broadband access and in-house networks – extending the ... - CiteSeerX

SYMPOSIA

AUTHORS

BROADBAND ACCESS AND IN-HOUSE NETWORKS – EXTENDING THE CAPABILITIES OF MULTIMODE FIBRE NETWORKS Ton Koonen, Henrie van den Boom, Giok-Djan Khoe COBRA Institute, Eindhoven University of Technology, The Netherlands, [email protected] Abstract: A number of optical signal processing techniques are reviewed which allow to realise high-capacity multi-service multimode fibre networks in the user environment. Introduction The residential user has a growing need for multiple groups of broadband services, such as (increasingly personalised) video services, fast internet, highquality audio, etc. Moreover, due to the liberalisation of the telecommunication market there are several operators who can offer (a number of) these services. The legal requirement for “access network unbundling” implies that in an access network a number of operators should be enabled to provide multiple groups of services to the residential users in an efficient and independent way. There is a widely-spread consensus that FTTH is the most powerful and future-proof access network architecture for providing broadband services to residential users. In the FTTH system concepts deployed up to now, single-mode optical fibre is used, which has a tremendous bandwidth and thus a huge transport capacity for many services; e.g., the ITU G.983.x ATM-PON system concepts standardised in the worldwide Full Service Access Network, FSAN, initiative of many world-leading operators and companies [1]. Research is ongoing to further extend the capabilities of shared single-mode fibre access networks, a.o. by dynamic capacity allocation by means of wavelength routing or –selection [2]. The installation of single-mode fibre, however, requires great care, delicate high-precision equipment, and highly-skilled personnel. This makes installation quite costly, which seriously hampers the large-scale introduction of FTTH. Reducing system costs with multimode fibre Multimode optical fibre provides a cost-reducing alternative for the commonly used single-mode fibre. Multimode fibre is easier to install than single-mode fibre with its tiny core: due to larger core diameter of multimode fibre, the coupling of light from a light source into the fibre and the (fixed or demountable) splicing of fibres together is easier, allowing less delicate splicing equipment and less requirements on the skills of the installation personnel. In particular in the access network, this may yield a considerable reduction of installation costs. The bandwidth-timeslength product of single-mode fibre is significantly higher than that of multimode fibre. As in the access

network the fibre link lengths are less than 10 km, however, the bandwidth of presently commercially available silica multimode fibres is quite sufficient. Also inside the customer’s building, there is a growing need for convergence of the multitude of communication networks. Presently, twisted copper pair cables are used for voice telephony, cat-5 UTP cables for high-speed data, coaxial cables for CATV and FM radio signals distribution, wireless LAN for high-speed data, FireWire for high-speed short-range signals, and also power lines for control signals and lower-speed data. These different networks are each dedicated and optimised for a particular set of services. Also no cooperation between the networks exists. It is therefore not easy to upgrade services, to introduce new ones, nor to create links between services (e.g., between video and data). By establishing a common broadband in-house network infrastructure, in which a variety of services can be integrated, however, these difficulties can be surmounted. Optical fibre is the prime choice as the transport medium in such a network. But even more than in access networks, reduced installation costs are of paramount importance, and thus the application of multimode fibre is attractive. Moreover, multimode fibre is already used extensively for inoffice data links up to Gigabit Ethernet services, benefiting from low-cost multimode fibre-compatible transceivers. It would be quite advantageous to introduce multiplexing methods for combining more service groups on the multimode fibre infrastructure. Using polymer optical fibre (POF) in stead of silica multimode fibre may provide additional reduction of installation efforts, due to its ductility and larger core diameter (beyond 100 µm). The losses of POF (some 10 dB/km) are still quite a bit higher than those of silica fibre, but are coming down steadily due to ongoing improvements in the production processes of this still young technology. Furthermore, link lengths in-house are short (less than 1 km), and thus the loss per unit length is of less importance. Although multimode fibre is easier to install, its bandwidth is still significantly lower than that of single-mode fibre. Graded-index POF may even have bandwidth-times-length products of only some 1 to 5 GHz⋅km. This limited bandwidth hampers the

desired integration of multiple broadband services into a common multimode fibre access or in-house network. Therefore techniques have to be developed to extend the capabilities of multimode fibre networks. This paper reviews three novel techniques for this: • mode group diversity multiplexing deploying in parallel subsets of the huge number of guided modes, thus enabling service integration by establishing several independent communication channels in parallel, and • optical frequency multiplying, which enables microwave radio signals to be carried along multimode fibre, and thus can simplify considerably fibre-wireless networks such as wireless LANs, and • subcarrier multiplexing using the higher-order transmission lobes of multimode fibre, which allows to adaptively position the transport of services in adequate spectral regions of the fibre’s transfer characteristics. Both silica multimode fibre and the upcoming polymer optical multimode fibre will be considered. Mode group diversity multiplexing In multimode fibre, the propagation time differences between the many guided modes can be significant, causing a lot of dispersion; this modal dispersion is the major cause of the fibre’s limited bandwidth. The disadvantage of having many modes can, however, also be turned into an advantage by dividing the large group of modes into smaller groups, and by then using each of these groups as a separate transmission channel. This novel method, which we have termed mode group diversity multiplexing, is illustrated in Fig. 1. electrical signal processing

. .

N lasers

. .

electrical signal processing

M detectors

. .

multimode fibre

. .

feedback detector

N input data streams

Fig. 1

laser

N recovered data streams

mode group 1

c11

transmitter 1

detector 1

c12 c21 transmitter 2

detector 2

c22 mode group 2

Fig. 2

Modelling the system matrix (for N=M=2)

transmission

At the other end of the fibre link, each of the M optical receivers receives a mixture of the signals arriving via the mode groups; see Fig. 2. This mixture is the result of the statistically varying mode mixing process in the multimode fibre network. After the receivers, an electrical signal processing circuit can reverse the mixing process, thus yielding the separate input data streams again. The signal processing circuit needs to know the mode mixing coefficients for that; these coefficients can be deduced by sending training sequences (similarly as done in wireless LAN multiple-input multiple-output, MIMO, systems). By repeating this training period when detecting flaws in the separated data streams, induced by changes in the mode mixing process, the coefficients can be updated timely. Detection of these flaws is enabled by signal processing at the transmitter site, adding redundancy by line coding) An analysis has been made how mode groups can be separately excited in polymer optical fibres, and thus how independent communication channels could be established. Fig. 3 shows the near-field patterns (NFPs) measured at the output of a 100 metres long multimode PMMA graded-index POF, under different mode launching conditions. Clearly, the NFP obtained when exciting low-order modes is complementary to the NFP when exciting high-order modes, so these two excitation conditions obviously can constitute two separate transmission channels. In general, it can be shown that different excitation conditions yield different circular NFPs at the multimode POF output.

Mode group diversity multiplexing

The concept is based on using a number N of independent optical transmitters at one end of the system, and M receivers at the other end. Each transmitter launches a data signal into a different group of modes. By exciting a subset of modes out of the total set of guided modes, the dispersion within such a mode group can be considerably less than that within the total set.

a)

b)

c)

Fig. 3 Near-Field patterns at the output of a multimode POF a) with excitation of all modes b) with excitation of low-order modes only c) with excitation of high-order modes only The knowledge of the system transmission matrix coefficients at the downstream receiving site may

also be used for selective mode group excitation in the upstream direction; this enables bi-directional system operation. The system functionality obtainable with mode group diversity multiplexing is similar to wavelength division multiplexing. Provided low-cost integrated electronic circuits can be used for the signal processing, the mode group diversity multiplexing method may economically outperform the wavelength multiplexing method, while offering similar performance. Thus widely differing services such as Gigabit Ethernet and analog video signals could be transported in a single multimode (polymer) optical fibre in-house network. Optical frequency multiplying Radio-over-fibre techniques enable considerable simplification and thus cost-reduction of the antenna sites, e.g. in wireless LANs evolving to higher carrier frequencies for delivering higher capacities. The signal processing functions, usually done in the antenna stations, can then be consolidated in the headend station, and also more comprehensive wireless system functions (such as macro-diversity, and MIMO) can be implemented more efficiently. Radio-over-fibre techniques are usually based on single-mode fibre (e.g. by deploying heterodyning between closely spaced optical carriers to generate a microwave carrier). For extensive high-capacity (inbuilding) wireless networks, the installation costs of single-mode fibre may be a major barrier; multimode fibre with its ease of handling is more attractive. Moreover, multimode fibre is already being used for in-building high-speed data communication networks, benefiting from low-cost transceivers. To realise transport of microwave radio signals over multimode fibre, however, methods to surmount its limited bandwidth have to be devised. Heterodyning optical carriers is not feasible in multimode fibre, due to the lack of phase correlation between the various guided modes. Headend station + data

fsw tunable LD

isw

+ϕ

λ0

-ϕ

Radio Access Point fmm = 2N ∙ fsw

periodic filter

GIPOF link

PD BPF

i(t)

f l1

f l2

transmission

- data ∆λFSR 0

1

2

.

N

filter λ

λ0

Fig. 4 Carrying microwave signals over a multimode fibre network by optical frequency multiplying Therefore, a method termed optical frequency multiplying has been devised, of which the basic

principle is illustrated in Fig. 4 [3]. At the headend station, a wavelength-tunable laser diode is used, of which the wavelength is periodically swept over a range ∆λopt with a sweep frequency fsw, while keeping the output power constant. The data is impressed on this wavelength-swept optical signal by means of chirp-free intensity modulation, by means of e.g. a differentially driven Mach Zehnder Interferometer. After having passed through the multimode POF network, the signal impinges on a periodic optical bandpass filter, e.g. a Fabry Perot filter. In sweeping across N transmission peaks of this filter (back and forth during one wavelength sweep cycle), light intensity bursts arrive on the photodiode with a frequency 2N⋅fsw . Thus, the output signal i(t) of the photodiode contains a microwave frequency component at 2N⋅fsw , and higher harmonics of which the strength depends on the bandpass characteristics of the periodic filter. After bandpass filtering in order to select the desired harmonic, and some amplification, the microwave carrier is radiated by the antenna. Note that only the optical sweep frequency is limited by the bandwidth of the POF network, and that the microwave carrier frequency can exceed this bandwidth by far due to the optical frequency multiplication mechanism. Simulations as well as experiments have shown that very pure microwave carriers can be generated by this optical frequency multiplying mechanism [4]. The data intensity modulation impressed at the headend station is not affected by this optical frequency multiplying process, and is carried on the envelope of the microwave carrier. Simulations have shown the feasibility of on-off ASK data modulation up to 225 Mbit/s on carrier frequencies of 5.4 GHz, and equally well on the second harmonic at 10.8 GHz. Instead of sweeping the optical frequency by direct laser modulation, it can also be swept by means of an external phase modulator. More comprehensive signal formats can be handled by putting these first on a subcarrier which is subsequently intensity-modulated on the frequencyswept carrier [3]. The feasibility of this approach for e.g. a 16-QAM signal carrying 56 Mbit/s on a subcarrier of 225 MHz and a carrier of 5.4 GHz is proven by simulations yielding the recovered inphase and quadrature-phase signals Bi-directional system operation is possible by deploying the generated microwave carrier at the antenna station for down-converting the upstream microwave signal received by the antenna, and transmitting that in baseband by a laser diode in a wavelength band different from the downstream one using coarse wavelength multiplexing.

Subcarrier multiplexing using higher-order transmission lobes of multimode fibre

however, put higher requirements on the linearity of the transmission system.

The transmission characteristics of multimode fibre are often assumed to follow a Gaussian low-pass frequency characteristic. However, when chromatic dispersion is small and coupling between the modes can be neglected (this is the case in today’s highquality multimode fibre), the impulse response of a length of multimode fibre can be considered to be a sequence of dirac impulses which correspond to the arrival times of the individual fibre modes. Therefore, the frequency characteristics of the fibre should show significant high-frequency components. This is indeed confirmed by measurements, as illustrated in Fig. 5.

Modal noise is a potentially serious multiplicative noise problem in multimode fibre systems [7]. High frequency intensity modulation of laser diodes may decrease the laser coherence time, which combats modal noise. In this system, the laser is modulated with high-frequency subcarriers for information transmission anyhow, so this may show helpful as well for reducing the modal noise. The position of the higher-frequency lobes depends on the fibre link length, and on the exact fibre characteristics, which may vary due to external circumstances such as induced stress by bending or environmental temperature variations. The system has to adapt to those variations, e.g. by monitoring the fibre link transfer curve by injecting some weak pilot tones, and allocating the subcarriers accordingly. Conclusions

Fig. 5 Transmission of 1 km of 62.5 µm core multimode fibre, measured with an 850 nm VCSEL (from [5]) These higher-frequency transmission lobes can be deployed by using subcarrier multiplexing techniques, which allow to transport information signals by modulating them on specific carrier frequencies. The modulated carriers can be positioned in such a way that they will optimally fit into the higher-frequency transmission lobes of the multimode fibre link. In [6], this approach is followed using two subcarriers at 1 and 3 GHz, respectively, each transporting 625 Mbit/s in BPSK (binary phase shift keying) format; thus 1.25 Gbit/s over 500 metres of multimode fibre is achieved. Each of the multiple subcarriers may constitute an independent transmission channel on the multimode fibre, which allows to integrate various groups of services independently in a single FTTH infrastructure. When extending the reach of the system (e.g., up to typical access network maximum link lengths of 5 km), it is needed to compress the width of the spectrum of the modulated subcarrier signal, as the higher-frequency transmission lobes will become narrower. Therefore, more information-rich formats such as QPSK (quadrature phase shift keying, 2 bits/symbol) and 16-level QAM (16 points of quadrature amplitude modulation, 4 bits/symbol) can be explored. These advanced modulation formats,

Multimode fibre enables low-cost installation of broadband networks in the access and in-building environment, as already shown for short-range data LAN communications. In particular Polymer Optical Fibre (POF) is attractive for short-range in-building applications, due to its easy handling. Notwithstanding its restricted bandwidth, a single multimode fibre network may carry a multitude of broadband services, by deploying mode group diversity multiplexing, or adaptive subcarrier multiplexing using the higher-order fibre’s transmission lobes. These services may e.g. encompass wired Gigabit Ethernet, and wireless high-capacity wireless LAN by deploying optical frequency multiplying. Thus, easy-to-install multimode fibre networks for access and in-house can be realised in which wirebound and wireless services are efficiently integrated. References [1] [2]

[3]

[4]

[5]

[6] [7]

David Faulkner, Yoichi Maeda, Proc. of ECOC 2003, Rimini, Sep. 21-25, 2003 Ton Koonen, Henrie van den Boom, Idelfonso Tafur Monroy, Jeroen Wellen, Rob Smets, Giok-Djan Khoe, Proc. of ECOC 2002, Copenhagen, Sep. 8-12, 2002, 2 p. Ton Koonen, Anthony Ng’oma, Peter Smulders, Henrie van den Boom, Idelfonso Tafur Monroy, Giok-Djan Khoe, Phot. Netw. Comm., Vol. 5, No. 2, March 2003, pp. 177-187 Anthony Ng’oma, Idelfonso Tafur Monroy, Ton Koonen, Henrie van den Boom, Peter Smulders, Giok-Djan Khoe, Proc. of ECOC 2003, Rimini, Sep. 21-25, 2003 L. Raddatz, D. Hardacre, I.H. White, R.V. Penty, D.G. Cunningham, M.R.T. Tan, S.-Y. Wang, Electr. Lett., Vol. 34, No. 7, Apr. 2, 1998, pp. 686-688 L. Raddatz and I.H. White, IEEE Phot. Techn. Lett., Vol. 11, No. 2, Feb. 1999, pp. 266-268 A.M.J. Koonen, IEEE J. on Selected Areas in Comm., Vol. SAC-4, No. 9, Dec. 1986, pp. 1515-1522