JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 28, NO. 12, JUNE 15, 2010 1811 Comparison of Orthogonal Frequency-Division Multiplexing and On-Off Keying in Amplified Direct-Detection Single-Mode Fiber Systems Daniel J. F. Barros and Joseph M. Kahn, Fellow, IEEE Abstract—We discuss the use of orthogonal frequency-division multiplexing (OFDM) for combating group-velocity dispersion (GVD) effects in amplified direct-detection (DD) systems using single-mode fiber. We review known OFDM techniques, in- cluding asymmetrically clipped optical OFDM (ACO-OFDM), DC-clipped OFDM (DC-OFDM) and single-sideband OFDM (SSB-OFDM), and derive a linearized channel model for each technique. We present an iterative procedure to achieve optimum power allocation for each OFDM technique, since there is no closed-form solution for amplified DD systems. For each tech- nique, we minimize the optical power required to transmit at a given bit rate and normalized GVD by iteratively adjusting the bias and optimizing the power allocation among the subcarriers. We verify that SSB-OFDM has the best optical power efficiency among the different OFDM techniques. We compare these OFDM techniques to on-off keying (OOK) with maximum-likelihood se- quence detection (MLSD) and show that SSB-OFDM can achieve the same optical power efficiency as OOK with MLSD, but at the cost of requiring twice the electrical bandwidth and also a complex quadrature modulator. We compare the computational complexity of the different techniques and show that SSB-OFDM requires fewer operations per bit than OOK with MLSD. Index Terms—Communications system performance, di- rect-detection, group-velocity dispersion, intensity modulation, maximum-likelihood sequence detection, maximum-likelihood sequence estimation, multi-carrier optical systems, orthogonal frequency-division multiplexing. I. INTRODUCTION R ECEIVER-BASED electronic signal processing in op- tical communication systems has been the subject of many recent studies. In coherent systems, linear equalizers have been shown to fully compensate linear fiber impairments in single-mode fiber (SMF), such as group-velocity dispersion (GVD) and polarization-mode dispersion (PMD) [1]. On the other hand, in systems using direct-detection (DD), linear equalizers offer little performance improvement, because the nonlinear photodetection process destroys information on the Manuscript received January 29, 2010; revised April 15, 2010; accepted April 18, 2010. Date of publication April 29, 2010; date of current version June 02, 2010. This work was supported by the Portuguese Foundation for Science and Technology scholarship SFRH/BD/22547/2005 and by a Stanford Graduate Fellowship. The authors are with the Department of Electrical Engineering, Stanford University, Stanford, CA 94305-9515 USA (e-mail: djbarros@stanford.edu; jmk@ee.stanford.edu). Digital Object Identifier 10.1109/JLT.2010.2048999 phase of the received electric field. Recently, maximum-likeli- hood sequence detection (MLSD) was shown to be effective in mitigating GVD and PMD impairments in DD links [2], [3]. Although the computational complexity of MLSD increases exponentially with the channel memory, using low-complexity branch metrics, near-optimal performance can be achieved with manageable computational complexity, provided the effective channel memory does not exceed a few symbol intervals [4]. An alternate approach to combating fiber impairments in am- plified DD systems is to use multicarrier modulation, such as or- thogonal frequency-division multiplexing (OFDM). There are three major approaches for combining OFDM with DD. The first two techniques are based on intensity modulation (IM), and so require some means to make the OFDM signal nonnegative. The first technique adds a DC bias to reduce the negative signal excursions and then clips the remainder of the negative excur- sions. This method is called DC-OFDM. The second technique clips the entire negative excursion of the waveform, avoiding the need for a DC bias [5], [6]. Clipping noise is avoided by appropriate choice of the subcarrier frequencies. This technique is called asymmetrically clipped optical OFDM (ACO-OFDM). The third technique is based on single-sideband modulation of the complex-valued optical electric field by an OFDM signal. A DC bias (carrier component) is added, leaving a guard band between the carrier and the OFDM signal in order to avoid in- termodulation products caused by photodetection [7], [8]. This method is called single-sideband OFDM (SSB-OFDM). There have been several studies of the different OFDM tech- niques (e.g., [6]), but there has been no comparison of power efficiencies among the various direct-detection OFDM methods in SMF, and to conventional baseband methods, such as on-off keying (OOK). Furthermore, in previous work, the DC bias and the powers of the subcarriers were not jointly optimized based on the channel response and the nonlinear beat noises, since there is no closed-form solution for this optimization. We present an iterative procedure based on known bit-loading algorithms with a new modification, the bias ratio (BR), in order to obtain the optimum power allocation. We compare the performance of the three OFDM techniques using optimized power allocations to the performance of OOK with MLSD. This paper is organized as follows. In Section II, we review methods for power and bit allocation for multicarrier systems and describe the optimal water-filling solution. We present our system model in Section III. In Section IV, we review the different OFDM formats and derive equivalent linear channel models for each one, assuming only GVD is present in the 0733-8724/$26.00 © 2010 IEEE Authorized licensed use limited to: Stanford University. Downloaded on June 29,2010 at 22:14:27 UTC from IEEE Xplore. Restrictions apply.