtained emission and absorption cross sections. The model has been used to optimize the fiber parameters for small-signal operation. The optimum cut-off wavelength of the LP₁₁ mode has been found to be 700 nm. It follows from the analysis that at 1310 nm a small-signal gain of 36.7 dB can be expected from the Pr³⁺-doped chalcogenide fiber amplifier for a pump power of 200 mW in contrast to 18.8 dB for the same pump power in the Pr³⁺-doped fluoride fiber amplifier.

An improved model of an optical phase-locked loop (OPLL) is presented. Based on this model and rate equations of an external cavity semiconductor laser, the effects of OPLL feedback on the FM noise and the AM noise of the external cavity semiconductor laser are discussed. A new linewidth formula of the local laser is derived.

A novel method for the channel interval stabilization in optical fiber networks has been presented and demonstrated experimentally using the optical phase locking technique. Semiconductor transmitter lasers are phase-locked to 6 consequent sidebands of a phase- modulated He-Ne gas master laser with a frequency stability better than 300 kHz.

This paper summarizes the wavelength allocation of services for Integrated Broadband Communication (IBC) Networks and the status of wavelength standardization in CCITT, ETSI, and RACE.

Wavelength division multiple access (WDMA) as a technique for achieving broadband communications in the future optical access network is introduced. This is followed by a brief review of WDMA protocol strategies which have been proposed in order to facilitate high capacity network throughput. In particular, new efficient WDMA demand assignment protocols, suitable for operation in such a high speed environment, are presented and their performance investigated using discrete event computer simulation techniques. It is demonstrated that protocols of this type could provide for broadband communications in the optical access network. Under symmetric traffic loading, the WDMA protocol performances are shown to exhibit an improvement over the single channel case which is particularly significant at higher transmission rates. It is then demonstrated that the WDMA protocol performances are much less sensitive to the degradation caused by asymmetric traffic loads resulting in a substantial improvement in the maximum throughput achievable. Hence, the potential capacity of this broadband strategy for the optical access network is explored.

A photonic packet switching device is designed for WDM based optical slotted ring networks which are suitable for B-ISDN access network and LAN/MAN. For the design, we devised a fiber optic delay line matched filter as a means of address filtering at the destination node and developed an efficient address decoding algorithm to accommodate the maximum available number of nodes in the ring network. Based on the algorithm, the maximum available number of node addresses in the ring is 3,432 for the case of 16-bit address. The design objectives of the photonic packet switching device are to eliminate the optical-to-electrical conversion at the address processing, to remove optical data from the ring at the destination nodes so that the slots can be reused a number of times in a rotation, and to be compatible with B-ISDN based on ATM concerning services, bit rate, interface, and switching node. The network architecture using the photonic packet switching device is also described. The simulation results show that the major part of the cell transfer delay in the network is coming from the propagation delay from a source to a destination. It was also observed that the average transfer delay characteristic of the network depends upon the aggregated transmission capacity of the network.

The research summarized in this paper is part of a work undertaken for the RACE (Research into Advanced Communications for Europe) program project 1044. It investigates the prospects for upgrading existing cable TV networks to an infrastructure capable of providing jointly cable TV and telecoms services. This is a topical issue as services are now being developed which blur the distinction between cable TV and telecoms services, and in some countries the broadcasting and telecommunications industries are coming closer together. Technological development is leading regulation in the sense that, while in technological terms broadcasting and telecommunications services are approaching each other, in many countries the regulation of the two industries is still staunchly separate. However, in the Netherlands and Belgium there are signs that this situation is changing and the European Commission may also facilitate the provision of some telecoms services over cable TV networks through its Open Network Provision program.

PTT Telecom, the Netherlands, will shortly introduce fiber optics on a large scale into the primary access network. Due to the large investment costs, a study has been done on the most cost effective structure for fiber optics in the primary access network. Cable costs, trench costs, and cable installation costs have been included in the cost function for structure optimization. Availability and transmission technology are the constraints that have been taken into account. A mathematical model is presented that is sufficiently flexible to encompass obstacles such as rivers and highways and the previously mentioned constraints. We present an optimization method for the implementation of structures. In a case study, we compare the chosen structures for various situations and we make a cost comparison of dual homing and double routing. We find that there is a significant cost difference between the structures. The most cost effective solution is an access network implemented without any structure constraint. Furthermore, we conclude that double routing requires higher investment costs than dual homing.

The asynchronous transfer mode ring protocol (ATMR), which is proposed to regulate multiple terminals within the Customer Premises Network (CPN) accessing the B-ISDN through a single user-network-interface (UNI), is described and its performance is analyzed. The mean packet access delay and the fair sharing of the non-guaranteed bandwidth at various levels of network load are evaluated by a discrete event simulation. Window size and reset period are the key parameters of the ATMR cycle-reset type access protocol. The simulation results show that depending on the choice of the window size the mean packet access delay can be optimized. The fairness among the stations in terms of mean packet access delay can be guaranteed under the examined load conditions. The results also demonstrate that the mean packet access delay increases according to the network size.

A software package is described which has been developed to assist in the process of both design and planning associated with passive optical access networks. The package constitutes a set of basic tools which enable the complete characterization of a number of key technical parameters of optical access networks. The optical networks which can be configured and analyzed using the package are specified visually on a design location and can adopt any topology. Hence a completely general interconnection of nodes and links is permitted. Furthermore an optical access network consisting of several independent networks can also be analyzed. The analytical output of the package consists of a complete spectral power budget specification along any path in the network, the optical return losses on any path, an optical time domain reflectometer (OTDR) display, and the spectral dependence of both the power budget and the return loss.

The Broadband Access Facility (BAF) is an affordable method of providing broadband services for the local loop. It is based on the transport of ATM Cells over a shared fiber passive optical network (PON) with a passive optical tree and branch topology. The PON supports 622 Mbits/s over 32 network terminations (BAF-NTs) using a passive optical splitter/combiner. The management architecture of the BAF is based on the CCITT Telecommunications Management Network (TMN) recommendations. The management and control requirements have been analyzed using the OSI decomposition into five functional areas: configuration management (controls the `ranging' procedure which compensates for different distances between the BAF-NTs and the BAF-LT); performance management (manages the allocation of bandwidth between the subscribers to provide statistical gain and monitors network parameters to measure how well the BAF MAC protocol provides the agreed service); fault management (set of functions which enables the detection, isolation, correction of abnormal operation); security management (supports encryption cells which is necessary due to their broadcasting to all the BAF-NTs); and finally, accounting functions.

Multiplexing of bursty sources and refined congestion control strategies are still the subject of numerous research activities. Broadband applications with very high peak-to-mean bitrate ratio and long silence periods like still picture video gave rise to different ideas of rate control at the B-ISDN network access. Contributions on Input Rate Control for source coded traffic as well as on Server Rate Control within a LAN/ATM Interworking Unit have recently been presented. This paper addresses a congestion avoidance strategy at the network access regarding the aggregated traffic of bursty sources. Depending on the number of active sources as well as on certain defined congestion levels the cell rate at the network access is controlled. The proposed analytical approach is based on the model of uniform and continuous arrival and service. The selected underlying Markov chain contains `split' states in order to handle the congestion correlation. The proposed model is extended to an adaptive Non-Markov system where the buffer filling level is evaluated using a switching hysteresis. This type of congestion measurement turns out to be very useful for an adaptive rate control mechanism that guarantees a certain quality of service while still achieving a good statistical gain. The analytical approach is confirmed by results of a computer simulation that is extended to the more complex case of adaptive rate control.

In this paper the introduction of coherent multi-channel technology in local loop networks is being discussed. A topology study is given and the possible evolution scenario to introduce coherent technology is discussed. The concept and the realization of a complete coherent multi-channel system, designed for application in the local loop, is presented. The system supports broadband distributive and interactive services over coherent channels, and includes network signalling and network security.

Asynchronous transfer mode (ATM) based transmission over passive optical network (PON) is seen as a promising technique for the access network connecting both residential and small business customers. One of the key points for the success of such a proposition is the integration of most of the electrical functions in order to help for a low system cost, a low power consumption, and a good system availability. This paper discusses the architecture of an ATM-PON system by pointing out its main characteristics, then the main features of the developed ASICs in this environment are presented.

This paper evaluates the experimental performance comparison between analogue and digital correlation techniques in optical spread spectrum systems. The digital correlation achieved a 3 cm spatial resolution, in comparison with the analogue case of 20 cm. To the authors' best knowledge this is the best resolution reported to date with a non-coherent system. The advantages of the approach are also discussed with reference to synchronization performance.

A two-tier optical fiber local-access physical network configuration is proposed consisting of a passive, fully transparent optical fiber core and a mixed transmission media periphery for connecting all sorts of subscribers to the network. The two tiers are separated from each other by a population of active remote nodes. The possibility of implementation of the most common logical networks on the proposed infrastructure is investigated.

We report on the performance of a 8 X 8 polarization-independent LiNbO₃ switch matrix on the base directional coupler with its full complement of the local area network (LAND). In this network the random access method with the exclusion of collisions has been applied. Such optical transmitting medium in LAN with a through-put up to 200 Mbit/sec has been realized by the authors.

The performance of spread spectrum code division multiple access (SS-CDMA) can be further enhanced by employing a hybrid that incorporates the advantages of SS-CDMA with those of pulse position modulation. This leads to an improved sensitivity performance and higher data rate capability. This paper considers an experimental hybrid SS-PPM configuration and reports original results that demonstrate the correlated and uncorrelated detection of the SS- PPM signal.