I. Introduction
Optical code-division multiple access (OCDMA) has always been of interest because of its inherent ability to support asynchronous bursty communications. As such, it was initially pursued for local area [1] and then for access network applications [2]–[5]. More recently, it is the subject of research for generalized multiprotocol label switching (GMPLS) versions of emerging networks [6]–[8]. Generally, OCDMA can be divided into direct-sequence pseudoorthogonal (PSO) pulse sequences [9], which are frequently referred to as optical orthogonal codes (OOCs) [10], [11]; direct sequence bipolar codes [6]; frequency or phase encoding codes [12], and so on; and two-dimensional (2-D) and higher dimensional codes. The 2-D codes, in turn, can be divided into time-spreading frequency-hopping types [13], [14]; depth-first search types [15], [16]; 2-D prime code types [17], [18]; sonar codes [19]; and folded optimum Golomb ruler types [20]–[26]. Examples of three-dimensional (3-D) codes are given in [27], using wavelength, space, and time for the codes and using wavelength, polarization, and time for the codes [28].