GB2138233A - Optical transmission - Google Patents

Abstract

In an optical transmission system three level differential coding of the optical signal is employed to transmit a two-level input signal. The three-level (dicode) optical signal avoids the ramping effect otherwise caused by long strings of binary ones or zeros, and improves the dynamic range of the receiver. <IMAGE>

Description

SPECIFICATION Optical transmission This invention relates to optical transmission, and in particular to methods of, and apparatus for, optical digital transmission.

In digital optical communication systems employing high impedance PIN-FET receivers, dynamic range limitations occur due to the ramping effect when, for example, non return to zero (NRZ) binary sequences are applied to the integrating front end.

In particular, the front end of a receiver normally requires a bleed resistor to drain the mean level input, causing the front end to behave as a leaky integrator. Receivers, whether PIN-FET or another kind with this characteristic behaviour will be referred to hereinafter as integrating receivers.

The present invention is based on the appreciation that a modified form of a known code called dicode, which has not previously been employed in optical systems, aids in overcoming the aforementioned dynamic range limitations. The code per se, in its application to electric signals, has been published by Valin J, "Codes used for the transmission of binary data", Thompson CSF Eng Rev Vol 11 No 2June 1979 and also in US Patents 2759047 (L A Meacham) and 3621141 (D E Mack). However, this code, because it requires positive and negative deviations from zero current, cannot be applied directly to optical transmission systems.

According to one aspect of the present invention, a method of optically transmitting binary digital information comprises encoding a two-level input signal as a three level optical signal such that a transition in the input signal from low to high is encoded as a first optical level, a transition from high to low is a second optical level, and absence of a transition as a third optical level intermediate the first and second optical levels.

Coding of NRZ binary data may be achieved by introducing a bit time delay into the two level signal and subtracting the delayed from the undelayed waveform.

According to another aspect of the invention, an optical digital transmission system comprises a transmitter, an optical transmission medium, e.g. an optical waveguide transmission line, and a receiver, wherein the transmitter comprises means to encode a two-level digital input signal sequence to a three level optical signal sequence for transmission, and the receiver comprises means to decode the three-level optical signal sequence to a two level signal sequence.

Said means to encode preferably operate to convert an electrical two-level input signal into the three-ievel optical output signal.

Said means to decode preferably operate to convert the three-level optical signal into a two-level electrical signal.

Preferably, the encoding means comprise means to generate a signal sequence delayed by one bit period relative to the input signal sequence, and means to generate the optical three level signal from a comparison, e.g. by subtraction, of the input and the delayed input signal sequences.

The present invention is particularly, though not exclusively, suitable for optical transmission employing intensity modulation.

Since the three-level code is essentially a differentiating code, it may be directly decoded by the integrating behaviour of the receiver. This, however imposes a linearity constraint on both the laser and its drive circuitry. Thus it is required that the three-level coder should produce pulses which are symmetrical about the mid-level which itself should remain independent of the input binary state. This disadvantage is, however, easily outweighed by the advantage provided by the present invention of overcoming the dynamic range limitations associated with conventional binary transmission.

Also, it has been found that receiver sensitivities close to that achievable with binary data can be obtained and that providing decision feedback may even allow some sensitivity improvement (1.1 dB). Three-level coding combined with decision feedback offers improvements in the dynamic range, and some potential for receiver simplification particularly at high bit rates.

Moreover, three-level optical transmission in accordance with the invention has the advantage of being largely compatible with high-impedance pin-fet receivers employed for ordinary binary transmission.

The invention will now be explained further by way of example and with reference to the accompanying drawings of which: Figure lisa block diagram of an NRZ binary signal to three-level optical signal convertor; Figure2 is a circuit diagram of a practical implementation of a coding circuit to generate the waveform b of Figure 3; Figure 3 illustrates some of the typical signal waveforms; Figure 4 is a schematic block diagram of a decision feedback equalizer circuit part of a receiver circuit; Figure 5 is a graph comparing the three-level code with binary code as a function of cut-in frequency 913 with and without decision feedback.

Referring now also to the drawings, and referring hereinafter to the three-level, modified dicode, code for the sake of brevity as "dicode", dicode is a coding scheme whereby positive transitions in a binary input signal are coded as 'l's corresponding to a high level light intensity, negative transitions as '0's, corresponding to a low level light intensity and the absence of transitions as a constant level midway between the 0 and 1 levels. Dicoding may be achieved, for example, by using a delay line 1 and a subtractor 2 as shown schematically in Figure 1. The same two-level signal (Figure 3a) is applied to the non-inverting input (+) of subtractor2, and to the input to the delay line 1.After a delay by 1 bit period in the delay line 1, the delayed signal is applied to the inverting (-) input of subtractor 2, and the undelayed signal and the delayed signals are subtracted to provide as output the three level signal shown in Figure 3b.

The signal of Figure 3b is subsequently applied to a laser light source 11 to drive the light source 11, thereby to modulate the light output directly. Alternatively, a constant output light source may be employed, and the signal of Figure 3b may be applied to an external modulator (not shown). The modulated light is then injected into an optical fibre 12 for transmission to a receiver. The modulation is conveniently intensity modulation, with the optical signal taking on one of three intensities, low, medium or high in accordance with the electrical signal applied to the laser source 11 or the modulator.

The optical signal transmitted over the optical fibre 12 (or via another suitable transmission medium) is received in a pin-fet receiver having a high-impedance front end, such as for example the receiver circuit of Figure 4, described in more detail below. By virtue of its high input impedance the three-level optical signal is converted into a corresponding electrical signal and integrated overtime to provide as output the integrated signal of Figure 3c.

A bit rate independent dicoder has been implemented as shown in Figure 2 using a dual D-type flip-flop 3, 4 a dual OR/NOR gate 5, and transistor circuits T1, T2, Ts and T4, T5, T6. For operation to 500Mbit/s, commercially available ECL components have been used as indicated in Figure 2.

The two D-type flip-flops 3, 4 produce the delayed and undelayed data streams at X and Y. These signals drive the high speed transistors T1-T6 configured as two differential amplifiers with their collectors corss-coupled to add the current generated by the erect and inverted versions of the delayed and undelayed data streams. Complementary versions of the output signal are available from the circuit at terminals 37 and 37'.

The choice of dual packaged components, transistor arrays, and careful biasing and resistor matching has resulted in an output with good balance of amplitude, pulse width and shape for both positive and negative going pulses. The maximum speed of operation is limited to 500Mbit/s by the dual D-type flip-flops.

Typical signal waveforms with an NRZ binary input signal are shown in Figure 3a and b. Integration of a dicoded signal with a mid-level at 0 volts results in the recovery of a uniform amplitude signal Figure 3c which may be subsequently restored to binary NRZ Figure 3d with the aid of a threshold detector of known construction such as the threshold detection and decision circuit 46 of Figure 4.

The receiver shown in Figure 4 comprises a conventional pin-fet detector stage, consisting of a photo-pin diode 41 and a resistor R43 in series, the common mode of the diode 41 and resistor R43 being connected to an FET preamplifier 42. The outer terminals of the diode-resistor 41/R43 series circuit are connected to a voltage supply rail 49 and to ground respectively. As discussed below with reference to a theoretical model, reception of the optical dicode signal carries a penalty in receiver sensitivity. However, by providing a fast feedback circuit 47 between the output of the threshold detection and decision circuit 46, the sensitivity of the receiver can be restored to, and under certain circumstances improved over, that for ordinary binary code.

The present invention may be better understood from the following theoretical model. Throughout this analysis the nomenclature used by Personick (Personick S D, Receiver design for digital fibre optic communication systems, part land II, Bell System Technical Journal, Vol 52, July-August 1973, pp 843-886) will be adhered to and the transmitted symbol rate 1/T is assumed to be identical for binary and dicode.

The dicoder output pulse spectrum GD (jw) is given by GD(jw) = GB (jW) (1 e ) where G5 (jew) binary input pulse spectrum and T is bit time.

For a rectangular NRZ binary input pulse it can be shown that the dicoded optical received pulse spectrum will be

The dc component of the optical signal, corresponding to the mid level of the dicode, is assumed to be subtracted out by the receiver.

In order to assess receiver sensitivity performance using dicode we need to specify a pulse spectral density at the pre-threshold detection filter output.

Normally with binary transmission a raised cosine spectrum is chosen. Such a spectrum is unacceptable for dicode since the filter would need an infinity at dc. Consequently a high pass filter characteristic HH(+} iS included for dicode transmission.

Thus HcuT(4)) 1/2 (1 + cos 4)). HH(+) -1 < 4) < 1 If HH (4)) iS realised as a single pole high pass RC filter the noise bandwidth integrals 12, 13 can be shown to be

where 4)3 < < (2r is the normalized 3dB cut in frequency

Letting a parameter Z be the normalized equivalent number of input noise current electrons per bit and receiver sensitivity Pr x z, then for NRZ binary transmission, including signal shot noise,

and for dicode

where q = electronic charge C SW = equivalent input noise current spectral density A2/Hz SE = equivalent input noise voltage spectral density V2/Hz CT = total input capacitance F 4Q2 = signal to noise ration 12B = 1.1275; 138=0.1738 The high pass filter however will introduce droop into the decoded and equalised signal causing 'eye' closure. Since a single pole RC network is assumed the penalty E arising from this effect is given by E = 2exp(-24)3N)-1 where N is the binary NRZ digital sum variation.

However it is possible to regain nearly all of the sensitivity loss arising from droop by adopting a decision feedback (DFB) equaliser structure of Figure 4.

In Figure (5) the sensitivity advantage, computed from equations 7, 8 and 9, of using dicode, with and without DFB, over normal binary transmission is shown as a function of 4)3 for a pin-FET receiver having the following parameters: Bias resistor = 10MOhm; Total leakage current = 10 nA FETtransconductance = 15 mS; Total capacitance = 0.5 pF Bit rate = 1.2 GBit/s; Coded Binary Digital Sum Variation = 6.

Using dicode transmission, receiving sensitivity is less dependent on FET channel noise (SE) than with binary transmission due to the lower value of 13; 0.1048 and 0.1738 respectively. However dicode transmission renders receiver sensitivity more dependent on shot noise and the combined gate leakage and photodiode dark currents plus bias resistor thermal noise (S1), due to the higher value of 12; 32 and 1.1275 for dicode and binary respectively. The last term of the equation 7 and 8 represents the respective signals shot noise limits; this term is higher for dicode than for binary. However with DRB 12 can be much lower than the value obtained for optimum performance without DFB, see Figure (5). At high cut-in frequencies 4)3 the dicode sensitivity advantage over binary tends to about 1.1 dB i.e.

lOlo 13B IsD this is the maximum obtainable for the high pass filter realisation considered. With DFB there is a small penalty (0.3 dB), even at optimum 4)3(2.4 x 10-3) for using dicode.

Dicodetransmission requires linear modulation of the optical source without pattern dependency (i.e.

dependency on the preceding signal sequence) for best performance. Also the pre-threshold detection filter for dicode will normally be chosen as a bandpass design rather than the simple low pass circuits normally used with dispersionless optical fibre systems binary transmission. Whilst these requirements are additional to those of binary transmission, the removal of bit rate specific components from a critical part of the receiver circuit nearest the input, and obviating the need for a code having a tightly bounded digital sum variation, are advantages, particularly for very high rates ( > 565 MBit/s) systems.

The term "pattern dependency" refers to the dependency of the output signal, particularly the mid-level signal, on the preceding signal sequence.

In summary, the modified dicode of the present invention offers as advantages that: 1. A differentiator circuit, which was previously needed in the receiver front end is not required. The differentiator circuit had to be a bit-rate specific network, and had to be changed for each different bit rate.

This in turn required access to a part of the circuit where the received signal was still very weak.

2. The receiver becomes more amenable to larger scale integration because of 1., since 'equalisation' can now comprise a pre-threshold-detection filter alone, and although the pre-threshold-detection filter is still bit rate dependent it follows, rather than precedes, the initial amplification stages so that the signal strength is already much greater.

3. Repeater or receiver dynamic range is improved since approximately uniform amplitude signals exist throughout a repeater or receiver.

4. The need for a bounded disparity code may be obviated since repeater dynamic range will no longer be dependent upon the code digital sum variant.

It should be noted however that dynamic range will still be dependent upon photodiode load resistor and supply voltage on voltage rails and also that pre-coding may still be required to ensure, for instance, adequate timing information at the repeater, although precoding is not required for dicode peruse.

Claims (12)

1. A method of optically transmitting binary digital information, which comprises encoding a two-level input signal as a three level optical signal such that a transition in the input signal from low to high is encoded as a first optical level, a transition from high to low as a second optical level, and absence of a transition as a third optical level intermediate the first and second light optical levels.

2. A method as claimed in claim 1 wherein said two-level signal is a binary non-return-to-zero signal.

3. An optical digital transmission system comprising a transmitter, an optical transmission medium, and a receiver, whereby the transmitter comprises means to encode a two-level digital input signal sequence as a three level optical signal sequence for transmission, and the receiver comprises means to decode the three-level optical signal sequence to a two level signal sequence.

4. A system as claimed in claim 3 wherein the transmission medium is an optical fibre waveguide.

5. A system as claimed in claim 3 or 4 in which the encoding means comprise means to generate a signal sequence delayed by one bit period relative to the input signal sequence, and means to generate the optical three level signal from a comparison, of the input and the delayed input signal sequences.

6. A system as claimed in any one of claims 3-5, employing intensity modulation.

7. A system as claimed in any one of claims 3 to 6, wherein the receiver comprises a high impedance pin-fet detector.

8. A system as claimed in any one of claims 3 - 7 further comprising decision feedback means.

9. A system substantially as hereinbefore described with reference to, and as illustrated by, the Figures of the accompanying drawings.

10. An optical transmitter comprising means to encode a two level electrical signal as a three-level optical signal such that a transition in the electrical signal from low to high is encoded as a first optical level, a transition from high to low as a second optical level, and absence of a transition as a third optical level intermediate the first and second optical levels.

11. An optical transmitter substantially as hereinbefore described and as illustrated by Figure 1 or Figure 2 of the accompanying drawings.

12. An optical receiver substantially as described with reference to Figure 4 of the accompanying drawings.