WO1985004064A1

WO1985004064A1 - Increased bit rate optical data stream

Info

Publication number: WO1985004064A1
Application number: PCT/US1985/000261
Authority: WO
Inventors: Kenneth Arthur Jackson
Original assignee: American Telephone & Telegraph Company
Priority date: 1984-03-05
Filing date: 1985-02-15
Publication date: 1985-09-12
Also published as: EP0172887A1; JPS61501363A

Abstract

Optical pulses from an unmodulated pulse generator (10) (e.g., Q-switched laser or mode-locked semi-conductor laser) are divided into optical paths of differing lengths (11-15), and hence differing optical transit times. A modulator (G) is inserted into at least some of the optical paths, and the outputs of the paths are combined onto a single optical fiber (17). This provides a total bit rate on the optical fiber equal to the sum of the modulation rates on each path. Hence, a relatively low pulse rate source and relatively low bit rate modulators are utilized to obtain a higher effective rate. Additional modulated paths may be provided at various locations along the optical fiber, for adding information thereto. A corresponding demodulation scheme is also possible, wherein a high bit rate optical data stream is divided into a plurality of lower rate streams.

Description

INCREASED BIT RATE OPTICAL DATA STREAM

Background of the Invention

1. Field of the Invention The present invention relates to a technique for obtaining an optical data stream on an optical path.

2. Description of the Prior Art

Optical fiber communications has made very great progress in the past few years in terms of increasing the distances that can be traversed without repeaters, as well as increasing the data rate that can be transmitted. It is well known that the chromatic dispersion in an optical fiber plays a significant effect in determining the maximum data rate that can be transmitted through the fiber for a given distance. In fact, the maximum distance between repeaters may be limited more by the chromatic dispersion of the fiber, causing a smearing of the transmitted signal in time, than by the inherent optical loss in the fiber.

By operating at a wavelength wherein the fiber has minimum chromatic dispersion, typically around 1.3 micrometers, the pulse distortion can be minimized. It is also possible to minimize chromatic dispersion by using an optical source having a very narrow chromatic spectrum. This is especially useful when operating at a wavelength other than the zero dispersion wavelength. For example, operation may be desired around 1.55 micrometers wavelength, where most silica optical fibers have a loss minimum. In short, the adoption of single mode optical fibers operating a zero dispersion wavelength, or the use of a very narrow spectral width source, has allowed substantial increases in repeater spacing, and increases in data rates obtainable.

However, even assuming an essentially perfect optical fiber in terms of dispersion, and an essentially monochromatic optical source, there still exists a substantial limitation on the rate at which data can be sent through an optical fiber. This limitation is the rate at which a laser or other optical source can be turned on and off; that is, modulated. This limitation in some cases is due to the physics of the optical source itself, wherein a certain time is required after the application of an electrical pulse for an optical pulse to be generated. Furthermore, a delay is inherent after the electrical signal is removed for the optical pulse to decay to an essentially zero value. Other effects include the capacitance and inductance in the modulator, which limit the electrical response times therein. Considering the necessity to preserve the monochromatic nature of the source, the maximum practical modulation rates are at present about 10⁹ bits per second (1 gigabit per second). While this is a substantially higher rate than was obtainable even a few years ago, even higher rates are desired. This is because a well designed single mode optical fiber is able to support optical pulse rates of many gigabits per second (i.e., over 10⁹ bits per second). In summary , there is a need for an optical source having a very narrow spectral width and capable of modulation at very high rates in order to take advantage of the potential offered by optical fibers. In other applications, there is a need for adding additional information onto an optical fiber at various points along its length. For example, office type applications need the ability for one work station to send information to one or more other stations, desirably on a single optical fiber. Summary of the Invention I have invented a technique for obtaining a high bit rate information stream in a single optical path typically comprising an optical fiber. In this technique an optical source, typically a laser, continually produces pulses having relatively short time duration, but having a relatively long repetition interval. The optical output of this source is divided into two or more optical paths of differing optical lengths. At least two of the optical paths have located therein an optical gate capable of controlling the optical pulse propagating therethrough. Each gate may be individually modulated by a separate information source. The outputs of the optical paths are then inserted onto a single optical path, typically a single mode optical fiber. An information stream results having a bit rate that is the sum of the modulated rates of the separate optical paths. The modulated optical paths may be provided at various locations along the optical fiber, thereby adding additional information thereto. In the inverse operation, an optical fiber carrying a high bit rate information stream is divided into a multiplicity of paths, each having an optical gate therein. Each gate is activated on so as to allow a single information pulse through in a given time period. In this manner a multiplicity of optical paths are obtained, each having a portion of the information in the optical fiber. Brief Description of the Drawings

FIG. 1 shows the basic modulation scheme whereby an optical source is divided into a multiplicity of optical paths that are separately modulated and combined onto a single optical fiber.

FIG. 2 shows a typical amplitude versus time graph of the optical source. FIG. 3 shows the combined outputs of the optical paths if all of the gates allow the optical pulses to pass therethrough.

FIG. 4 shows a typical modulation sequence for the different gates in the optical paths, and a resulting combined data stream.

FIG. 5 shows an embodiment for adding additional data at various points along the fiber.

FIG. 6 shows a demoduation technique according to the present invention, whereby a high bit rate optical data stream is demodulated into lower bit rate data paths. Detailed Description

The following detailed description relates to a technique for obtaining a high bit rate optical data stream utilizing optical components that individually operate at a relatively lower modulation rates than the resulting data stream. Both a modulation and a demodulation technique are shown. As used herein, the term "optical length" relates to the transit time of an optical pulse through an optical path. The term "gate" refers to a device that can change the amplitude or other characteristic of an optical pulse applied thereto, in response to a modulating or demodulating signal.

Referring to FIG. 1, an optical source 10 provides a continuous stream of optical pulses. These pulses have a relatively short duration (t) in comparison with the relatively long pulse repetition interval (T); see FIG. 2. For convenience, a given pulse repetition interval T is also referred to herein as a "frame". The optical output of source 10 is divided and directed through a plurality of optical paths (11 through 15). Located within the various paths are optical gates (G₁ through G₅ that communicate with a modulator (16), which independently controls the opening of each gate. Since each optical path has a different optical length than the others, the delay of the unmodulated pulses (P₀) from source 10 will be different. The outputs of the optical paths 11 through 15 are then combined on a single optical fiber 17. Due to the differing delays, each modulated signal will then occupy a different position in time along the single optical fiber. Source 10 can be any of a number of conventional sources, including for example, a Q-switched gas laser, a dye-laser, but is preferably a mode-locked semiconductor laser. To form a mode-locked laser, a semiconductor laser can be placed within a Fabry-Perot cavity. The time interval between the pulses is determined by the length of the cavity. For example, one face of the semiconductor laser may have a mirror formed thereon, and can be placed at one end of the Fabry-Perot cavity. The other face of the semiconductor laser can then be optically coated, to allow transmission of the optical energy therethrough. Then, the opposite end of the Fabry-Perot cavity may have another mirror formed thereon, allowing multiple reflections of the optical pulse within the cavity. This can achieve mode locking of the laser to obtain a spectrally pure output, and hence reduce the chromatic dispersion of the optical pulse as it is transmitted along the optical fiber. The interval of the optical pulses is then determined by the length of the Fabry-Perot cavity. Considering that light travels about 1 foot per nanosecond in free space, a cavity length of six inches provides a round trip distance of 12 inches, or about 1 pulse per nanosecond, providing a pulse repetition rate of 1 gigabit per second. In addition, such a source can produce pulses having very short durations (t), with 1 picosecond duration pulses being readily obtainable. As will be seen, the present technique allows for obtaining a modulated pulse stream at a rate significantly higher than the basic pulse repetition rate of the optical source. The operation of the system shown in FIG. 1 is further illustrated by assuming that all of the optical gates G₁ through G₅ are initially turned "on" so that optical pulses from source 10 are allowed to propagate through optical paths 11 through 15. FIG. 3 then shows the output of the five paths as they are combined at the input of optical fiber 17. The pulse P₁ is that due to the optical pulse traveling through the shortest modulated path (11), with the succeeding pulses traveling through the longer paths as indicated. Thus, the pulse traveling through gate 2 (optical path 12) arrives at a time later than the pulse through path 11, due to the additional length of path 12, and so forth for the other pulses. It is also possible to send a pulse P₀ through unmodulated path 18 for synchronization with another modulator, or with a demodulator, discussed below. Note also that the basic repetition time between the beginning of a subsequent pulse propagating through a given path (for example path 11 and gate 1) is given by time T, which is the same as the basic pulse repetition time of the optical source 10. Note that the pulses associated with gates G₁ through G₅ as shown equally spaced in FIG. 3 and have a pulse spacing

However, this is not necessarily so, and the pulses may be unequally spaced, if the paths are of unequal optical lengths. Furthermore, not all of the time available in each frame needs to be occupied by a data channel. Rather, some of the time can be reserved for future system expansion, including the addition of data at various locations along the optical fiber, discussed below.

The maximum number of modulated paths that can be combined onto a fiber is given typically by the pulse repetition interval divided by the pulse duration (T/t) , which defines the maximum number of time "slots" available in each frame. The ratio T/t is at least 2, and typcially at least 5. Allowing a sufficient guard space between each slot to provide suitably low error rates may reduce the number of slots somewhat from the maximum. Since the optical output from source 10 is divided into a multiplicity of paths, the amplitude at the input of each path will be less than the total output of the optical source. If the gates G₁ through G₅ do not provide amplitude gain, the resulting outputs from the paths will therefore be less than the output of source 10. However, as indicated further below, the gates G₁ through G₅ may in some cases provide for optical gain, and hence provide for an optical output higher than at the input of each path.

As indicated in FIG. 1, optical source 10 and optical modulator 16 are synchronized to provide for applying the modulation signal to each gate at the proper time in each frame. This can be accomplished by providing an optical output pulse from source 10 that is converted into an electrical signal by a detector, thus providing a reference time point to the modulator. Alternately, the modulator 16 may provide an electrical or optical signal to source 10 for controlling the time instants at which source 10 produces its pulses. Another possibility is to control both the optical source and the modulator from a system timing source, which may provide an electrical or optical synchronization pulse. To simplify synchronization, the optical gates are desirably all placed the same optical distance from the source 10. Then, the modulator can provide the modulating signal simultaneously to all the gates necessary to obtain the desired data sequence in a given frame. However, it is possible to place a gate at any location in its path, including the other end, with appropriate adjustment to the timing of its modulating signal; see e.g., FIG. 4.

Referring now to FIG. 4 it will be seen how information is modulated onto the pulses in each path. In an exemplary frame, it is seen how the gates G₁ and G₄ are turned on for a time period t_G, whereas gates 2, 3, and 5 are turned off. The time period t_G is typically less than the basic pulse repetition time interval (T), and is typically greater than the duration of the individual optical pulses (t). Hence, by providing the modulation shown, the modulated output of the five paths as combined onto the optical fiber, wherein pulses P₁ and P₄ are obtained as shown on the upper portion of FIG. 4. The modulator can then change to a new state in order to provide a different sequence of modulated pulses during the next frame. The modulator typically provides for independently modulating each gate by separate information sources. However, two or more gates can be modulated by the same information source to provide for redundancy in transmission for reduced error rates, or for other purposes. Also, the time period t_G can be greater than T, to provide two (or more) frames having the same information, as for redundancy purposes.

It is apparent that the above scheme offers numerous advantages in optical communication. Firstly, it allows the use of spectrally pure optical source types that are not otherwise easily modulated at high rates. This is because the present scheme avoids modulating the optical source, and allows the mode locking scheme or other stabilization scheme to produce high spectral purity in the source . Secondly, certain types of optical sources , even though they produce very short pulses, are not capable of rapid modulation. The present scheme allows such sources to be utilized. Thirdly, it can be seen that each of the modulation gates are operated at a relatively low modulation rate. This allows devices to be more readily manufactured and modulated then would be the case if they were required to operate at a high rate. Fourthly, the electrical portions of the modulator can also operate at a relatively low rate. Furthermore, additional advantages relate to flexibility in pulse spacing, and the ability to increase the number of available channels as capacity needs increase. The present technique also allows for information to be added at any point along the fiber's length. Referring to FIG. 5, consider an optical fiber 51 carrying therein a stream of information pulses P₁...P₅ traversing left to right as viewed. Note that slots P₁ and P₄ are "on", producing pulses P₁ and P₄, whereas slots P₂, P₃, and P₅ are "off" for the exemplary frame of FIG. 4. Optical fiber 52 carries a continuous stream of unmodulated pulses (P₀) from an optical source (not shown). At the desired location, an optical splitter (54) directs a portion of the optical energy from fiber 52 onto path 53. An optical gate G₆ is placed in path 53 , and modulated by modulator 56 to produce a modulated pulse (P₆), shown as being "on" during its corresponding slot in the exemplary frame. The optical length of path 53 is chosen to delay modulated pulse P₆ with respect to P₁...P₅. Then the output end of path 53 is combined onto fiber 51 , so that P₆ is added to the information stream on fiber 51 at point 55. The modulator 56 is synchronized with the continuous pulse stream P₀ on fiber 52. This synchronization can be accomplished by detecting a portion of the optical energy of pulse P₀ traversing fiber 52 and applying an electrical signal to modulator 56, or by other means.

Note that additional delay paths can be split off at point 54, independently modulated, and combined at 55, as desired, for adding still more independent information channels. Still more access points can be obtained in a like manner at any point along the fiber's length. Units may be added until all time slots in a frame are utilized. This capability gives the present invention broad applicability to optical fibers local area nets (LANS), useful in office-type environments, among others. For such uses, a very high data rate is less important than the ability to add and extract information from the fiber at various points along its length. In fact, in some cases, light emitting diodes (LEDS) are suitable at the unmodulated optical source, and the fiber may be of the multimode variety. Furthermore, in some applications, there is no need for modulating the optical stream near source 10, as shown in FIG. 1. Rather, all of the modulated optical paths may be located at various distances from the source, as indicated in FIG. 5.

The above technique can also be applied to a corresponding demodulation technique. Referring to FIG. 6, the optical output from optical fiber 17 is split into a multiplicity of optical paths 61 through 65. The demodulation gates D₁ through D₅ are controlled by a demodulator 66 which clocks the opening of the demodulation gates. For the unequal path lengths shown in FIG. 6, the demodulation gates may be opened simultaneously, but for a time short enough to ensure that only one pulse per frame passes therethrough. The output of each demodulation path will correspond to the modulated pulse in the corresponding modulation path. Since pulse P1 arrives first at the demodulation path inputs, it will be demodulated in the longest path (65), and pulse P₅ in the shortest path (61) if all of the demodulation gates are opened simultaneously by the demodulator. The outputs of the demodulation paths are typically suppled to detectors (67-71) for converting the optical pulses to electrical pulses. The above noted advantages also apply in that case, in that the detectors need operate at a pulse rate only a fraction of the unmodulated optical source rate. It is also possible to couple the optical outputs of the demodulator gates to other optical fibers for transmission to other locations. It is apparent that, by analogy to the modulation scheme of FIG. 5, demodulation paths can be supplied at any point along the fiber's length. To provide for synchronizing the opening of gates

D₁...D₅ the demodulator 66, a synchronizing pulse may be sent from the modulator. This can be accomplished using an unmodulated path (18) from the optical source, as shown in FIG. 1. This synchronizing pulse may have an amplitude different from those of the modulated pulses as shown in FIG. 3. Alternately, a different pulse characteristic can be used for synchronization, such as optical polarity, pulse position, pulse wavelength, etc. Still other synchronizing techniques are possible, including the use of a separate optical fiber to carry the synchronizing pulse, as shown in FIG. 5.

It is also possible to use an optical demodulation technique which relies upon detecting the coincidence of a synchronizing optical pulse and an information optical pulse. For example, a non-linear optical medium can be used that produces a frequency doubled optical output when two optical inputs are present simultaneously. One of the optical inputs is the synchronizing pulse carried by a first optical fiber (e.g., P₀ on fiber 52 in FIG. 5). The other optical input is the information stream e.g., P₁...P₆ on fiber 51 in FIG. 5). To obtain the desired optical coincidence, the synchronizing pulses may be divided into optical paths of unequal lengths, as shown in FIG. 6, and applied to the coincidence detectors in place of the gates D₁...D₅. The synchronizing optical pulses (P₀) are also divided and applied to each coincidence detector. Each coincidence detector will then produce an output on each demodulation path for the pulses on the corresponding modulation path, as noted above. Note that it is alternately possible to provide the relative delays between the divided synchronizing pulses, while not delaying the information pulses relative to each other at the coincidence detector input. The optical coincidence detection technique eliminates the need for an electrical demodulator that must open each modulation gate for a time period short enough to separate optical pulses, as required in FIG. 6.

Such a coincidence detection technique, making use of the non-linear optical properties of lithium iodate, has been demonstrated. Lithium iodate and other non-linear optical crystals can double the frequency of light (i.e., convert red light to green light) incident on the crystal near the phase matching angle. The geometry can be arranged so that frequency doubled light is produced only when two beams are simultaneously incident on the crystal, and either of the two beams alone will not produce frequency doubled light. One such beam carries the information pulses, and the other beam carries the synchronization pulses. Each beam is directed into the crystal at a slight angle (e.g. 5 degrees) to the phase matching angle. The beams are directed to the same spot on the crystal. When the path differences are adjusted properly, the required information stream can be detected as pulses of green light coming from the crystal at the initial single channel modulation rate. In one test a mode locked krypton pumped dye laser produced about 3 milliwatts at 0.83 micrometers wavelength for synchronizing pulses, and a semiconductor gate produced about 1 milliwatt information pulses on the crystal. The pulses were about 1 psec duration, and spaced about 10 nsec. Changing the path length by 10 psec for either the information stream or the synchronous pulse stream eliminated the frequency doubled output. The optical paths 11 through 15 and 61 through 65 can be free space paths, with discrete modulators G₁ through G₅ and demodulators D₁ through D₅ associated therein. However, it is especially convenient to utilize optical fibers for these paths. It is also possible to form the optical paths and the associated gates on a single optical substrate, whereby an integrated optical device is obtained. The different optical lengths may then be obtained, for example, by different dopant concentrations, producing differing refractive indices among the paths. In some materials, impressing an electric field across the material produces a change in the index of refraction. In some cases, applying acoustical energy to a material changes its refractive index, and hence changes the optical length of a path having a given physical length. Of course, the use of differing physical lengths for the paths is also possible. In addition, the modulator 16 and demodulator 66 can be formed on a separate integrated circuit chip, or may be additionally combined on an electro-optical chip comprising the optical paths and gates. While the demodulation gates (D₁...D₅) and the detectors (67...71) can be separate devices, as illustrated in FIG. 6, their functions can be combined in a single device. For example, a detector that converts optical energy into electrical energy may be electrically or optically controlled to respond to the information pulses only when the synchronization pulse is present.

It is also possible to use a non-linear optical medium for reducing the duration of the optical pulses, either before or after modulation. For example, the use of an optical fiber having a non-linear dependence of the fiber refractive index with electric field is known in the art for producing so-called "soliton" pulses; see U. S. patent 4,406,516 co-assigned with the present invention. For pulse compression, an appropriately designed optical fiber or other non-linear device may be used in the optical paths (11-15) either before or after the gates (G₁-G₅). In that case, the pulse duration (t) referred to herein means the duration after compression.

The basis of gates devices can be a GaAlAs laser diode. This device, when biased by an electronic pulse, can amplify an external optical pulse which is focussed into the active area and which is coincident in time with the current pulse. Optical amplification has hitherto been observed in GaAs structures under dc current injection. A co-worker of the present inventor has discovered that amplification of pulses which are only a few psec long can be accomplished in a double heterostructure GaAlAs traveling-wave type diode amplifier. When the diode is injected with current pulses overlapping the optical pulses, amplification by a factor of 10 occurs when the current pulse is on, while attenuation by a factor of 10 occurs when the current pulse is off, leading to modulation depths of at least 100. Modulation rates greater than 1 GH_z are possible. In this way information in the form of TTL-like electronic pulses can be transformed with gain into optical pulses of arbitrarily short length. One example of a suitable gate, and its operation, are as follows: A Ga₀₉Al_0.1AS was used that had a standard proton-bombarded stripe geometry, with no optical guiding in the active area; see "Physics and Technology of Semiconductor Emitters and Detectors", L. A. D'Asako, Journal of Luminescence, Vol. 7, page 310 (1973). The facets of the diode were anti-reflection coated with a layer of SiO₂ to reduce the reflectivity to about 1%, thereby preventing self-oscillation of the diode. The diode was cw bonded to a brass stud and injected with 2 nsec voltage pulses from a pulse generator. The optical source was a dye laser synchronously pumped by a modelocked Krypton laser. The dye was oxazine 750 which gave 10 psec pulses at 81 MHz rate and tunability from 7400 Å to 8400 Å covering the gain spectrum of the diode. The pulse generator was synchronized to the modelocked rf source which operated at half the optical pulse rate. The dye laser output was focussed onto one facet of the diode active area and the emerging beam from the opposite facet was detected by a Si detector with less than 1 nsec risetime. The signal was displayed on an oscilloscope and photographed. The output wavelength of the diode was at 8300 Å, the diode peak of the diode gain curve. The diode was injected with a 40.5 MHz electronic pulse train with 2.5V amplitude, and each pulse in the train had a duration of 2 nsec. When the relative delay was adjusted so that each current pulse overlapped an optical pulse in time, every second optical pulse experienced an increase in intensity by a factor of about 100. A comparison of the output to the input intensities of these pulses showed a net gain by a factor of 8-10. This included all coupling losses to the diode. The optical pulses passing through the diode in the absence of current pulses were attenuated by a factor 10 due to absorption across the bandgap. Hence a modulation depth of about 100 occurred under these conditions. For high input intensities, saturation of the gain was observed. At 5 mw average input power, the gain was reduced to about 1, giving a modulation of 10 only. At input powers less than about 0.1 mw the gain was linear. It was determined that under non-saturating conditions, a modulation rate of at least 1 GHZ was possible.

The gates need not provide amplitude gain. A relatively high-power source (10), for example a gas laser, may be used to overcome the losses incurred by dividing the continuous pulse stream. Furthermore, the gate may themselves be controlled by optical or acoustical signals (instead of electrical signals) from the modulator. While gates, G₁-G₅ and D₁-D₅ have been described in terms of controlling the amplitude of pulses in their respective paths, other properties of the optical pulses can be controlled for purposes of modulation and demodulation. For example, the optical wavelength of the pulses can be shifted in response to the modulating signal by the use of an appropriate heterodyne device. Alternately, a switch can be used to divert the pulses out of the optical path. Alternately, the polarization of the optical pulses can be modulated. In that case, a polarization-preserving optical fiber (17) may be used for transmission. The appropriate devices for performing these functions on the optical pulse are included in the term "gate means" as used herein. The means for dividing the pulses from the source into a multiplicity of paths (19 in FIG. 1), and for combining the outputs of the paths (20 in FIG. 1) may be implemented by components known in the art; for example, "An Engineering Guide to Couplers", J. C. Williams et al. Laser Focus October 1981, pages 129-134. Still other component variations are possible. While the present invention affords significant benefits to optical fiber transmission systems, other optical path media are possible. For example, direct transmission through the atmosphere between buildings or between computers is known in the art, and can benefit from the present technique. All such utilizations of the inventive teaching herein are within the spirit and scope of the present invention.

Claims

1. A system for transmitting optical information CHARACTERIZED BY means for producing a continuous stream of optical pulses having a duration (t) and an interval (T) between adjacent pulses; means for dividing said stream of pulses into at least two optical paths having unequal transit times; gate means for modulating the stream of pulses transiting through each of at least two of said paths; and means for inserting the outputs of said optical paths onto a single optical paths.

2. The system of claim 1, CHARACTERIZED IN THAT said dividing said stream of pulses occurs for at least two of said paths at the same optical distance from said source.

3. The system of claim 1 , CHARACTERIZED IN THAT said dividing said stream of pulses occurs for at least two of said paths at different optical distances from said source.

4. The system of claim 1 further CHARACTERIZED BY means for synchronizing said means for producing the stream of optical pulses and said gate means for modulating the stream of pulses.

5. The system of claim 1, CHARACTERIZED IN THAT the ratio of said interval to said duration (T/t) is at least 5.

6. The system of claim 1 , CHARACTERIZED IN THAT said means for producing optical pulses is a modelocked solid state laser.

7. The system of claim 1,

CHARACTERIZED IN THAT said means for producing optical pulses is a Q- switched laser.

8. The system of claim 1 further CHARACTERIZED BY a non-linear optical medium for compressing the duration of said pulses prior to said inserting.

9. The system of claim 1 further CHARACTERIZED BY means for inserting an unmodulated stream of said pulses onto said single optical path.

10. The system of claim 1 further

CHARACTERIZED BY means for inserting an unmodulated stream of said pulses onto an optical path separate from said single optical path.

11. The system of claim 1

CHARACTERIZED IN THAT said single optical path comprises an optical fiber.

12. A method of transmitting optical information CHARACTERIZED BY producing a continuous stream of optical pulses having a duration (t) and an interval between adjacent pulses (T), dividing said stream of optical pulses into at least two of optical paths having controllably different optical transit times, gating the pulses transiting at least two of said paths in response to at least two modulating signals, and inserting the output of said optical paths onto a single optical fiber.

13. The method of claim 12 further

CHARACTERIZED BY the step of compressing the duration of said pulses by means of a non-linear optical medium.

14. The method of claim 12 CHARACTERIZED IN THAT the ratio of said interval to said duration (T/t) is at least 5.