Classifications

• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04B—TRANSMISSION
  • H04B10/00—Transmission systems employing electromagnetic waves other than radio-waves, e.g. infrared, visible or ultraviolet light, or employing corpuscular radiation, e.g. quantum communication
  • H04B10/25—Arrangements specific to fibre transmission
  • H04B10/2507—Arrangements specific to fibre transmission for the reduction or elimination of distortion or dispersion
  • H04B10/25077—Arrangements specific to fibre transmission for the reduction or elimination of distortion or dispersion using soliton propagation

Definitions

• the present invention relates to fibre optic transmission systems, and in particular to systems using optical amplification for long distance transmission at high bit rates.
• a transmission system refers in the present context to a modulation scheme at the transmitter, the spectral and temporal characteristics of the light, the physical path of transmission including the fibre, optical amplifiers, filters isolators and any other components, and the receiver.
• the components can be individually characterised, their response as a system is not a trivial function of the individual characteristics.
• the over-riding reason for this is the interplay between non-linearity and dispersion effects which become important in long-haul transmission systems.
• the main constraint on transmission distance is the loss in the system due to the constant attenuation of the fibre, and dispersion due to the chirping of the transmitter.
• These systems can often be considered as being linear in their response, and the transmission over the optical fibre can be treated as simply adding a certain amount of dispersion, and attenuating by a determined amount.
• Semiconductor laser diodes are almost universally used as the transmitter in current and proposed fibre optic communications systems.
• Most systems employ some form of digital communication whereby the signal which is being transmitted is represented as a sequence of 1 's and 0's, independent of nature of the information being transported.
• the way in which these 1's and 0's are represented in a transmission system depends upon the type of modulation employed.
• the most common type of modulation is known as Amplitude Modulation, where the intensity of the light indicates the state of a particular bit of information, high intensity for 1's and low intensity for zeros. This is usually achieved by turning the laster on and off by modulating the current to the laser.
• RZ Return to Zero
• NRZ Non Return to Zero
• amplitude modulation is that the wavelength (and hence frequency) of the light is chirped, i.e. the wavelength of the laser changes during the transmission of the bit because the laser transmission wavelength is a function of the current applied.
• This effect is detrimental to long distance transmission because optical fibre is dispersive, i.e. different wavelengths travel at slightly different speeds down the fibre.
• the light from one bit arrives at the receiver at different times, and so is distorted and can interfere with the other bits. This effect limits the distance that data can travel down an optical fibre, and becomes increasingly important at high bit rates.
• An alternative to modulating the current to the laser diode is to apply a constant current to the laser diode, and to externally modulate the light from the laser using a device such as a lithium Niobate crystal modulator. This effectively reduces the chirp, but is difficult to realise because of the high drive voltages required. Additionally, Brillouin scattering can be a problem over long distances because of the large wavelength component at a single frequency.
• Frequency Shift Keying encodes the bits of information not by on or off, but by transmitting at two different frequency for the ones and the zeros. These rapid changes in the frequency of the light can be achieved by small changes in the current applied to the laser diode. This light can be decoded by filtering the light to only allow one frequency pass. This is shown in Figure 3. This results in a conversion to amplitude modulation which can be detected in the normal way at the receiver.
• the filtering can take place either after the transmitter or before the receiver, and the filter can be realised in several different ways, such as Mach-Zender interferometer or a Fabry-Perot filter. As only small changes in frequency are required (dependent upon the bit rate), the wavelength spread is minimal and the effect of dispersion is lessened.
• Kerr effect is that the speed of transmission of light through a fibre is a function of the intensity of the light. Although this is usually only a very small effect, at sufficiently high intensities and over long enough distances the net effect can be quite dramatic, including extreme pulse narrowing and chaotic behaviour.
• Solitons, or solitary waves are pulses of a mathematically defined shape (solutions of the non-linear Schrodinger Equation) which can travel along a dispersive non-linear medium without change of shape.
• the pulse narrowing effect of the non-linearity exactly cancels the pulse broadening effect of the dispersion and the pulse propagates undisturbed indefinitely.
• each soliton represents a 1 in the data stream and absence of a soliton indicates a zero.
• Soliton propagation requires both constant intensity of the light, and that the pulses be solitary.
• the pulses must be separated from each other by a distance much larger than the width of each individual pulse. This requirement results from the non-linear interaction which occurs between pulses that are too close leading to chaotic behaviour. Indeed, by definition, solitons must be propagated in a solitary way.
• Solitons are introduced at a higher intensity and decay to a lower intensity than the required average intensity.
• Objects of Invention It is an object of this invention to provide a long distance transmission system which utilises non-linearity to overcome dispersion, but which is not limited to short span lengths between amplifiers. Disclosure of Invention The proposed transmission system involves the transmission of near- transform limited pulses.
• the invention is based on the discovery that by specifying the pulse width and bit rate of an optical transmission system, in such a way that the rules of soliton transmission are simultaneously violated, then by controlling the intensity throughout the system by controlling the optical amplification, the system length can be extended by a large factor over the dispersion limit for a comparable linear system.
• the system according to the invention also allows for an large increase in span length compared to a soliton system.
• the pulses no longer propagate as solitons, as even at constant intensity they will eventually interact with each other, but by choosing the pulse widths according to the bit rate and dispersion of the system they can be made to propagate for thousands of kilometres.
• FIG. 2 illustrates the distinction between RZ and NRZ data
• FIG. 3 illustrates modulation by the FSK technique
• Figure 4 illustrates one form of modulation suitable for providing pulses according to the present invention
• Figure 5 further illustrates the waveforms present for sample data at various stages in the modulation technique
• Figure 6 illustrates schematically in block form one system for generating pulses suitable for use with the present invention
• Figure 7 illustrates schematically an example of the operation of a system according to the invention.
• Figures 8, 9 and 10 show the same digital signal as transmitted and as received over 1000 and 2000 km of the system of Figure 7 for the preferred embodiments short pulse system and a conventional NRZ system respectively.
• Description of Embodiment Figure 1 illustrates a very simple example of the type of system contemplated, comprising transmitter 10, transmission fibre 15 and optical amplifier 20, terminating in receiver 30. It will be understood that many amplifiers 20, and accordingly many fibre spans, may be present as required in the actual implementation.
• the system comprises a transmitter 10 which is able to produce a data train of near-transform limited pulses of the desired width and bit rate.
• the specifications for these are an integral part of the system.
• one solution of the non ⁇ linear Schr ⁇ dinger equation can have a whole series of physical interpretations, where only the time and length scans change according to (i) and (ii).
• the systems with which the present invention is concerned are ones in which the characteristic length is large compared to the span length. This ensures the pulses no longer react adiabatically, or in other words.that their shape is determined by an average power instead of the instantaneous power, ensuring the pulses do not expand and contract too much with the attenuation and gain along a fibre.
• the bit rate thus appears limited to about 1 Gbit s over standard fibre, and about 10 Gbit/s over dispersion shifted fibre,where the fibre has an average dispersion of -2 ps/nm.km, for transoceanic distances. For distances of a few thousand kilometres, 2.5 Gbit s transmission should be possible using standard fibre. This could be particularly useful in applications such as low cost island hopping systems, and certain terrestrial applications.
• the following design rules are put in place of the soliton transmission rules: i /
• the pulse width is greater than 20 % of the bit period.
• the dispersion of the system must be such that over any one span the broadening of the pulse width due to dispersion in the linear regime is a small fraction (less than about 10 %) of the pulse width.
• the greater the total system length the smaller the allowable dispersion in each span.
• the intensity at the beginning of each span is chosen to be larger than the corresponding 'soliton' intensity by a factor such that the non-linear compression counteracts the dispersion over that span.
• Condition (ii) is made possible at reasonably high bit rates because of the trade off involved in the first condition.
• a pulse width which is wider by a factor of 4 ( say) for a given bit rate
• the frequency bandwidth is reduced by a factor of 4
• the dispersion is reduced by a factor of 4.
• this is now 1/16th the relative size for a fibre of a given dispersion.
• an alternative implementation may use dark pulses travelling in the normal dispersion regime of a fibre, rather than the anomalous regime
• a dark pulse is the absence of a "pulse" on a constant higher intensity background, rather than a signal against a low or zero intensity background.
• These dark pulses may also be generated using the modulation technique described below, substituting a filter to exclude the transition frequencies.
• non-dispersion shifted (standard telecommunications single mode) fibre is used and the signal is transmitted at
• a distributed feed back (DFB) laser 35 61 is adequate to use as the single mode high coherence light source.
• the DFB 61 is biased with a constant current above threshold, and lases at a wavelength near the peak of the gain spectrum of the amplifiers in the system, say 1535 nm for an erbium-doped fibre amplifier.
• the output of the DFB 61 is coupled into a single mode fibre 62 and passed through a fibre pigtailed Lithium Niobate, Mach-Zender external modulator 63.
• the data 64 at a bit rate of 2.5 GBit/sec is used to modulate the External modulator 63 in such a way that a pulse of 141 ps temporal duration (full-width half-maximum) is generated for each bit of information.
• This is achieved by first generating an appropriate electrical pulse train via pulse generator 65 or by the use of other methods as described below (i.e. using the transition of a square wave to generate the pulse).
• pulse generator 65 or by the use of other methods as described below (i.e. using the transition of a square wave to generate the pulse).
• sech squared intensity shape for the pulses. It will be appreciated that while sech squared is a preferred shape, it is not essential to the invention and any suitable pulse shape may be used in practice.
• Each pulse has an electric field vector opposite to the preceding pulse, to ensure that in spite of pulse overlap there is a zero intensity region between the pulse where the electric fields are equal in magnitude but opposite in sign. Having the pulses with alternating sign produces a pushing apart of pulses upon non-linear propagation. If they were in phase, there would be a pulling between pulses, in a similar fashion to that encountered in soliton propagation. Either the pushing or pulling regime may be used, but the former is used for the purposes of this example.
• the resulting optical pulse train is then amplified by the booster amplifier 66 to the correct power level before being transmitted along the first span of fibre 67 0 . In this numerical model the power level corresponds to a peak pulse power of
• the first booster amplifier 66o and subsequent in line amplifiers 66 n are chosen for this example to be Erbium-doped fibre amplifiers with the length chosen to give maximum gain at the signal wavelength of
• this example may underestimate the power required slightly and so the power chosen in practice should be determined exactly for the fibre characteristics used in order to as nearly as possible replicate the original pulse train after one span and amplification period, as will be understood by those skilled in the art.
• the span length we have chosen is 100 km of standard telecommunications fibre or pure silica core fibre when an absolute reduction in losses is required. In cases where noise considerations become dominant it may be necessary to decrease this length slightly, but for total system lengths of 2000 km this span length is feasible.
• 19 in line amplifiers 68 0 - 68 ⁇ have been used, and at each amplifier the signal power is maintained at the correct value to achieve minimal pulse broadening or distortion.
• each of the amplifiers 68 n will have identical output power, though some slow transition to high powers at the end of the span to overcome noise floor limitations may be imposed.
• the signal is regenerated by a combination of a erbium preamplifier 69, an optical filter 70 and a receiver 71 (a pin photodiode for example).
• the system described in the example can employ non-dispersion shifted fibre and large repeater spacings over distances of several thousand kilometres. This provides a clear cost advantage over systems requiring specialised fibre.
• soliton systems generally require amplifier spacings at least every 40 km, creating additional reliability and cost burdens on the system simply because more amplifiers are required.
• a conventional NRZ system over the distance proposed in the example would require dispersion shifted fibre, with the dispersion minimum shifted to nearly the signal wavelength of 1.55 ⁇ m.
• the design of such a system is complex, to ensure the average dispersion requirements are met. Further, the cost of the fibre, higher losses associated with the fibre and hence need for more repeaters again necessitate a more complex (and hence less reliable) and more costly system.
• the exact average dispersion requires less rigid tolerances, so that it is easier to operate at any wavelength in the given region.
• filtering is minimised or avoided and multiple wavelength transmission, i.e. wavelength division multiplexing, may be used to upgrade the system to higher bit rates.
• Techniques for this include gain switching which involves applying an electrical pulse to the laser, and mode-locking, which involves applying gain at a constant repetition rate related to the time for the pulse to travel around a resonant cavity formed by a loop or a mirror.
• the first method results in large chirping and consequent dispersion problems
• the second method requires that the pulse stream be externally modulated to remove the zeros and can only be used at discrete repetition rates. It is therefore desirable to provide a system able to generate a pulsed stream of data suitable for implementation in the transmission system described above, with minimal chirp at high bit rates, requiring only small drive voltages and no external modulation.
• the present invention may be implemented using a modulation system which takes an electrical NRZ data stream and converts it to an optical pulsed data stream with very low chirp.
• the idea is to generate a pulse every time there is a transition in an NRZ signal, from 1 to 0 or from 0 to 1.
• the electrical NRZ is used to frequency shift the laser diode in a similar fashion to FSK.
• a filter is then used to pass the frequencies not of the 1 "s or 0's but the intermediate frequencies, corresponding to the transition. As such light will only pass during the transition from a one to a zero, in the form of a short pulse of light at each transition. This is illustrated in Figure 4.
• This system will be described hereinafter as PTFSK (Pulsed Transition Frequency Shift Keying).
• the filter pass band will generally be smaller than that required for FSK, and is chosen according to the width of the pulse required and the rise and fall times of the driving current.
• the filter is capable of being less sensitive to non ideal frequency modulation response of a laser because it acts on a transition which can be made as sharp as required by increasing the frequency difference between the two levels.
• a properly chosen filter will give a transform limited pulse, i.e. the minimal range of wavelengths which is theoretically possible, and accordingly pulses suitable according to the present invention.
• the pulsed data stream achieved is not equivalent to the original data stream so it is necessary to decode the data either at the transmitter or the receiver.
• One way of implementing this is to construct from the original data at the transmitter a second NRZ data stream which will produce the correct pulse stream.
• This derived NRZ data involves generating a transition or a flip from one electrical level to another every time a one passes. This is illustrated in Figure 5. This can be implemented with straightforward electronic logic as will be apparent to those skilled in the art.
• PTFSK ( Figure 6) consists of an electrical converter stage 42 which converts a NRZ or RZ electrical signal 41 to what we can term the transition triggering NRZ (TTNRZ) signal 43.
• TTNRZ current 43 can be combined together in the usual fashion employing a bias Tee 45.
• This current signal is then applied to a Distributed Feed Back (DFB) laser 46, with the bias set well above threshold.
• the optical output 47 is then filtered using a Fabry-Perot filter (or similar) 48 at a frequency midway between the frequency of the 1's and 0's.
• the pass band of the filter is chosen to exclude the two frequencies for one and zero, but to allow a range of transition frequencies to pass.
• the output is then a PTFSK signal 49.
• Both the bias current to the laser and the pass band of the filter may need to be stabilised in a feed back loop to ensure that the filter picks out only transition frequencies.
• An optical amplifier 50 may be used to amplify the system to a sufficient output power for transmission through fibre 51 , and to compensate for the losses through the filter.
• PTFSK may be implemented using other hardware, for instance using external modulation of a CW laser.
• the drive voltage to the modulator corresponds to an off or both the one and the zero bits. These two voltages can be chosen so that there is an 'on' state in between them. As such the light will only pass when there is a transition between a one and a zero.
• the width and amplitude of the pulse depend upon the rise and fall time of the driving voltage. The rest of the implementation would remain substantially unaltered.

Landscapes

• Physics & Mathematics (AREA)
• Electromagnetism (AREA)
• Engineering & Computer Science (AREA)
• Computer Networks & Wireless Communication (AREA)
• Signal Processing (AREA)
• Optical Communication System (AREA)

Applications Claiming Priority (2)

Publications (2)

Family

ID=3775018

Family Applications (1)

Country Status (3)

Families Citing this family (16)

Family Cites Families (2)

• 1991
  • 1991-10-18 WO PCT/AU1991/000484 patent/WO1992007430A1/en not_active Application Discontinuation
  • 1991-10-18 JP JP3517598A patent/JPH06504415A/ja active Pending
  • 1991-10-18 EP EP19910918557 patent/EP0553193A4/en not_active Withdrawn

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events