Technology of 400G optical modules

With the IEEE 100 Gbit/s (hereinafter referred to 100G) Ethernet standard-setting discussions with the end of the world's major manufacturers are promoting the global deployment of 100G and 400G focused its attention to even 1 Tbit/s system up. With 40G / 100G, like, 400G deployment should be a gradual manner. In order to make more effective use of existing DWDM (dense wavelength division multiplexing) line resources, reduce investment costs, operators can expect 400G deployment on existing networks, rather than re-design and build a new network to accommodate 400G of transmission. This means that you must adapt to the network design and planning 400G 100G / 40G or 10G to achieve a mixed deployment of 400G, 100G / 40G's.

1. Technical analysis 400G LAN Interface optical modules

400G LAN (Local Area Network) interface to optical modules may 100G Ethernet will continue to use proprietary parallel transmission. In February 2011, Finisar on "Beyond 100GE" seminar presented 400GE modules standard recommendation, primarily to support 400GE-LR16 and 400GE-SR16 two applications. Where 400GE-LR16 using 16×25G LAN WDM (1330,1310,1290 and 1270nm 4 waves channels) to achieve, and 400GE ~ SR16 is using a 16×25G multimode fiber interfaces. In addition, in the physical layer defines the CAUI (Attachment Unit Interface) -16, CPPI (parallel physical interface) -16 electrical interface standard. 16×25G 100GE just linear expansion, as long as the process to meet the requirements, there is no other technical difficulties. In contrast, the optical fiber parallel development will have more space, but supports higher densities needed photonic integration technology can make 400G commercially possible.

In addition, Finisar also proposed other possible ways 400G LAN applications: the first type is the use of already commercialization 4OG of EML (electro-absorption modulated laser) technology architecture composed of 10×40G; the second was that rely on technology to improve the EML, using 4-level amplitude modulation and DSP (digital signal processing) of dispersion compensation framework 8×50 G; third is 4×100G of architecture that must be complex amplitude phase modulation, such as PM-QPSK ( polarization multiplexing - Quadrature Phase Shift Keying), there is no possible commercialization of the technology demonstration.

NTT research report also pointed out that in the serial data transmission, 16×25G, 10×40G and 8X 50G of these types of architecture are possible. For 50G, modulation is also based MZ (Mach - Zehnder) the DQPSK (differential quadrature phase shift keying) modulation or OOK (on-off keying) modulation. Each architecture in size, cost, power consumption, etc. have their own advantages and disadvantages. From the current research point of view, DML (directly modulated laser) production process is simple, low power consumption, but the ER (extinction ratio) is small. EML complex production process, power consumption is relatively large, but the larger ER, you can get a very clear eye. Further, the quantum well EAM (electroabsorption modulator) InGaA1As based valence band offset is reduced due to a hole caused by accumulation in the modulation process, and therefore suitable for use as high-speed modulation. Figs. 1 to 3 are NTT in OFC 2011 show for the 400 GE of 1300nm, 50G EML spectrum, into the relationship between the current and the optical power and the transmission 10 and 40 km after the output eye diagram.

For 400GE systems, 50G of OOK modulation due to the physical size of the entire transmitter end, is a good compromise choice. In this modulation, DML comparison more difficult to implement. So we can predict, EML and 8×50G of OOK modulation prospects in 400GE systems.

2. Technical analysis 400G long-distance transmission optical modules

In the ITU-T / IEEE Joint seminars, Alcatel-Iucent report stated: interface cable OTU5 rate will reach 449.219 Gbit/s. With the increase rate, the system OSNR (optical signal to noise ratio), CD (chromatic dispersion), PMD (polarization mode dispersion) and nonlinear increasingly demanding. 400G signal dispersion tolerance of only 0.5 ps/nm, the 100G is 1 / 16.400G on OSNR also encountered a challenge, 6dB higher than the 100G. In particular, the use of higher than the current 7% FEC (forward error correction) after expenses, can achieve longer distances. The current discussion was more of a 25%. 400G 100G in PMD than the challenge faced bigger, 400G PMD tolerance of only 0.25ps, the 100G 1/4.

2.1 laser linewidth requirements

With the development of digital coherent receiver technology, the high-end multi-level modulation format because of its high spectral efficiency characteristics in DWDM systems are increasingly compelling. Phase noise characteristics of the laser transmitter and receiver optical local oscillator determines the bit error rate performance of the system. Table 1 lists the different modulation formats, 400G and 100G comparison with the laser line width requirements. Table, △fTX represents transmitter laser linewidth, △fLO represents wide receiver local oscillator light.

2.2 modulation formats and channel spacing feature

To meet the 400G DWDM transmission system in the current requirements and improves overall system capacity modulation pattern of the most important requirements are: to meet the SE (spectral efficiency) and OSNR sensitivity requirements, and have a very strong non-linear tolerance.

Theoretical capacity of single-mode fiber is 8 bit/s/Hz, the actual long-distance transmission equipment and optical fiber, a maximum of 4 bit/s/Hz. In modern optical communication systems, modulation format a great impact on system performance, in order to achieve with the existing network of 10G, 40G hybrid deployment, to achieve 80 wave 50 GHz spacing, it must meet high SE, which can be single-carrier high ary modulation or multi-carrier transmission is achieved. For 448G transmission system, consider the device frequency drift and R0ADM (reconfigurable OADM) non-ideal characteristics, requires practice must be 45G of 32QAM (quadrature amplitude modulation) modulation or 28G of PM ( polarization multiplexing) -256QAM. Electrical domain OFDM (Orthogonal Frequency Division Multiplexing) modulation of a single carrier can be substituted, as the complexity of both the DSP's, but due to the extra information OFDM cyclic prefix, preamble and training overhead symbols, etc., usually higher than the corresponding single-carrier format The SE is lower.

In order to try to meet a 50GHz DWDM channel spacing, many theoretical studies in 2010 are based on multi-level amplitude modulation, which PM-256QAM, a total of 65,536 constellation points, compared to 100G PM-QPSK, the density increases 8 times, and noise and light XPM (cross-phase modulation) / SPM (self-phase modulation) is very sensitive to transmission distance is very short. From the current situation reported demonstration of single-carrier QAM speed opinion, short term 448G transmission, whether it is a single carrier OFDM PM ~ 256QAM or electricity of 32QAM, it is not yet commercialized.

The first one kind of relaxed solution required SE: abandon 50GHz spacing WDM mandatory requirements, such as a 16QAM using 56G PM and flexibility of 70 ~ 80 GHz spacing WDM, SE of 6 ~ 5 bit/s/Hz, and the need There are more than enough to support ROADM systems. Data center users tend to use this flexible solution, but with large-scale, multi-service mesh network carriers to stick with standard spacing to 50GHz. For compatibility with a 50GHz boundary conditions, the use of inverse multiplexing can channel into two 448G 224G wavelength. 28G PM-I6QAM modulation up to 4 bit/s/Hz net SE, compared to 100G PM-QPSK, doubling the capacity of each fiber is increased in WDM. In addition, in order to achieve 10G, 40G / 100G to 400G seamless upgrade deployment of WSS (wavelength selective switch) presents an adjustable bandwidth requirements.

The first two kinds of solutions: using higher order modulation 32QAM or low photon carrier orthogonal multiplexing to replace the single-carrier 448G signal. This method is called coherent WDM or coherent optical OFDM. It is with inverse multiplexing DWDM different nature, because it can be obtained at a specific single carrier modulation format and the same SE coherent reception OSNR tolerance. A 448G of the transmitter can be used 10 separate photons modulated quadrature carriers. The receiver can be divided into two groups (each five as a group) to detect the reception. From the above discussion it seems, in order to obtain the highest possible rate of subcarriers can be processed on the electric field parallel to maintain a minimum light the large ones would be a more practical and more economical solution. Table 2 lists the OFDM and single carrier 400G system performance comparison.

Due to high sensitivity compared to a single carrier and excellent CD / PMD tolerance, multi-polarization digital coherent receiving CO (g coherent light) -OFDM becoming more promising and began to receive widespread attention in the industry. In the future exploration of 400G optical modules based on OFDM modulation structure, the main three kinds of architecture: Based on the FFT (Fast Fourier Transform) OOFDM (optical orthogonal frequency division multiplexing), all-optical OFDM and electro-optical OFDM.

Traditional OOFDM based DSP / DAC of IFFT (Inverse Fast Fourier Transform) and FFT signal synthesizing demodulation, CD and PMD tolerance can be inserted by a cyclic prefix or a guard interval, promoted training symbols, but this will resulting in a 10% or 20% of the overhead, and increase line speed. Especially in need of period CD compensation transmission line, the transmission performance DSP-based multi-carrier OFDM will be limited fiber nonlinear characteristics. The use of silicon-based PLC (Planar Lightwave Circuit) and hybrid integration technology LN (lithium niobate) lightwave circuit, the two carriers have been able to make the QPSK (quadrature phase shift keying) modulator to achieve single polarization modulator 100G (25G) and dual polarization modulator 111G (13.9G). Taking into account the complexity of the transmitter, unlike conventional DSP-based OFDM, the number of sub-carriers must less (usually 2 to 4), because of less number of subcarriers can effectively reduce PAPR. In addition, the transmitter does not require DSP and DAC. The use of a small amount of carrier, loop overhead will lead to additional costs or limit the ability to compensate. Because we need a receiver using a linear filter based on CD / PMD compensation.

All-optical OFDM, must be inserted GI (guard interval) to improve, CD and PMD tolerance, and requires a long symbol periods (many sub-carriers) to offset the cost of GI caused by electro-optical OFDM therefore proposed architecture to solve issue, electro-optical OFDM architecture to meet the higher speed requirements. Table 3 presents a comparison of the performance of 400G systems in a variety of different modulation formats.

In summary, the use of fewer sub-carriers, all-optical OFDM has the following two advantages: no emission end of DSP / DAC, has a relatively low electrical, optical complexity; As a result of fewer subcarriers , thereby reducing the PAPR of the signal, there are in the CD or the low dispersion compensating optical fiber line, having good nonlinear suppression. Therefore, from the cost, performance and complexity, etc. to achieve the look, all-optical OFDM modulation technique (2SC-DP-16QAM format) and channel spacing flexible optical module vendors can attract attention, it will play an important in the early commercial 400G role.

3. 400G detection technology

3.1 micro-optics and hybrid integration technology

For coherent detection, the use of separate free space 90° balanced mixer and a light detector to build coherent receiver system, this complex configuration to achieve commercialization is difficult. From the 2009 ECOC, U2T and HHI demonstrates monolithic integrated PLC 90°. Mixers and two pairs of high-speed balance PD receiver, to the 2010 ECOC, U2T and HH1 once again demonstrated monolithic two-way PLC 90°. Mixer and eight high-speed balancing PD receivers. From the evolution and development trend in recent years, it seems 100G transmission technology, 400G line receiver technology is gradually moving towards integration.

The integrated use of monolithic integrated receiver and free space optics, there is no way to get satisfactory performance, reliability and low cost. In this regard, NTT PLC technology using a silicon PBS (polarizing beam splitter) and 90°as a monolithic integrated optical mixer DPOH (polarized light mixer). Another use of a new multi-channel collimator DPOH the coupling loss between the lower and the PD, and to suppress coupling deviation caused by temperature changes. In addition, NTT has developed a high-speed chip-level compact photoelectric conversion structure, and use these techniques to produce an integrated coherent receiver. Micro-optical alignment technology allows greatly improved sensitivity based PD matching and temperature performance PLC hybrid integrated devices.

Since the silicon-based microelectronics plane technology, optoelectronic devices is three-dimensional technology. Compared to hybrid integration, PIC (Photonic Integrated Circuit) can significantly reduce the size of the optical module, save packaging costs and flat connection allows the optical path to match and balance easier, thus effectively control deviation, it is the future mainstream. Currently Bell Labs has been research on monolithic silicon integrated coherent detection technology, but there are many technical difficulties to be a breakthrough. 400G early in the silicon PLC and free space optics hybrid integration will be more mature commercial programs.

3.2 with detection technology

Digital coherent reception technology in the field of high-speed transmission is generally regarded as a promising technology because it can increase the sensitivity of the system OSNR, CD and PMD compensation line transmission damage. Because electricity rates "bottleneck", ADC sampling rate will be limited to a very long period of time of less than 100 GS/s. In order to effectively address this issue in 400G even T bit/s channel transmission, the use of multi-carrier modulation format is an effective method.

The use of wavelength-independent detection, the receiver hardware complexity can bring lower by detecting a plurality of sub-carriers, not only from ADC sampling rate bottleneck constraints, and reduce the load on DSP. The structure shown in the block diagram of an optical transceiver module 4.

3.3 processing capabilities within power

Currently in 100G systems, most of the optical device has reached the extent of commercial, but in coherent receiver technology, the key to the biggest problem ADC and DSP chip production business is the processing power and power consumption. Although in 2010, Alcatel-Lucent has 112 G's long-distance system uses a 70M + door of 56 GS/s of ADC / DSP, but faces the same prediction in the 400G system bottleneck.

4. Conclusion

400G long-distance communication and transmission of the light entering a new era, the multi-carrier multi-level phase modulation and an array of optical communications is moving from single-carrier modulation with coherent detection of the polarization multiplexed coherent change detection. Photonic integration and electronic integration, ADC / DSP technology will be 400G optical communication modules and systems business of the key. With the urgent need for standardization of Ethernet, parallel light of the requirements of photonic integration technology will be a huge boost. In the next 2 to 3 years, 400G electro-optical OFDM-related technologies will gradually mature, although these devices have a certain distance from the commercial cost and power consumption, but as these technologies have matured and related standards of discussion and development, 400G commercial system is about to pull the curtain.

Popular Posts