In future optical networks that employ wavelength division multiplexing (WDM), the use of optical switching technologies on a burst or packet level, combined with advanced modulation formats would achieve greater spectral efficiency and utilize the existing bandwidth more efficiently. All-optical wavelength converters are expected to be one of the key components in these broadband networks. They can be used at the network nodes to avoid contention and to dynamically allocate wavelengths to ensure optimum use of fiber bandwidth. In this work, a reconfigurable wavelength converter comprising of a Semiconductor Optical Amplifier (SOA) as the nonlinear element and a fast-switching sampled grating distributed Bragg reflector (SG-DBR) tunable laser as one of the pumps is developed. The wavelength conversion of 12.5-Gbaud quadrature phase shift keying (QPSK) and Pol-Mul QPSK signals with switching time of tens of nanoseconds is experimentally achieved. Although the tunable DBR lasers can achieve ns tuning time, they present relatively large phase noise. The phase noise transfer from the pump to the converted signal can have a deleterious effect on signal quality and cause a performance penalty with phase modulated signals. To overcome the phase noise transfer issue, a wavelength converter using tunable dual-correlated pumps provided by the combination of a single-section quantum dash passively mode-locked laser (QD-PMLL) and a programmable tunable optical filter is designed and the wavelength conversion of QPSK and 16-quadrature amplitude modulation (16-QAM) signals at 12.5 GBaud is experimentally investigated. Nonlinear distortion of the wavelength converted signal caused by gain saturation effects in the SOA can significantly degrade the signal quality and cause difficulties for the practical wavelength conversion of sig nal data with advanced modulation formats. In this work, the machine learning clustering based nonlinearity compensation method is proposed to improve the tolerance to nonlinear distortion in an SOA based wavelength conversion system with 16 QAM and 64 QAM signals.