Ultra-short- and short-reach optical interconnects are the new high growth applications for optical communications. High capacity density, high spectral efficiency, low cost, low power consumption, and fast configurability are some of the key requirements for potential optical transmission technology candidates. Based on recent progress in orbital angular momentum multiplexed optical transmission and optical device technologies, this paper discusses the potentials and challenges of using orbital angular momentum multiplexing in optical interconnect applications scenarios to meet above requirements.
© 2015 Optical Society of America
Driven by the even-growing data volume from fixed and mobile Internet usage, new models of operating the Internet, such as the ‘Cloud’, are increasingly centered around large information technology (IT) infra-structures. The most prominent of these include data centres and supercomputers, where data (content) storage, processing and provisioning take place. Such business models concentrate data in these centres in an unprecedented way, resulting in significantly higher growth compared to the global overall data growth. Some estimates put the growth of data in these centres at annualized rates of around 70-80%, significantly out-strip the average growth rate of the entire Internet, which is widely quoted as about 50-60% p.a .
At the physical layer, the expansion of large IT infrastructure is enabled by the Moore’s Law in terms of the information processing and storage powers of integrated circuit (IC) chips. The growth rate of transistors in processor chips is about 60% p.a., which roughly translates to the same growth rate of data on the chip. According to frequently quoted roadmaps (such as ITRS), on-chip data volume at present is in the order of 5-10Tb/s, rising to 300 Tb/s in the next decade .
The demand to exchange such amounts of data via interconnects becomes increasingly challenging. Interconnects are essentially communications links between various parts of the IT infrastructure. The transmission capacity of each interconnect link largely follows the growth rate of communications technologies. While data volume grows at rates of 50-80% p.a. as stated above, communications link capacity has been growing at a much slower long term average pace of only about 10-15% p.a , despite the use of advanced technologies including optical communications. Only occasionally the growth rate of communications technologies was able to match that of the data volume - when high growth results from the introduction of disruptive technologies. For example, the introduction of wavelength division multiplexing (WDM) in the mid-1990’s resulted in growth of about 70-80% p.a . in communications link capacity over a few years, before it falls back to the long term rates as the new bandwidth resources are depleted.
This imbalance between the data growth rate and interconnect bandwidth growth rate creates what is known as the ‘bandwidth wall’ around data intensive locations or hot-spots, including processor (or processor cores) and memory devices. The inadequate bandwidth linking these hot-spots results in slow data exchange, limiting the overall processing power and performance of systems. In the case of memory devices, the mismatch is even more profound as the data storage density has out-grown communications bandwidth by as much as 75 times .
Interconnects have thus become a major bottleneck for future expansion in scale and performance in IT systems at all levels including intra-chip, inter-chip, inter-board, and inter-rack links. These interconnects have transmission reaches between sub-millimeter and a few hundred meters.
For interconnects, the key metric should be transmission capacity density rather than transmission capacity itself. This is because the implementation of interconnects is fundamentally limited by available space. At the intra- or inter-chip levels, the available chip real estate is very precious and significant cost saving could be gained if interconnects occupy less chip area whilst moving increasingly large amounts of data, i.e., capacity density must be increased. This is to be achieved by significantly increasing the transmission capacity per interconnect thus reducing the number of interconnects, and in the meanwhile making each interconnect more compact. At the inter-board and inter-rack levels, space is slightly less restrictive, but is still a very valuable resource due to the very large numbers of interconnects in modern IT infra-structures.
2. Electric interconnects
Current interconnects are predominantly electrical transmission lines, with perhaps the exception of inter-rack links which is quickly adopting the active optical cable technology.
Electrical interconnects use conductor tracks to carry signals in the form of electrical current and voltage. The performance limiting factors of electrical interconnects have been very well studied , and have been attributed to the ohmic resistance R of the metal track and the capacitance C between the metal track and the ground. Both R and C scale with the length L of the interconnect, resulting in a interconnect bandwidth that scales with 1/L2. Generally, it would be difficult to increase the bandwidth of individual electrical interconnects to much higher than 10 GHz, in particular over relatively long distances (e.g., tens of cm). As it is difficult to implement multiplexed transmission schemes on electrical tracks without requiring significantly additional resources, modern processor ICs need interconnect pins numbered in the thousands to provide adequate accumulated interconnect bandwidth. 1000 pins running 10Gb/s each provides a total fan-out capacity of 10Tb/s. It is hard to see how such technology can be scaled to provide fan-out capacity of > 100Tb/s.
The ohmic resistance R also results in signal attenuation – with on-chip metal track width scaling down towards 10 nm this loss can be significant, giving rise to the need of regeneration amplifiers even for on-chip interconnects, which takes up expensive real estate and reduces capacity density. The increased power consumption of amplifiers will significantly push up power consumption. Some estimates  put the overall chip power consumption at several hundred Watts if multi-Tb/s total data fan-out capacity is to be achieved. Thermal challenges aside, the total electric current needed to delivery such power at CMOS voltages run into hundreds of amperes and would require large number of power pins.
3D integration - stacking chips to form larger memory and couple them closely with processors - is an effective approach for improving performance . 3D integration requires 3D interconnects, which is currently being implemented in the form of through-silicon via-holes (TSVs). In this approach, however, all data has to pass through the bottom layer of the stack. In the absence of interconnect technologies that provides at least one order of magnitude higher bandwidth over TSVs and other electrical interconnects, this concentration of data flux will only exacerbate the bandwidth wall problem at the bottom layer where the chip stack interfaces external world.
3. Optical interconnects
The ability of optical communications technologies to provide very high bandwidth has been abundantly proven in long haul transmission systems, with commercially available single lane capacity of 100Gb/s and wavelength multiplexed capacity in excess of Tb/s in a single optical channel (e.g., one single mode fibre core). The concept of optical interconnect is therefore very attractive in its potential to solve the bandwidth wall problem. Optical transmission does not suffer from RC bandwidth limit and its performance is virtually independent of transmission distance over the length scale of typical interconnect applications of less than several hundred meters. However, interconnect applications pose specific challenges to optical transmission technologies that do not feature prominently in long haul systems. These challenges are mainly manifested in the requirements for high capacity density, low cost, and low power consumption.
The physical density of waveguide-based optical interconnect (which could be implemented on chip, in an photonic layer in a chip stack, or on an silicon interposer layer) is not only limited by the waveguide size (which in silicon can be as small as <0.5μm). The size of transmitting and receiving devices, especially that of the optical modulator and its driver circuit, and that of the receiving amplifiers, would dominate. As photonic components are generally larger in size compared to electronic components, it is therefore necessary to significantly increase the data flux per optical channel in order for optical interconnects to compare favourably with electrical interconnects in terms of the capacity density. Yet it is not very obvious how this could be achieved.
A light carrier wave can in principle carry extremely high data rate by means of time division multiplexing (TDM). Experimental Tb/s TDM transmission systems have existed for sometime [9–11]. However, electronic systems would typically operate at clock rates of several GHz (off-chip) to several tens of GHz (on-chip), which would be difficult to increase very significantly. Implementing TDM at Tb/s and beyond would therefore require time domain multiplexing electronic circuits and quite possible optical circuits, as no electronics yet operate at THz clock rates and even if in future they do, the cost would be prohibitive for interconnects. TDM optical circuits  rely on nonlinear optical effects that are very power inefficient and would take up large chip areas, therefore unlikely to be a favourable candidate for optical interconnects without fundamental breakthroughs in nonlinear nano-photonics.
One of the most important features of optical transmission technologies is the ability to increase capacity density by means of optical multiplexing, which enables massive parallelism that can increase the capacity density for orders of magnitude. The proven WDM technology may have an important role to play in inter-rack and perhaps inter-board interconnects, should the cost of WDM components be brought down for one or two orders of magnitude than current levels, which is potentially achievable with the advent of mass production photonic integration technologies, especially silicon photonics.
However, it is difficult to foresee how WDM could be deployed in intra-chip interconnects. Here cost is not the only issue. Rather, the additional chip real estate needed to implement WDM may negate the increased capacity per physical channel, and therefore any gain in capacity density. Even for silicon interposer [13, 14] based planar (2D) inter-chip optical interconnects, where substrate area is larger, it is not clear yet if WDM could become viable.
In the absence of viable WDM solutions, other multiplexing schemes that could provide the necessary capacity density becomes more important to optical interconnect than to longer haul optical communications systems.
Spatial division multiplexing (SDM) is a terminology that has been used to describe a number of schemes of increasing capacity density in optical communications. In strict sense and in analogy to other ‘division multiplexing’ schemes, this should mean that the channels are spatially separate, yet increasingly closely packed. Multi-core fibre is a good example of SDM. It has been intensively investigated recently [15–17], with experiments demonstrating up to 49 cores per fibre and total capacity of 1Pb/s per fibre (12 core) over 52km. But cores will couple with each other if they are packed too closely, and so there exists an upper limit to the capacity density of multi-core fibres. Even in the relatively mature fibre-based interconnect applications, the need to further increase capacity density is evident as the large bundles of fibres in data centres become difficult to manage and the slots on the backplane of the racks become fully populated.
When pushed to extreme, SDM channels will start to overlap spatially. This is not a problem for free space optical interconnects because photons are bosons and light beams can cross each other in space without interfering, as long as they are not projected onto the same receiver. This feature is exploited in so-call 3D optical interconnect schemes [18–22], where large number of optical channels are implemented between source and detector arrays, and the light paths do cross each other.
3D schemes could be especially valuable to inter-chip and inter-board interconnects. Much of the demonstrated schemes use 850nm vertical cavity surface emitting lasers (VCSELs) as a natural candidate for the sources, as these devices are low-cost, low power consumption, and readily integrate into 2D matrices and can be hybridized onto silicon ICs. The 850 nm wavelength signals are readily detected by Si photo-detector (PD) matrices.
Yet existing 3D optical schemes do not afford high capacity density. This is fundamentally because VCSELs have limited modulation bandwidth – the best reported values are in the order of several tens of GHz [23–25], and is unlikely to increase much further. VCSEL-based interconnects therefore would only out-perform electrical interconnects by about one order of magnitude over ultra-short reaches (over longer distances it clearly wins out). Assuming each VCSEL cell occupies a very small area of ~10x10 μm2, the capacity density of VCSEL-based 3D optical interconnect schemes is actually lower than waveguide based WDM schemes. In fact such array density is already very challenging to achieve due to fundamental limitations of the bulk optics linking the transmitter and detector arrays, which images the transmitter array onto the detector array. Such limitations include diffraction and aberration, and result in crosstalk between adjacent channels .
3D integrated chip stacks may face serious bandwidth bottleneck at the bottom layers. Proposals of ‘optical via’  have been made, where through-silicon optical links could be made using wavelengths of > 1100nm. This would enable any pair of layers to exchange data directly without the signal being carried on intermediate layers, as long as a clear optical window is kept open throughout the stack. While this proposal is attractive, in practice how to implement without significant crosstalk is challenging, as the optical via is a shared transmission medium to which all transmitters and receivers will have to be attached. This represents an extreme form of SDM in which all signals are projected onto all receivers. To avoid crosstalk and to maximize the capacity density, it is necessary to implement some kind of multiplexing schemes through the same ‘optical via’. WDM is unlikely to be applicable here due to cost and chip area issues.
A very important consideration arising along with optical interconnect is the need to perform optical routing. This is because relying on electronic routing will necessitate optical-electrical and electrical-optical (O/E/O) conversion. O/E/O conversion is very costly in many ways, including chip area and power consumption, and would significantly negate benefits of optical interconnects. Therefore, it is important to note that potential optical interconnect technologies should be evaluated not only by their ability to deliver bandwidth or capacity density, but also by their ability to provide optical routing functions. In fact, one of the reasons why WDM is less likely to be a good candidate for interconnects is that wavelength routing would require complicated technologies such as precise tunable lasers and all-optical wavelengths convertors – the latter still not yet available even for long haul systems. Currently, switching in the spatial dimension which works by deflecting the direction of light seems the only viable optical routing scheme in optical interconnect applications. Extra routing resources complementing the spatial switching dimension would be very valuable to optical interconnect networks.
4. Mode division multiplexing and OAM communications
Mode division multiplexing (MDM) has recently become an intensively researched approach for providing extra capacity density. In MDM schemes, all signal channels are of the same wavelength and share the same physical space, but are launched into different transverse modes in the physical channel. It is therefore possible to increase both the spectrum efficiency and the capacity density due to the increased signal channel number. Mode division allows spatial overlap of signal channels, which provides potentially even higher density over SDM schemes such as multi-core fibre. MDM could be seen as extreme forms of SDM
In comparison, WDM increases capacity density but not spectrum efficiency. Where WDM is not a viable solution as is likely the case for many interconnect scenarios, MDM could provide very valuable multiplexing resources.
The concept of MDM is based on the orthogonality between the eigen-modes that should remain uncoupled when propagating in longitudinally invariant waveguides or homogeneous media. The multiplexed mode channels thus should not interfere with each other. However, this ideal scenario is more often broken than upheld in practical scenarios, and the ease of linear mode coupling due to perturbations in the physical media is well known. Linear mode coupling is the main limiting factor of MDM schemes as it gives rise to severe crosstalk noise between channels. On the other hand, the ease of causing mode coupling may give rise to opportunities in optical routing, as the data can be converted between different mode channels relatively easily.
One kind of MDM scheme is exemplified by few-mode communications schemes in optical fibres. Here the transverse mode space is characterized by two mode indices [m, l], with m being the mode index in the radial dimension and l the mode index in the azimuthal dimension of the cylindrical coordinate.
Few-mode communications experiments use all [m, l] values, i.e., a two-dimensional mode space supported by the optical fibre. The mode coupling crosstalk is dealt with using multi-input and multi-output (MIMO) algorithms, originally developed for wireless communications where multiple antennae are used to increase channel numbers. This has been successful, as reported experiments have demonstrated total transmission capacity of 57.6Tb/s over 119km using up to 6 modes [28–31]. As total used mode number M increases further, however, MIMO algorithms require increasing overheads and the required processing resources scales roughly with M2, therefore quickly become impractical. The problem is exacerbated by the fact that many modes populating the [m, l] space tend to be degenerate with their propagation constant very close to each other, which significantly promotes mode coupling and results in a MIMO matrix that is close to full rank. The scalability of few-mode communications schemes therefore remains unclear.
Orbital angular momentum (OAM) modes are a subset of the cylindrical eigen-modes. Usually those with m = 1 are used. The concept that the azimuthal mode index l is associated with the quanta of the orbital angular momentum of photons was first proposed in , while the other momentum components of the photon, including the linear propagating momentum and the spin angular momentum, and their classical manifestations as the wave vector k and the circular polarization states [left, right] respectively, have been well known.
There exists a large body of literature [33–37] about the physics and optics of light wave with OAM. In a classical picture as shown in Fig. 1, OAM is associated with the projection of the propagation momentum or the wave vector k in the azimuthal dimension. This projection causes a phase shift of 2lπ around the optical axis z, with the cyclic boundary condition imposing a self-consistent phase with integer l. The azimuthal mode index l is hence also the quantum of the OAM in a quantum physics picture, in which each photon in the OAM mode has an amount of OAM equaling to l ħ, where ħ is the reduced Planck constant. The classical manifestation of OAM is a beam with a l-fold spiral phase front. The central axis of the beam therefore must be a phase singularity hence has zero field amplitude. The OAM modes therefore have annular intensity cross-section with a dark spot in the centre. OAM modes with different mode index l are orthogonal to each other, as a group of eigen-modes of the electromagnetic wave in the cylindrical space. It is this orthogonality that provides the potential of OAM multiplexing.
By limiting m to 1, OAM multiplexing uses a significantly reduced mode space that is one-dimensional, i.e., with only one mode index variable l. This is a simplification likely to provide significant benefits without sacrificing the number of accessible mode channels, as l can take infinite integer numbers.
First of all, it is relatively easy to multiplex and de-multiplex OAM modes optically, i.e., to launch OAM modes in a concentric manner so that they propagate co-axially, and then separate them into individual light paths, as have been experimentally demonstrated [38–41]. This is very important as a passive optical de-multiplexer is zero power and can be data rate or modulation format transparent, hence much more favourable over MIMO processing. In contrast, optically de-multiplexing a large 2D mode space [m, l] is much more difficult if not impossible.
Within the OAM mode group, degeneracy is completely lifted as all modes have different and well define propagation constants. This reduces linear mode coupling significantly and limits coupling to between adjacent modes with the closest propagation constants. The corresponding MIMO matrix is likely to be much more sparsely populated with significantly values in near-diagonal terms, making any need for signal processing (in order to suppress remaining crosstalk noise) less intensive. Optical engineering of the OAM channels (such as OAM fibres [42–44]) can further reduce the mode coupling be enlarging the non-degeneracy between the modes. Having said above, mode coupling remains the single most important technological issue for OAM communications.
A significant potential benefit of using OAM multiplexing is that it affords optical routing by means of controlled mode coupling that converts data between OAM mode channels. This has been experimentally demonstrated [45, 46] using passive optical components. As a linear mode coupling process, the conversion efficiency can be very high – approaching 100% – compared to typical efficiency in the order of −20dB  of broadband wavelength conversion based on 3rd order optical nonlinearity driven by significant levels of optical pump power. OAM conversion components can have zero static power consumption. Optical routing in the MDM domain, however, is unlikely to be practical for few-mode schemes using the 2D [m, l] mode space, as both controlled mode coupling and mode de-multiplexing would be extremely difficult.
5. Recent progress in OAM communications research
OAM multiplexing has been investigated relatively recently, with much of the literature appearing after 2010.
Free space OAM multiplexing transmission experiments have demonstrated rapid progress in terms of the total capacity carried in a single coaxial beam as well as the spectrum efficiency. In 2012, transmission capacity of 2.5 Tb/s and spectrum efficiency of 95.7b/s/Hz was reported , followed by 100 Tb/s transmission in 2013  and 1 Pb/s transmission in 2014 . Much of these experiments combine the various dimensions of the light wave, using high order amplitude-phase domain modulation schemes and multiplexing in wavelength, polarization, and OAM dimensions. Latest spectrum efficiency record is 230 b/s/Hz over a single optical carrier . These experiments unequivocally demonstrate the potentials of using OAM multiplexing for extremely high capacity communications, in conjunction with existing modulation and multiplexing techniques. OAM switching of data channels in free space has also been demonstrated.
Fibre-based OAM transmission has also been studied since the 1990s, firstly focusing on the mode stability in cylindrical and twisted fibres . Recently, OAM communications over useful distance of 1.1km and high capacity of 1 Tb/s has been demonstrated . OAM fibres supporting higher number of OAM modes with low mode coupling have also been demonstrated .
Significant progress in key integrated OAM components, including transmitters and (de-)multiplexers, has also been made. In particular, compact silicon based integrated OAM transmitting device  and (de-)multiplexers device  are promising enabling technologies for interconnects. Recently, fast (20μs) switchable integrated OAM transmitting device has also been demonstrated . Massive parallelism in OAM (de-)multiplexing has also been demonstrated using specially designed Damann grating .
6. Potentials of OAM multiplexing in optical interconnects
Advancement in free space and fibre based OAM communications experiments and various OAM components hold promising prospects for applications in optical interconnects where extremely high capacity (density) is needed for ultra-short and short-reach links (sub-mm to few hundred meters) within confined spaces.
The most likely interconnect scenarios where OAM communications could become useful include inter-chip, inter-board and inter-rack links. Free space OAM transmission links could find applications at the inter-chip and inter-board levels, while fibre based OAM transmission links could find application at the inter-rack level.
One potentially very useful example of the inter-chip interconnect application would be the ‘optical via’ through 3D stacked chips. Here an effective multiplexing scheme is needed that can penetrate the multiple layers of silicon chips, allowing multiple interconnects between arbitrary pairs of chips to share the optical window hence minimizing chip area used for interconnect, without blocking the links between other pairs. It would enable the bottom layer (or external carrier layers such as the silicon interposer) to optically ‘drill through’ directly to individual layers in a chip stack without affecting the intermediate layers, realizing Peta-bit/s capacity in the 3D stack. Using OAM channels, this could be possible as each layer can integrate mode selective detectors (e.g., using micro-ring based OAM devices as in ) that will detect one OAM mode l while other modes pass through undetected, as shown in Fig. 2. Such a scheme effectively constitutes an optical code division multiple access (OCDMA) network using OAM value l as the address code of each node (layer). Reconfiguration of the network using fast switchable OAM transmitters and detectors similar to those demonstrated in  would also enable very flexible and dynamic networking functions such as multicasting.
Free space 3D inter-chip and inter-board links can also benefit from the extremely high capacity density provided by OAM multiplexing as demonstrated in recent experiments [48–51]. Surface-normal OAM multiplexed coaxial transmission and detection can be made using silicon devices [55–57], interconnected by free space optics [20–22, 27], providing multiple OAM channels between the chips or boards, as shown in Fig. 3. This allows channels numbers to be scaled without having to increase the optical aperture to accommodate more spatially separate channels. It also eliminates the crosstalk between adjacent channels resulting from diffraction. Furthermore, the optical components between the transmitter and receiver can also act as routers by performing OAM conversion. As widely demonstrated , gratings can readily perform OAM conversion and switching. However, currently such components convert the entire set of incoming OAM channels to a different out-going set, shifting their OAM mode number l by the same amount. This is undesirable as it is restrictive to routing algorithms. One challenge will be to realize strictly non-blocking routing by selectively convert any one incoming OAM channel to any other OAM channel.
VCSEL-based parallel transmission could also improve its capacity density by exploring the OAM dimension. Recently we have experimentally demonstrated VCSELs that emit designated OAM modes by integrating simple optical structures in the VCSEL aperture that can be mass-produced as a simple addition to existing VCSEL fabrication process. Figure 4 below shows two examples, one of a VCSEL emitting a single OAM mode and the other a VCSEL emitting a combination of two concentric OAM modes. This is promising as such devices can be used in both free space or fibre-based OAM multiplexed links as low-cost and low power consumption OAM sources. By integrating electrically isolated VCSELs that emit multiple concentric OAM modes, each OAM channel can be individually modulated so that OAM multiplexed communications can be achieved. Implementing such VCSELs in arrays, in which each VCSEL pixel is a multi-OAM source, can further extend the OAM channel count. Combination and separation of such spatially separate OAM channels into co-axial OAM beams can be implemented using specially designed gratings.
A highly interesting grating based OAM multiplexing scheme has been reported while this paper is been revised . In this demonstration, a pair of Dammann optical vortex gratings have been used to multiplex 10 OAM modes originating from 10 spatially separate non-OAM sources. Using 80 wavelengths and 2 polarizations in addition to the 10 OAM modes, a single spatial channel carried a total 1,600 optical data channels with aggregate capacity of 80-160Tb/s. Although this particular experiment was carried out using fibre sources, the scheme is essentially applicable with the VCSEL arrays as sources, which can significantly reduce the cost of OAM interconnects. This experiment, as well as other (e.g., as reported in ) high capacity density experiments, clearly demonstrate the potential of OAM multiplexing for optical interconnect applications.
In inter-rack, fibre based interconnects, OAM multiplexed communications could be a route to further capacity density increase beyond what is offered by multi-core fibres, as high packing density multi-core fibres will start show inter-core crosstalk.
7. Challenges to OAM multiplexing in optical interconnects
Challenges to the application of OAM communications in optical interconnect applications remain severe. The central issue to OAM communications is the crosstalk due to mode coupling. Other challenges include compact integrated OAM multiplexer, mode filter, de-multiplexer as well as OAM mode converter components.
In free space transmission, the main cause for crosstalk is the wavefront distortion due to air turbulences, which damages the ideal spiral wavefronts of the OAM modes and therefore breaks mode orthogonality. This could be especially serious in interconnect between chips and board inside IT equipment, as forced air ventilation is often necessary to cool the equipment. For chip-chip interconnects on the same board, it is possible to seal the optical path in solid media (such as glass or polymer) to avoid turbulence. In solid media (including 3D chip-stacks), the turbulence problem does not exist, but media non-homogeneity will be an issue. Where the OAM optical path has to be in air (e.g., inter-board links), adaptive optics schemes could be employed to correct the wavefront distortion. This has been shown to improve performance significantly by as much as 11-12dB .
In OAM fibres, mode coupling happens because of imperfections in OAM fibres such as bends, twists, non-circularly symmetric cross sectional geometry and refractive index profile, etc. Just as in polarization maintaining (PM) fibres, the main approach to reducing mode coupling is to increase the effective refractive index difference Δneff between OAM mode so that the beat length between modes is reduced – it is generally acknowledged that a Δneff of > 10−4 is required although further experimental verification is needed. As neff is bounded between the core and cladding refractive index values, high mode number and low crosstalk is a pair of trade-off. To support large number of OAM modes and in the meanwhile maintain low crosstalk, it is necessary to use high refractive index contrast fibres, as shown in the recent example of the 16-mode OAM fibre  with air inner cladding. Further development of the OAM fibre may need to use structures similar to photonic crystal fibres. High index contrast fibres tend to have high attenuation. If they are to be used in interconnects of up to 1km length, maximum attenuation allowed is likely to be in the order of 10 dB/km. For each OAM mode, reasonably low dispersion is also necessary in order to support the necessary data modulation bandwidth.
Although good progress has been made in integrate OAM components, to further increase the channel count in OAM (de-)multiplexer device in a compact form will be challenging, especially with high channel isolation performance. The research on integrated OAM conversion and routing devices is in a very initial stage. Although reconfigurable integrated OAM emitters have been demonstrated which could also be used a reconfigurable OAM receivers, an integrated OAM mode converter is yet to be conceptualized, let alone experimentally demonstrated. To achieve these components, new principles and innovative device concept will likely to have to come out of nanophotonics research.
The need for higher transmission capacity density in order to interconnect chips, boards and racks in fast growing large IT infrastructures, such as data centres and supercomputers, provides strong incentive for innovative optical multiplexing schemes. OAM multiplexing could potentially provide capacity density increases of more than an order of magnitude over other advanced modulation formats and multiplexing scheme in a compatible manner. Morever, OAM multiplexing also provides valuable optical routing resource as highly efficiency OAM conversion can be used to route signals. These features make OAM multiplexing an attractive candidate for optical interconnects, in particular where share media transmission is inevitable and other multiplexing schemes – such as WDM – is not viable.
The short transmission distances in optical interconnect application scenarios alleviates the requirement for signal attenuation. The main technical challenge to be overcome is the crosstalk due to unwanted coupling between OAM modes. The non-degeneracy between OAM modes helps to reduce mode coupling compared to other MDM schemes, but engineering of low coupling OAM fibres and wavefront repair technique would still be needed. Despite recent progress, significant challenges also exist in integrated OAM components that can perform selective OAM mode transmission, selective OAM mode reception, and OAM mode (de-)multiplexing, and are low cost, CMOS compatible and low power consumption.
The author acknowledges funding from the National Basic Research Program of China (973 Program) Project No. 2014CB340000, The Natural Science Foundation of China (NSFC) Key Research Project No. 61490715 and the EU Horizon2020 program under project ROAM.
References and links
1. E. Ayanoglu, “Green cellular wireless networks: Where to begin,” CoRR abs/1210.8433 (2012).
2. R. D. Williams, T. Sze, D. Huang, S. Pannala, and C. Fang, “Server memory road map,” Presentation: Server Memory Forum, 2011.
3. B. M. Rogers, A. Krishna, G. Bell, K. Vu, X. Jiang, and Y. Solihin, “Scaling the bandwidth wall: challenges in and avenues for CMP scaling,” in 36th Annual International Symposium on Computer Architecture, Austin, TX, USA, 371–382 (2009). [CrossRef]
4. A. R. Chraplyvy, “The coming capacity crunch,” in European Conference on Optical Communication (ECOC’09), plenary talk (2009).
5. F. Liu, X. Jiang, and Y. Solihin, “Understanding how off-chip memory bandwidth partitioning in chip multiprocessors affects system performance,” in HPCA-16 pp. 1–12 (2010).
6. D. A. B. Miller, “Rational and challenges for optical interconnects to electronic chips,” Proc. IEEE 88(6), 728–749 (2000). [CrossRef]
7. A. Louri and Hongki Sung, “3D optical interconnects for high-speed interchip and interboard communications,” Computer 27(10), 27–37 (1994). [CrossRef]
8. J. U. Knickerbocker, P. S. Andry, B. Dang, R. R. Horton, C. S. Patel, R. J. Polastre, K. Sakuma, E. S. Sprogis, C. K. Tsang, B. C. Webb, and S. L. Wright, “3D silicon integration,” in Proc. of 58th Electron. Components Technol. Conf., Lake Buena Vista, FL, May., pp. 538–543(2008). [CrossRef]
9. C. Schmidt-Langhorst, R. Ludwid, D. –D. Gross, L. Molle, M. Seimetz, R. Freund, and C. Schubert, “Generation and coherent time-division demultiplexing of up to 5.1 TB/s single-channel 8-PSK and 16-QAM signals,” OFC/NFOEC’ 2009, paper PDPC6 (2009).
10. H. C. Hansen Mulvad, M. Galili, L. K. Oxenløwe, H. Hu, A. T. Clausen, J. B. Jensen, C. Peucheret, and P. Jeppesen, “Demonstration of 5.1 Tbit/s data capacity on a single-wavelength channel,” Opt. Express 18(2), 1438–1443 (2010). [CrossRef] [PubMed]
11. T. Richter, E. Palushani, C. Schmidt-Langhorst, M. Nolle, R. Ludwig, J. K. Fischer, and C. Schubert, “Single wavelength channel 10.2 Tb/s TDM-data capacity using 16-QAM and coherent detection,” OFC/NFOEC’ 2011, paper PDPA9 (2011). [CrossRef]
12. T. Yamamoto, E. Yoshida, and M. Nakazawa, “Ultrafast nonlinear optical loop mirror for demultiplexing 640 Gbit/s TDM signals,” Electron. Lett. 34(10), 1013–1014 (1998). [CrossRef]
13. Y. Urino, Y. Noguchi, M. Noguchi, M. Imai, M. Yamagishi, S. Saitou, N. Hirayama, M. Takahashi, H. Takahashi, E. Saito, M. Okano, T. Shimizu, N. Hatori, M. Ishizaka, T. Yamamoto, T. Baba, T. Akagawa, S. Akiyama, T. Usuki, D. Okamoto, M. Miura, J. Fujikata, D. Shimura, H. Okayama, H. Yaegashi, T. Tsuchizawa, K. Yamada, M. Mori, T. Horikawa, T. Nakamura, and Y. Arakawa, “Demonstration of 12.5-Gbps optical interconnects integrated with lasers, optical splitters, optical modulators and photodetectors on a single silicon substrate,” Opt. Express 20(26), B256–B263 (2012). [CrossRef] [PubMed]
14. N. Hatori, T. Shimizu, M. Okano, M. Ishizaka, T. Yamamoto, Y. Urino, M. Mori, T. Nakamura, and Y. Arakawa, “A hybrid integrated light source on a silicon platform using a trident spot-size converter,” J. Lightwave Technol. 32(7), 1329–1336 (2014). [CrossRef]
15. M. Koshiba, K. Saitoh, and Y. Kokubun, “Heterogeneous multicore fiber: proposal and design principle,” IEICE Electron. Express 6(2), 98–103 (2009). [CrossRef]
16. B. Zhu, T. F. Taunay, M. F. Yan, J. M. Fini, M. Fishteyn, E. M. Monberg, and F. V. Dimarcello, “Seven-core multicore fiber transmissions for passive optical network,” Opt. Express 18(11), 11117–11122 (2010). [CrossRef] [PubMed]
17. H. Takara, A. Sano, T. Kobayashi, H. Kubota, H. Kawakami, A. Matsuura, Y. Miyamoto, Y. Abe, H. Ono, K. Shikama, Y. Goto, K. Tsujikawa, Y. Sasaki, I. Ishida, K. Takenaga, S. Matsuo, K. Saitoh, M. Koshiba, and T. Morioka, “1.01-Pb/s (12 SDM/222 WDM/456 Gb/s) Crosstalk-managed Transmission with 91.4-b/s/Hz Aggregate Spectral Efficiency,” presented at the ECOC’2012, Amsterdam, The Netherlands, Jun. 2012. [CrossRef]
18. M. W. Haney, M. P. Christensen, P. Milojkovic, G. J. Fokken, M. Vickberg, B. K. Gilbert, J. Rieve, J. Ekman, P. Chandramani, and F. Kiamilev, “Description and evaluation of the FAST-Net smart pixel-based optical interconnection prototype,” Proc. IEEE 88(6), 819–828 (2000). [CrossRef]
19. M. P. Christensen, P. Milojkovic, M. J. McFadden, and M. W. Haney, “Multiscale optical design for global chip-to-chip optical interconnections and misalignment tolerant packaging,” IEEE J. Sel. Top. Quantum Electron. 9(2), 548–556 (2003). [CrossRef]
20. B. Ciftciuglu, R. Berman, J. Zhang, Z. Darling, S. Wang, J. Hu, J. Xue, A. Garg, M. Jain, I. Savidis, D. Moore, M. Huang, E. G. Friedman, G. Wicks, and H. Wu, “A 3-D integrated intrachip free-space optical interconnect for many-core chips,” IEEE Photon. Technol. Lett. 23(3), 164–166 (2011). [CrossRef]
21. J. Xue, A. Garg, B. Ciftcioglu, J. Y. Hu, S. Wang, L. Wavidis, M. Jain, R. Berman, P. Liu, M. Huang, H. Wu, E. Friedman, G. Wicks, and D. Moore, “ An intra-chip free-space optical interconnect,” Conf Proc Int Symp C, 94–105 (2010).
22. B. Ciftcioglu, R. Berman, S. Wang, J. Hu, I. Savidis, M. Jain, D. Moore, M. Huang, E. G. Friedman, G. Wicks, and H. Wu, “3-D integrated heterogeneous intra-chip free-space optical interconnect,” Opt. Express 20(4), 4331–4345 (2012). [CrossRef] [PubMed]
23. X. Zhao, D. Parekh, E. K. Lau, H.-K. Sung, M. C. Wu, W. Hofmann, M. C. Amann, and C. J. Chang-Hasnain, “Novel cascaded injection-locked 1.55-mum VCSELs with 66 GHz modulation bandwidth,” Opt. Express 15(22), 14810–14816 (2007). [CrossRef] [PubMed]
24. P. Westbergh, J. Gustavsson, B. Kögel, Å. Haglund, A. Larsson, A. Mutig, A. Nadtochiy, D. Bimberg, and A. Joel, “40 Gbit/s error-free operation of oxide-confined 850 nm VCSEL,” Electron. Lett. 46(14), 1014–1016 (2010). [CrossRef]
25. P. Westbergh, R. Safaisini, E. Haglund, B. Kögel, J. S. Gustavsson, A. Larsson, N. Geen, R. Lawrence, and A. Joel, “High-speed 850 nm VCSELs with 28GHz modulation bandwidth operating error-free up to 44Gbit/s,” Electron. Lett. 48(18), 1145–1147 (2012).
26. N. Al-Ababneh, “Crosstalk reduction in free space optical interconnects systems using microlenses with Gaussian transmittance,” Opt. Commun. 318, 79–82 (2014). [CrossRef]
27. C. Favi and E. Charbon, “Techniques for fully integrated intra-/inter-chip optical communication,” in Proc. 45th ACM/IEEE Design Automat. Conf, DAC, Jun. 2008, pp. 343–344 (2008). [CrossRef]
28. R. Rvf, S. Randel, A. H. Gnauck, C. Bolle, R. Essiambre, P. Winzer, D. W. Peckham, A. McCurdy, and R. Lingle, “Space-division multiplexing over 10km of three-mode fiber using coherent 6×6 MIMO processing,” in Optical Fiber Communication Conference, OSA Technical Digest CD, (Optical Society of America, 2011), paper PDPB10 (2011).
29. S. Randel, R. Ryf, A. Sierra, P. J. Winzer, A. H. Gnauck, C. A. Bolle, R.-J. Essiambre, D. W. Peckham, A. McCurdy, and R. Lingle Jr., “6×56-Gb/s mode-division multiplexed transmission over 33-km few-mode fiber enabled by 6×6 MIMO equalization,” Opt. Express 19(17), 16697–16707 (2011). [CrossRef] [PubMed]
30. R. Ryf, S. Randel, A. H. Gnauck, C. Bolle, A. Sierra, S. Mumtaz, M. Esmaeelpour, E. C. Burrows, R.-J. Essiambre, P. J. Winzer, D. W. Peckham, A. H. McCurdy, and R. Lingle, “Mode-division multiplexing over 96km of few-mode fiber using coherent 6×6 MIMO processing,” J. Lightwave Technol. 30(4), 521–531 (2012). [CrossRef]
31. V. Sleiffer, Y. Jung, V. Veljanovski, R. van Uden, M. Kuschnerov, Q. Kang, L. Gruner-Nielsen, Y. Sun, D. Richardson, S. Alam, F. Poletti, J. Sahu, A. Dhar, H. Chen, B. Inan, T. Koonen, B. Corbett, R. Winfield, A. Ellis, and H. De Waardt, “73.7 Tb/s (96×3×256-Gb/s) mode-division-multiplxed DP-16QAM transmission with inline MM-EDFA,” in European Conference and Exhibition on Optical Communication, OSA Technical Digest (online) (Optical Society of America, 2012), paper Th.3.C.4 (2012).
32. L. Allen, M. W. Beijersbergen, R. J. C. Spreeuw, and J. P. Woerdman, “Orbital angular momentum of light and the transformation of Laguerre-Gaussian laser modes,” Phys. Rev. A 45(11), 8185–8189 (1992). [CrossRef] [PubMed]
33. R. A. Beth, “Mechanical detection and measurement of the angular momentum of light,” Phys. Rev. 50(2), 115–125 (1936). [CrossRef]
34. M. E. J. Friese, T. A. Nieminen, N. R. Heckenberg, and H. Rubinsztein-Dunlop, “Optical alignment and spinning of laser-trapped microscopic particles,” Nature 394(6691), 348–350 (1998). [CrossRef]
35. E. Santamato, “Photon orbital angular momentum: problems and perspectives,” Fortschr. Phys. 52(11–12), 1141–1153 (2004). [CrossRef]
36. G. Molina-Terriza, J. P. Torres, and L. Torner, “Twisted photons,” Nat. Phys. 3(5), 305–310 (2007). [CrossRef]
37. S. Franke-Arnold, L. Allen, and M. Padgett, “Advances in optical angular momentum,” Laser Photon. 2(4), 299–313 (2008). [CrossRef]
38. Y. Awaji, N. Wada, Y. Toda, and T. Hayashi, “World first mode/spatial division multiplexing in multi-core fiber using Laguerre–Gaussian mode,” in Proc. ECOC 2011, Geneva, Switzerland, Sep., pp. 1–3, paper We.10 (2011).
39. Y. Awaji, N. Wada, and Y. Toda, “Observation of orbital angular momentum spectrum in propagating mode through seven-core fibers,” in Conference on Lasers and Electro-Optics 2012, OSA Technical Digest (Optical Society of America, 2012), paper JTu2K.3 (2012). [CrossRef]
40. W. Nasalski, “Vortex and anti-vortex compositions of exact elegant Laguerre-Gaussian vector beams,” Appl. Phys. B 115(2), 155–159 (2014). [CrossRef]
41. G. Xie, Y. Ren, H. Huang, N. Ahmed, L. Li, Y. Yan, M. Lavery, M. Padgett, and A. Willner, “Experimental analysis of multiplexing/demultiplexing laguerre Gaussian beams with different radial index,” in Forntiers in Optics 2014, OSA Technical Digest (online) (Optical Society of America, 2014), paper FTh4B.6 (2014). [CrossRef]
42. P. Gregg, P. Kristensen, S. E. Golowich, J. Ø. Olsen, P. Steinvurzel, and S. Ramachandran, “Stable transmission of 12 OAM states in air-core fiber,” in CLEO 2013, OSA Technical Digest (online) (Optical Society of America, 2013), paper CTu2K.2 (2013). [CrossRef]
44. C. Brunet, P. Vaity, Y. Messaddeq, S. LaRochelle, and L. A. Rusch, “Design, fabrication and validation of an OAM fiber supporting 36 states,” Opt. Express 22(21), 26117–26127 (2014). [CrossRef] [PubMed]
45. N. Ahmed, H. Huang, Y. Ren, Y. Yan, G. Xie, and A. E. Willner, “Reconfigurable 2x2 Orbital-Angular-Momentum-Based Optical Switching of 50-Gbaud QPSK Channels”, paper Th.1.C.3. ECOC 2013.
46. B. Guan, R. P. Scott, C. Qin, N. K. Fontaine, T. Su, C. Ferrari, M. Cappuzzo, F. Klemens, B. Keller, M. Earnshaw, and S. J. B. Yoo, “Free-space coherent optical communication with orbital angular, momentum multiplexing/demultiplexing using a hybrid 3D photonic integrated circuit,” Opt. Express 22(1), 145–156 (2014). [CrossRef] [PubMed]
47. S. J. B. Yoo, “Wavelength conversion technologies for WDM Network Applications,” J. Lightwave Technol. 14(6), 955–966 (1996). [CrossRef]
48. J. Wang, J. Yang, I. M. Fazal, N. Ahmed, Y. Yan, H. Huang, Y. Ren, Y. Yue, S. Dolinar, M. Tur, and A. E. Willner, “Terabit free-space data transmission employing orbital angular momentum multiplexing,” Nat. Photonics 6(7), 488–496 (2012). [CrossRef]
49. H. Huang, G. Xie, Y. Yan, N. Ahmed, Y. Ren, Y. Yue, D. Rogawski, M. Tur, B. Erkmen, K. Birnbaum, S. Dolinar, M. Lavery, M. Padgett, and A. E. Willner, “100 Tbit/s free-space data link using orbital angular momentum mode division multiplexing combined with wavelength division multiplexing, ” in Optical Fiber Communication Conference/National Fiber Optical Engineers Conference 2013, OSA Technical Digest (online) (Optical Society of America, 2013), paper OTh4G.5 (2013).
50. J. Wang, S. Li, M. Luo, J. Liu, L. Zhu, C. Li, D. Xie, Q. Yang, S. Yu, J. Sun, X. Zhang, W. Shieh, and A. Willner, “N-Dimentional multiplexing link with 1.036-Pbit/s transmission capacity and 112.6-bit/s/Hz spectral efficiency using OFDM-8QAM signals over 368 WDM pol-muxed 26 OAM modes,” ECOC’2014, paper Mo.4.5.1 (2014) [CrossRef]
51. J. Wang, S. Li, C. Li, L. Zhu, C. Gui, D. Xie, Y. Qiu, and S. Yu, “Ultra-high 230-bit/s/Hz spectral efficiency using OFDM/OQAM 64-QAM signals over pol-muxed 22 orbit angular momentum (OAM) modes,” in Optical Fiber Communication Conference, OSA Technical Digest (online) (Optical Society of America, 2014), paper W1H.4 (2014). [CrossRef]
53. N. Bozinovic, Y. Yue, Y. Ren, M. Tur, P. Kristensen, H. Huang, A. E. Willner, and S. Ramachandran, “Terabit-Scale Orbital Angular Momentum Mode Division Multiplexing in Fibers,” Science 340(6140), 1545–1548 (2013). [CrossRef] [PubMed]
54. C. Brunet, B. Ung, Y. Messaddeq, S. LaRochelle, E. Bernier, and L. Rusch, “ Design of an optical fiber supporting 16 OAM modes,” in Optical Fiber Communication Conference, OSA Technical Digest (online) (Optical Society of America, 2014), paper Th1A.24 (2014).
55. X. Cai, J. Wang, M. J. Strain, B. Johnson-Morris, J. Zhu, M. Sorel, J. L. O’Brien, M. G. Thompson, and S. Yu, “Integrated compact optical vortex beam emitters,” Science 338(6105), 363–366 (2012). [CrossRef] [PubMed]
56. N. K. Fontaine, C. R. Doerr, and L. Buhl, “Efficient multiplexing and demultiplexing of free-space orbital angular momentum using photonic integrated circuits,” in Optical Fiber Communication Conference, OSA Technical Digest (Optical Society of America, 2012), paper OTu11.2 (2012). [CrossRef]
57. M. J. Strain, X. Cai, J. Wang, J. Zhu, D. B. Phillips, L. Chen, M. Lopez-Garcia, J. L. O’Brien, M. G. Thompson, M. Sorel, and S. Yu, “Fast electrical switching of orbital angular momentum modes using ultra-compact integrated vortex emitters,” Nat Commun 5, 4856 (2014). [CrossRef] [PubMed]
59. Ting Lei, Meng Zhang, Yuru Li, Ping Jia, Gordon Ning Liu, Xiaogeng Xu, Zhaohui Li, Changjun Min, Jiao Lin, Changyuan Yu, Hanben Niu and Xiaocong Yuan. Massive individual orbital angular momentum channels for multiplexing enabled by dammann gratings. Light: Science & Applications accepted article preview 18 December 2014; e257; doi: [CrossRef] .
60. Y. Ren, G. Xie, H. Huang, C. Bao, Y. Yan, N. Ahmed, M. P. J. Lavery, B. Erkmen, S. J. Dolinar, M. Tur, M. Neifeld, M. J. Padgett, R. Boyd, J. H. Shapiro, and A. E. Willner, “Simultaneous turbulence compensation of multiple orbital-angular-momentum 100-Gb/s data channels using a Gaussian probe beam for wavefront sensing,” in Conference Optical Communication, London, Sep. 2013.