I. Introduction
From the viewpoint of effective utilization of wavelength band as well as the increase in aggregate capacity, the use of 40 Gb/s channel bit rate is desirable for long-haul wavelength-division multiplexing (WDM) optical network applications. Recently, 100 GHz-spaced 32×42.7 Gb/s 9000 km transmission experiment was reported using carrier-suppressed return-to-zero on-off-keying (CSRZ-OOK) signal [1] as the first demonstration of an encouraging future possibility of 40 Gb/s-based transpacific systems. To make such systems a commercial reality, however, we still have to work on several formidable issues. Higher spectral efficiencies (i.e., smaller WDM channel spacing) are the key to the cost-effectiveness and high upgradeability for the future high-capacity optical transmission systems, because we can relax the requirement for repeater bandwidth and broadband fiber dispersion and loss management. However, the spectral efficiency demonstrated in [1] was only 40%, and for the transoceanic transmission the highest spectral efficiency of 57% has been demonstrated only over 6100 km by using 32×42.7 Gb/s pre-filtered CSRZ-OOK signals [2]. Therefore, we have to explore the possibility to extend the transmissible distance with a similar degree of high spectral efficiency. Another issue is the complexity of the transmission line configuration; in [1], the repeater span was intricately configured with a 4-segment dispersion-flattened fiber (two concatenating sets of 19 m2 negative dispersion fiber and 195 m2 positive dispersion fiber) and a Raman-amplifier with 4-wavelength pump lasers. Simpler and more cost-effective wet plant configurations are desired for practical applications. Furthermore, the assessment of the impact of polarization mode dispersion (PMD) in the transmission line is one of the major concerns for 40 Gb/s-based transmission systems, but none of the 40 Gb/s-based transoceanic transmission experiments reported so far have evaluated the expected bit-error-rate (BER) performance averaged over all the possible signal states of polarization [3], [4].