I. Introduction
The rapid growth in the demand for higher system capacity to support a range of new services opens opportunities for new multiplexing techniques that maximize the benefit of the available bandwidth within the deployed optical fiber infrastructure. Optical code-division multiple access (OCDMA) is one promising candidate capable of supporting bursty and variable data-rate traffic in an asynchronous manner and is well suited to fulfill the needs of local area and access networks [1]–[3]. Among the different types of reported OCDMA systems, the two-dimensional (2-D) time-spreading wavelength-hopping (TW) scheme offers desirable advantages, particularly in terms of access to a large number of users. In TW schemes, each data pulse is spread into many chips constituting the code length; some/all chips are then transmitted at different wavelengths according to a certain algorithm. In general, several parameters impact system performance, particularly the code properties. A comparison between several symmetric and asymmetric 2-D TW codes has been presented [4], highlighting the superiority of some symmetric codes and introducing a desired feature, referred to as “flexibility.” The term flexibility reflects the degree of independence of time chips, or equivalently the code length, on the number of wavelengths, or code weight. If these two properties are independent, the code then offers a high degree of flexibility. A lack of code flexibility reflects dependence between time and wavelengths.