Sie sind auf Seite 1von 2

Fiber Capacity Limits: Information Theory meets Optical Communication and Fiber Physics

Ren-Jean Essiambre
1

Bell Labs, Alcatel-Lucent, 791 Holmdel-Keyport Road, Holmdel, New Jersey, 07733, USA channel, the capacity (C) per unit of channel bandwidth (B) or the spectral efficiency (SE) is given by,

Abstract Summary Determining a fundamental limit to the rate of transmission of information in optical fibers, or fiber capacity, requires bringing together concepts of information theory, nonlinear dynamics and optical physics. An overview of an approach to estimate fiber capacity is presented. Keywords - optical communication; information theory; fiber nonlinearity; amplified spontaneous emission.

where SNR is the signal-to-noise ratio. The SNR relates to the optical signal-to-noise ratio (OSNR) commonly used in optical communication [2] by,

I. INTRODUCTION Experiments have demonstrated a tremendous growth in the ability of optical fibers to transport information over thousands of kilometers. An astounding one-thousand fold increase in the rate of transmission has been achieved in laboratory experiments over the past two decades. Similar growth has occurred in commercial systems, with a lag of about 5 years relative to record experiments. Such dramatic growth has fueled the widespread use of the Internet. Can it continue unabated? We present here a procedure and the results of a capacity limit estimate analysis for optical fibers. We start from the first principles of Shannons information theory [1] and present the challenges associated with determining a fiber capacity. We focus our attention on the instantaneous Kerr nonlinear effect which is the physical phenomenon limiting signal power, and therefore capacity, in optical fibers. We incorporate a series of advanced technologies in our capacity estimation, such as distributed Raman amplification, arbitrary waveform generation for producing compact-bandwidth signals and advanced modulation formats, coherent detection, advanced digital signal processing based on fiber nonlinear back-propagation and ideal forward error correction. We compare our fiber capacity estimate to current experimental records and consider the impact of the development of novel optical fibers on capacity. II. INFORMATION THEORY Shannon considered the rate at which information can be transported reliably through a noisy medium [1]. In particular, he considered the channel described by Y = X + N, where X is an input state, N represents a source of additive white Gaussian noise (AWGN) and Y is the output state. For the AWGN

where p is the number of polarization states of the signal, RS is the symbol rate, and Bref is a reference bandwidth approximately equal to 12.5 GHz. Shannons theory has been applied to a variety of channels beyond the AWGN channel. The capacity calculations are often based on a discrete representation of a channel given by a set of discrete transition probabilities between M input states of X and the Q output states of Y as represented in Fig. 1.

Figure 1. Transition probabilities describing a channel.

For a given channel represented by a set of transition probabilities, the capacity at a signal power P is given by

where is the capacity at constant average power and I(X;Y) is the mutual information defined as

for discrete input and discrete output signals [1,3]. The set of transition probabilities PY|X(b|a) can be established after nonlinear propagation over fibers (described in Sec. III), digital signal processing at the receiver to compensate signal distortions followed by matched filtering (see [2] for more details).

III. NONLINEAR DYNAMICS Optical fibers are made of fused silica, a material that exhibits the optical Kerr nonlinearity [4]. The presence of the fiber Kerr nonlinearity results in signal distortions after propagation. These distortions are dependent on power and, if uncompensated, result in a loss of information. The stochastic nonlinear Schrdinger equation (see [2] and references within) that governs propagation in fibers is given by

instance, ingenious fiber designs and new materials. This fiber capacity CF can be expressed as where C(E2,J,DL) is the capacity defined in (3) and L the system length. The optimization is over the physically realizable combinations of E2, J and D. Figure 3 shows the impact of a change in nonlinear coefficient J on fiber capacity. The values of D and E2 are identical to Fig. 2 while the distance of transmission has been increased to 1000 km. The capacity of the 64-ring constellation used to compute the nonlinear capacity estimate (in both Figs. 2 and 3) is shown for the AWGN channel, i.e. in the absence of fiber nonlinearity in (5). The nonlinear capacity results of Fig. 3 shows that a lower value of J leads to a higher capacity. Notice that a large reduction (by a factor of ~10 or higher) in nonlinear coefficient is necessary to increase capacity significantly. For comparison, the value of J has been reduced by only 30% in the last two decades.

where E(z,t) represents all the fields present in the fiber, while E2, J and N(z,t) represent the chromatic dispersion, fiber nonlinear coefficient and the AWGN field, respectively. The field N(z,t) incorporates the fiber loss coefficient D and the energy of the photon at the signal wavelength (see [2] for a detailed description). The nonlinear dynamics in (5) can be classified in three forms of nonlinear interactions involving: 1) the signal field only; 2) both signal and noise fields; and 3) the noise field only. For type 1), nonlinear distortions can be compensated exactly by electronic back-propagation if all interacting signal fields are known exactly, if there is no bandwidth limitation and in the absence of chaotic behavior. Back-propagation consists of reversing the direction of propagation in (5), after neglecting the noise source N(z,t). For type 2), partial compensation of nonlinear distortions can be achieved, while it is unknown if some compensation of nonlinear interactions of type 3) are possible. In the capacity results below, back-propagation of the received signal is used. IV. CAPACITY RESULTS Figure 2 shows the fiber capacity estimate for a 500-km system. The fiber loss coefficient D, nonlinear coefficient J, and chromatic dispersion E2 are 0.15 dB/km, 1.27 (W km)-1 and 17 ps/(nm-km), respectively. The four symbols in Fig. 3 represent the latest record SE experiments of 2009 and 2010.

Figure 3. Impact of lower fiber loss coefficients on nonlinear capacity limit.

ACKNOWLEDGMENT This work is a collaboration with G. Kramer, G. Foschini, and P. Winzer. Many helpful discussions with other Bell Laboratories colleagues are acknowledged. REFERENCES
[1] [2] [3] [4] [5] C. E. Shannon, A mathematical theory of communication, The Bell System Technical Journal, vol. 27, pp. 379423 and 623656, 1948. R.-J. Essiambre et al., Capacity limits of optical fiber networks, J. Lightwave Technol., vol. 28, pp. 662701, 2010. R. G. Gallager, Information Theory and Reliable Communication, New York: Wiley, 1968, chap. 2. G. P. Agrawal, Nonlinear Fiber Optics, San Diego, CA: Elsevier, 2006. H. Takahashi et al., DWDM transmission with 7.0-bit/s/Hz spectral efficiency using 865.1-Gbit/s coherent PDM-OFDM signals, Proc. OFC, paper PDPB7, 2009. A. H. Gnauck et al., 10112-Gb/s PDM 16-QAM transmission over 630 km of fiber with 6.2-b/s/Hz spectral efficiency, Proc. OFC, paper PDPB8, 2009. A. Sano et al., 69.1-Tb/s (432171-Gb/s) C- and extended L-band transmission over 240 km using PDM-16-QAM modulation and digital coherent detection, Proc. OFC paper, PDPB7, 2010. X. Zhou et al., 64-Tb/s (640107-Gb/s) PDM-36QAM transmission over 320km using both pre- and post-transmission digital equalization, Proc. OFC, paper PDPB9, 2010.

Figure 2. Nonlinear capacity limit and record capacity experiments.

[6]

V.

IMPACT OF FIBER CHARACTERISTICS

[7]

In contrast to many communication channels, the fiber channel properties can be altered significantly through, for

[8]

Das könnte Ihnen auch gefallen