This paper will review the measurements and characterization necessary to deal with advanced optical communication systems. These issues go beyond the optical power and bandwidth budgets used for simple point-to-point links that do not push to the limit the performance capabilities of advanced optoelectronic components. The paper will address primarily fiber optic communications, but some characterization issues raised by free-space optical communications will also be discussed.

The results of the progress made by the EIA in the standardization of single-mode fibers is reported. Emphasis is on the fiber's optical parameters as these have been some of the more controversial areas for standardization. Finally, the results of the recent agreement on the method of specifying single-mode fibers is reviewed.

The four most popular definitions of mode field diameter in single mode fibers are compared : Near-Field (or Far-Field) Gaussian fit ; Transverse Offset ; Variable Aperture Far Field ; Far-Field R MS width. We then use the values calculated for step-index matched-cladding, step-index depressed-cladding, and triangular-core dispersion-shifted fibers, for predicting the splicing losses between fibers. A classification of these various definitions is presented with respect to the exact splice loss computed for various fiber parameter mismatch and alignment errors, at 1300 nm and 1550 nm.

We have defined new Equivalent Step-Index parameters for single-mode fibers having a graded core profile and/or a central index dip. These parameters are obtained from the actual index profile through simple analytical expressions with a straightforward physical interpretation. As compared to previous ESI theories, they allow a better accuracy in predicting the transmission characteristics of the fibers. In particular, the use of these ESI parameters with the usual models for calculating the chromatic dispersion in single mode fibers proves very efficient : the accuracy in predicting the zero dispersion wavelength is better than 2 nm for nearly step index fibers designed for the 1.3 μm window, and it remains better than 15 nm for triangular core profile fibers designed for the 1.55 μm window.

Both time domain and frequency domain techniques have become well established for measuring the bandwidth of multimode optical fibers. The bandwidth of single mode fibers is much more source-dependent, however, and hence dispersion (or the zero dispersion wavelength) is used instead to characterize the information capacity. Dispersion can be determined by measuring pulse delay, phase shift, or fringe visibility as a function of wavelength.

An interferometer with a short length of fiber in one arm and a moving mirror in the other, with LED illumination, is described. The photocurrent generated in a detector viewing the output of the interferometer as the mirror moves contains an ac component whose instantaneous frequency is proportional to the optical frequency of one spectral component of the power spectrum of the source. This frequency is a unique function of the instantaneous position of the moving mirror, a function whose inverse yields the group delay as a function of frequency for the fiber under test. Simple counter-timer circuits are used to measure the instantaneous frequency of the signal as a function of mirror position and are monitored by counting reference fringes generated by HeNe radiation simultaneously traversing the interferometer. The HeNe signal provides an accurate measure of relative delay without the imposition of special requirement (e.g., uniform velocity) on the mirror motion, as well as a frequency calibration for the instantaneous frequency of the LED radiation being observed.

There has been much concern of late over the accuracy of Power Meters used in making loss measurements on long lengths of fiber cable, but comparatively little over the effects of source wavelength and spectral width. Being a function of wavelength, attenuation measurement errors can occur if the source wavelength is not controlled or adequately known. Presently available stable Laser Sources operate within a fairly broad wavelength range, usually being specified by equipment manufacturers at some nominal value of wavelength (1300 or 1550 nanometers) with a wide tolerance (30 to 40 nm). Measuring the same fiber with two "1300 nanometers" sources can result in a significant error. Measuring loss with an LED Source relates little to factory attenuation measurements due to their inherently wide spectral widths. This paper describes analysis and measurements on long lengths of concatenated Singlemode fibers using various Laser and LED Sources. Measurement precisions and repeatabilities are shown for various source types and wavelengths and the source wavelength tolerance required is determined for meaningful measurements.

Advances in fiber optic technology in the past few years have firmly established the superiority of optical fiber to coaxial cables, particularly for large-capacity, long-haul transmission systems. Recently, several undersea fiber optic cable systems have been proposed by both common and noncommon carriers. This paper addresses the techno-economic implications of these applications, and includes a brief review of the current status of undersea cable technology and a projection of future demand and capabilities. The prospects for using high-speed, multifiber undersea cable systems for international communications, extension of these systems through fiber optic terrestrial distribution, and future developmental trends are critically assessed.

Various architectures for optical distribution of digital data streams in large-capacity earth stations are described, analyzed, and compared. The distribution schemes are evaluated with respect to their optical power requirements, synchronization requirements, electrical and optical component costs, development costs, and complexity. A modular star distribution network emerges as the most attractive architecture.

Dispersion power penalties arising from chromatic dispersion and mode partition noise are discussed. Using chromatic dispersion measurements made on fifty 30 km spans in a MCI fiber-optic system, maximum repeater spacings for a hypothetical 565 Mb/s system (using conventional fibers and lasers at 1300 nm) and a 1.6 Gb/s system (using conventional fibers and DFB lasers at 1550 nm) are calculated. For the DFB laser, power penalties due to chirping, existence of a side mode, and mode partition noise are calculated.

Two methods employed during fusion splicing as well as two methods for calculating the average splice loss per fiber-span are presented. The two splicing methods described are the telemetry and optical time domain ref lectometer (OTDR) methods. The cut-back and OTDR measuring methods for calculating the average splice loss per fiber-span are also described. A statistical model relating the various measuring uncertainties with the sample size uncertainties is used to predict the population parameters from the sample statistics. Laboratory results are also included to analyze the important parameters needed for splice optimization. The accuracy of splice loss measurements obtained using OTDRs is studied in the laboratory and is used to analyze field data. Our results show that there is no significant difference in the quality of the telemetry splicing method as compared to the OTDR splicing method. In addition, the average splice loss per fiber-span calculated from cut-back measurements was found to yield a distribution that is only 80% as accurate as the one derived from the OTDR measurements.

Experiments show that optical fiber attenuation measurement systems used by many laboratories in China meet the requirements of standard measurement methods recommended by the International Telegraph and Telephone Consultative Committee (CCITT) and the International Electrotechnical Commission (IEC). A simplified, practical criterion for checking equilibrium mode distribution (EMD) has been obtained. Round-robin results show good agreement of attenuation measurement for six Chinese laboratories using this criterion.

The state of the art in optical time domain reflectometry is described from the viewpoint of the instrument designer, with emphasis on single-mode technology. The fundamental OTDR range/resolution tradeoff is reviewed, along with its practical implications for measurement accuracy. Finally, prospects for future OTDR performance improvements are discussed.

The presence of transient high-order-mode losses in multi-mode fiber optic cable test samples makes accurate data extrapolation to system lengths difficult. Extensive research has resulted in several methods for launching an equilibrium modal distribution (EMD) into the test sample. Most approaches require coupling the launching apparatus to the sample via a fiber optic splice. However, in many cases, splices disturb the modal distribution thereby reducing the effectiveness of the EMD launcher. In these instances, it is valuable to be able to determine the severity of EMD disruption and correct the problem. A technique employing an OTDR in conjunction with a throughput loss measurement system has recently been developed. The system is used to monitor the quality of EMD transfer into a cable sample prior to testing. Test results illustrating the effectiveness of the integrated system are included.

The measurement of longitudinal strain of optical fibers using several optical techniques is discussed. A review of the optical principles used to design each system is included. Each technique is based upon directly or indirectly measuring the change in the transit time of an optical signal injected into the test fiber. Equations are provided relating strain to the change in fiber transit time. Examples of calibration results and cable tests are given.

With the advent of fiber optics and optical communications technologies, much attention has been paid to the emitter and detector components associated with these technologies(1). These semiconducting devices are small chips which have been fabricated from crystalline wafers upon which "epitaxial" layers have been deposited. (Epitaxy is the phenomenon whereby the deposited solid layer adopts the crystal structure of the substrate upon which it is deposited). The purpose of this article is to compare some of the epitaxial techniques used to "grow" or synthesize these layers and then to focus in on one of the primary techniques used to grow commercial devices: vapor phase epitaxy (VPE).

In an active guided wave medium, the solutions of Maxwell's wave equation, with a driving polarization that is electrooptically induced, may be attained through a derived set of first-order coupled-mode equations that provide a powerful formalism for solving the transfer characteristics of a variety of integrated optics control devices. This paper will present the results of such application to two important interactions in guided wave structures: electrooptic tunable filtering and single-sideband frequency translation.

Optical domain inversion effects on the performance of electro-optic devices in Ti-indiffused C-cut LiNbO3 are studied. The C- surface was found to be considerably more immune to domain inversion than the C+ surface under similar diffusion conditions. The extent of domain inversion was found to depend on the diffusion temperature, time, initial Ti film thickness and waveguide width. Lower electro-optically induced phase mod lation was observed in the drive voltage in Mach-Zehnder interferometers fabricated on C-1- substrates compared to that of C- substrates under the same voltage drive. The increase in crosstalk levels and change in the switching behavior of directional couplers attributable to domain inversion are also discussed.

Single-mode integrated optic waveguide demultiplexer structures were fabricated consisting of a 4-pm-wide input channel waveguide, an adiabatic beam expansion horn 130 μm at its output, a planar waveguide region, and several 3-μm period grating structures embedded in the planar waveguide region. The angular orientation of the gratings was adjusted to varying at specific wavelengths. The planar waveguide, the channel waveguide, the horn, and the grating structures were simultaneously formed by dilute silver nitrate ion exchange on an aluminum masked slide. High diffraction efficiencies were observed in good agreement with the theoretical model for the horn-grating interaction at 0.6328-, 0.789- and 0.82-μm wavelengths. This is the first reported operation of a single-mode demultiplexer structure with simultaneously fabricated input waveguide, beam expansion, and recollimation horms and multiple diffraction gratings.

Miniaturized optical transmitter and receiver modules for optical fiber data communications up to 200 Mbit/s is described. The modules use an InGaAsP LED operating at 1.3um and an InGaAs p-i-n photodetector. All the electronic circuits are integrated on one IC in the transmitter module and two ICs in the receiver module. They are packaged in a standard dual-in-line with integral optical connectors. The transmission distance can be as far as 5 Km.

A wideband electro-optic direction finding (DF) processor employing a network of fiber optic delay lines is described. This DF filter offers a potential instantaneous bandwidth of 20 GHz and allows for multiple, simultaneous beam angular responses with peaks which are independent of frequency. Preliminary results of an eight-laser feasibility test model DF device with two transversal filters show good agreement between predicted and measured angular responses. This experimental matched delay filter operates in the 200-1000 MHz frequency range and can simultaneously monitor two angles of arrival. The laboratory results obtained in this investigation suggest that the fiber optics matched delay filter can be useful for wideband direction finding.

A numerical method to solve the rate equations is presented. This is applied to the superluminescence diodes revealing the spatial gain and photon density profiles at high frequency modulation.

Link parameters, including Bit-Error-Ratio (BER), jitter, instantaneous and transient losses, and effective bandwidth, are defined, and their effects on fiber optic LAN (Local Area Network) link design are quantified. Measurement conditions appropriate to LAN applications and economical options available to fiber optic systems designers are discussed.

The cost-effective application of optical communications components requires that the product characterization measurements also predict the actual performance of the component in-situ, particularly for fundamental attributes such as attenuation. In this work, attenuation characteristics of individual components by several measurement methods were compared to the observed power reduction (loss) over a range of complexities of a practical LAN link topology for the ANSI X3T9.5 Fiber Distributed Data Interface (FDDI). The results clearly indicate the need to use quasi-steady-state measurement methods for LAN components, even when the total link length is less than 100 meters.

An optical fiber cable distribution architecture and a ring interface are described. The unique synergism of the ring configuration coupled with a widespread optical fiber cable facility are explored. The ring interface adapts a token passing network of work-stations from coaxial cable to 50 micron core, telecommunications type, fiber optic cable. The optical fiber cable links a series of communications centers to as many as 288 optical fibers. The loop distribution of optical fibers forms the backbone for a 6.7 km ring operating at 6-12 Mbps.

Local area networks are usually built with multimode fibers. Since insertion loss of all components is mode dependant, a mode power matrix method was developed that allows to take this mode dependancy into account with moderate mathematical expense. Here, an extention is presented that is closely related to the power matrix method but is able to include dispersion properties as well.

A method, using Mode Transfer Matrices (HIM) to characterize step index fiber components and predict Local Area Network (LAN) power budget, is presented. The results show this method is well adapted to describing modal power distribution variations.

AstraNet* is a heirarchical, active-star configured, Ethernet-compatible, fiber optic local area network. The use of active nodes makes all fiber links in the network point-to-point in nature, permiting the use of cost. effective LEDs and PINs and assuring reliable collision detection. AstraNet's heirarchical topology significantly reduces the amount of optical fiber required, reducing both cable and installation costs. We describe the AstraNet architecture and modules and report the results of field trials involving up to 176 network ports and both wall and under-carpet outlets.

Various means of adding fault tolerance and survivability to fiber optic local area networks are reviewed. A method of building a high level of fault tolerance into a hierarchical star topology network is introduced. This attribute makes the network attractive in military Ci applications or anywhere a rugged, survivable network is required.

Longwave fiber optic systems require accurate power measurements in the spectral region from 1.0 to 1.6 microns. We have studied the characteristics of germanium photodiodes with the hope that these devices might achieve the accuracy, stability, and dynamic range in the longwave region that silicon photodiodes have achieved in the shortwave region.1'2 A natural concern is the effect of temperature on the performance of germanium photodiodes, since any long wavelength quantum photodetector, such as germanium, must necessarily have a small bandgap and thus a considerable and temperature variable number of thermally generated electron-hole pairs. Additionally, uniformity of response over the detector surface plays an important role in the accuracy of transfer calibrations.

A system for measuring optical fiber bandwidth utilizing the Pulse Spectrum Analysis method (PSA) has been established. This paper will discuss problems inherent to that system such as signal-to-noise ratio and off-peak error. Included are the results of bandwidth measurements on multimode telecommunication grade fibers. Finally, the PSA method is compared to other bandwidth measurement methods: the frequency domain (FD) and the time domain (TD) methods.