1. Introduction

Future high-performance transmission systems and networks may require a shift from electronic signal processing to high-speed all-optical signal processing. All-optical logic gates and logic functional devices are key units in future all optical network or computing. Many digital devices are required to be realized in the optical domain.

Semiconductor optical amplifiers (SOAs) have been widely used to realize various all-optical processing functions, since they can exhibit a strong refractive index variation together with high gain and allowing photonic integration characteristics [1–6]. Various basic logic gates [3–6] have been demonstrated with SOAs, and many schemes for all-optical logic functional devices have also been reported with SOAs recently, such as 2.5 Gb/s optical half adder based on a single SOA and a PPLN waveguide [7], 10 Gb/s all-optical half-adder with interferometric SOA gates [8], 10 Gb/s all-optical recirculating shift register with SOA assisted Sagnac switch and SOA feedback [9], 2.5 Gb/s all optical half-subtracter and half-adder using two parallel SOAs and a PPLN waveguide [10], 10 Gb/s all-optical full adder using cross-gain modulation (XGM) in SOAs [11]. Based on cross-polarization modulation and XGM in SOAs, Ref. [12] reported 2-to-4 level encoder with 4 level amplitude shift keying (ASK) signal output. However, to our best knowledge, a unit that can perform digital encoder and comparator simultaneously in optical domain has not been demonstrated yet. As we know, a module that performs digital encoder and comparator simultaneously may offer much more flexibility in optical processing systems.

In this paper, we demonstrated simultaneously all-optical digital encoder and comparator at 40 Gb/s based on XGM and four wave mixing (FWM) in three parallel SOAs. We use three SOAs to achieve five optical logic gates which are needed for digital encoder and comparator. Detuning filter was used to improve the dynamic characteristics of the SOAs especially when XGM mechanism was used. The proposed scheme has advantages such as simple structure, high speed and potential much more logic functions.

2. Concept and operation principle of digital encoder and comparator

Operation principle for all-optical digital encoder and comparator can be shown in Fig. 1. As shown in Fig. 1(a) and Fig. 1(c), digital encoder consists of four logic outputs Y0, Y1, Y2, Y3, which are corresponding to four different input conditions. These four different outputs are achieved by four different logic gates: A̅ · B̅, A̅B, AB̅ and AB, respectively. For input signal A and B with bits “00”, “01”, “10” and “11”, output bit “1” appears only at port Y0, Y1, Y2 and Y3, respectively.

For digital comparator, three logic outputs are needed to represent three results after comparing the two digital signals. When A is bit “0” and B is bit “1”, only the A<B output port is bit “1”, and this operation can be represented by A̅B logic. When A and B are both bit “0” or bit “1”, only A=B output port is bit “1”, and this operation can be represented by A⊙B or XNOR logic. When A is bit “1” and B is bit “0”, only the output A>B port is bit “1”, this operation can be represented by AB̅ logic. From above discussions, we can find that Y1 output in digital encoder is identical with A<B output in comparator and Y2 output is identical with A>B output. In other words, all-optical digital encoder and comparator can be achieved by five different logic functions: A̅ · B̅, A̅B, A̅B, AB and A·B.

 

Fig. 1. Concept and operation principle of digital encoder and comparator, (a) digital gate-level diagram of encoder/comparator; (b) optical implementation of encoder/comparator; (c) logical truth table for the encoder/comparator

Download Full Size | PDF

Figure 1(b) shows the principle diagram of proposed scheme for all-optical digital encoder and comparator. Three SOAs are exploited in this scheme. Signal A and B are input signals with wavelength λ_A and λ_B, respectively. SOA1 is used to achieve A̅B logic function at wavelength λ_B based on XGM effect while the optical power of signal A is much larger than signal B. Contrarily, SOA2 is used to achieve AB̅ logic function at wavelength λ_A while signal B is much stronger than signal A. Signal A and B are injected into SOA3 together with a continuous wave λ_cw. FWM and XGM effects occur simultaneously in SOA3. Based on XGM effect, we can get NOR logic at wavelength λ_cw. On the other hand, we can achieve logic AND at the new generated channel based on FWM effect while the optical power of two data signals is nearly equal. Based on the output AND and NOR gates, we can get the XNOR gate by coupling the two outputs together with proper power equalization. Therefore, we can obtain five different logic gates based on XGM or FWM effects in three SOAs, which can be exploited to achieve all-optical digital encoder and comparator simultaneously.

As we know, FWM effect is bit-rate transparent but XGM effect is limited by the carrier recovery time of the SOA. With the assistance of filter detuning, the output performance of the XGM scheme can be improved [13], and 160 Gb/s wavelength conversion was reported. Recently, the same group demonstrated 320 Gb/s wavelength conversion using the same scheme [14]. In that scheme, the blue shifted filter can balance the power of the chirped component and the power of the probe signal during gain recovery process, so the effective recovery process of the scheme can be accelerated. With the help of filter detuning, 40 Gb/s all-optical logic encoder and comparator can be demonstrated in this paper. And we believe that higher operational rates can be achieved with this scheme.

3. Experimental demonstration and discussion

The experimental setup for digital encoder and comparator is described in Fig. 2. The wavelengths of three CW beams generated by LD1, LD2, and LD3 are 1550.7nm (λ_A), 1549.3nm (λ_B), and 1557.3nm (λ_cw), respectively. The data signals with wavelength λ_A and λ_B are modulated by two Mach-Zehnder Modulators (MZMs) to form 2³¹-1 return-to-zero (RZ) pseudo random binary sequence (PRBS) signals at 40 Gb/s. The duty cycle of these RZ pulses is 33%. Two data signals will be separated into two channels by the demultiplexer (DMUX) and one of them is delayed for several bit periods by an optical delay line (ODL). Therefore, two data signals with different data pattern are obtained. The three SOAs (Kamelian NL-SOA) are biased at 250mA, and their recovery times are about 60ps, which is much longer than one bit period. The small signal gain at 1550nm is about 22dB. The polarization control (PC) is used before SOA3 to ensure that the polarization state of the two data signals is nearly parallel and the FWM conversion efficiency will be higher. Tunable four narrow BPFs with 0.32nm bandwidth are used to filter the Y3, Y0, Y1/A<B and Y2/A>B output. We use ATT1 and ATT2 to control the two data power before SOA1 and SOA2. The Y3 output signal is amplified by the EDFA2, and it was combined with Y0 output to realize A⊙B logic. Finally, the optical spectrum analyzer (OSA) and communication signal analyzer (CSA) are used to observe the optical spectrum and waveform of the converted signals respectively.

 

Fig. 2. Experimental setup for all-optical digital encoder and comparator with three SOAs

Download Full Size | PDF

The data A and B before the SOA3 are shown in Fig. 3(i) and 3(ii), respectively. Both signals have a peak power of 2.6mW with extinction ratio (ER) over 13dB. The probe signal has a power of 0.6mW. From spectra shown in Fig. 4(a), we can see that three input signals are both gain and phase modulated by SOA3, and conjugated light is generated at 1547.7nm at the SOA output. The conjugated light is filtered by BPF3 and amplified by EDFA2, then the output Y3 is achieved with good eye pattern as shown in Fig. 3(iii). The output ER of Y3 is 7.94dB. We use BPF4 to filter out the probe wavelength with small detuning of -0.2nm (blue shift), then the output signal shown in Fig. 3(iv) represents Y0 output. The ER of Y0 is 10dB. The Y0 (NOR) output has a high power level, while the Y3 (AND) output has a low power level due to low conversion efficiency of FWM. With the assistance of EDFA2, the Y0 output and Y3 output have a similar power level. Combined by optical coupler, the mixed signal shown in Fig. 3(v) represents A=B output with ER about 6dB. We can observe much noise appears in bit “1”, which is caused by different intensity in the Y0 and Y3 outputs. The output spectra for Y3, Y0, A=B outputs are shown in Fig. 4. We get the Y3 output in conjugated wavelength at 1547.7nm and the Y0 output in probe light wavelength with a little blue shift at 1557.1nm. And the Fig. 4(b) represents the spectrum for the A=B output, with the assistance of coupler. The detail Y0 spectrum assisted with the detuning filter is shown in Fig. 5(a), with small detuning of -0.2nm (blue shift).

In order to obtain the Y1/A<B output in SOA1 or Y2/A>B output in SOA2, the key point is to control the power properly. As shown in Fig. 5(b), the average power of the data A is 17dB larger than data B before SOA1, the output Y1/A<B is shown in Fig. 3(vi). The output ER of Y1/A<B is 13dB. The average power of the data A is 12dB lower than data B before the SOA2. The output Y2/A>B is shown in Fig. 3(vii). The output ER of Y2/A>B is 11.3dB.

From experimental results, we can see that the output ER of the XGM-based process is high and the ER of the FWM-based process is not high due to its low conversion efficiency. The signal quality of A=B output in the comparator is needed to be improved. The effective way is to enhance the FWM conversion efficiency. Because the XGM conversion efficiency is much higher than FWM, the mixing signal quality is degraded.

 

Fig. 3. Output waveforms for different output of digital encoder/comparator, (i) input signal A (ii) input signal B (iii) Y3 (iv) Y0 (v) A=B (vii) Y1/A<B (vii) Y2/A>B

Download Full Size | PDF

 

Fig. 4. Measured spectrum of SOA3 from OSA, (a) dash line: spectrum before SOA3, solid line: spectrum changed after SOA3 (b) the Y3, Y0, A=B output spectrum assisted with filter and coupler

Download Full Size | PDF

We experimentally observed that the probe light for the XGM can also be used to enhance the signal to noise ratio. Recently, it was reported that assisted beam can improve the FWM output performance. Reference [15] demonstrated that using an assisted beam can enhance the FWM conversion efficiency and the signal to noise ratio. Reference [16] demonstrated that a counter propagating CW pump can be used to improve the dynamic characteristic of SOA. If the assisted light is injected to improve the dynamic characteristic, the whole result will be better.

Regarding the tunability of this scheme, it is mainly determined by the operational range of the SOAs. However, the conversion efficiency of FWM process is highly related to the frequency detuning, the frequency span between signal A and B could not be too large in order to achieve high conversion efficiency and good output performance. If signal A and B are tuned with a certain span, this scheme can operate in the whole operation range of the SOA.

 

Fig. 5. Spectrum for Y0, Y1 and Y2, (a) solid line: spectrum of Y0 output, dash line: filter shape (b) solid line: input power for Y1/A<B dash line: input power for Y2/A>B

Download Full Size | PDF

The proposed scheme is multifunctional. In this paper, the function of digital encoder and comparator has been realized simultaneously. It should be noted that it is easy to fulfil other logic functions by properly alternations because four basic logic functions of all input digital condition have been realized.

4. Conclusions

We have presented a simple and multifunctional scheme of the 40 Gb/s digital encoder and comparator based on three parallel SOA structures, utilizing FWM and XGM in SOAs. We obtained five optical logic functions: A̅ · B̅, A̅B, AB̅, AB and A⊙B to fulfil the needed seven outputs of digital encoder and comparator:Y0, Y1/A<B, Y2/A>B, Y3, A=B, respectively. When the power of A is 17dB higher than power of B before SOA1, AB based on XGM is obtained. When the power of B is 13dB higher than power of A before SOA2, AB̅ based on XGM is obtained. When the power of A and B is nearly equal, A̅ · B̅ based on XGM and AB based on FWM are obtained simultaneously. A⊙B is the mixing result of A̅ · B and AB. The filter is blue shift 0.2nm to improve the logic A̅ · B. The output ER for the XGM operation is above 10dB, and the ER for FWM operation is around 8 dB. Wide and clear eye patterns for the five logic outputs can be observed. Much more potential logic functions and higher operation bit rate can be achieved in our proposed scheme.