Combined effectiveness of optical hard-limiting and guard-time in optical CDMA systems

摘要

Optical hard-limiting has been proposed as a means to reduce the deleterious effects of multi-access interference (MAI) in optical CDMA systems. The improvements possible with optical hard-limiting decrease with increased user density, and there appears to be a crossover in the number of simultaneous users where a system without optical hard-limiting has a better bit-error-rate performance than one which incorporates optical hard-limiting. This is important because the crossover is at about ten users. Optical CDMA is attractive because of its ability to support asynchronous, bursty, simultaneous data flows, but it must be able to compete with other multiple access schemes from the viewpoint of user density. In view of the increasing WDM density, this means increasing the number of optical CDMA users well beyond 10. Therefore, it is important to identify other MAI mitigation techniques which may enhance the MAI suppression techniques of optical hard-limiting. In this paper we explore by means of computer simulations the combined effectiveness of optical hard-limiting and guard-time applied to a set of 32 optical orthogonal matrix codes. The guard-time technique applies the code to a fraction of the bit time (the remainder of the bit is a space). The benefits of the guard-time are that of reducing inter-symbol interference (ISI) and distributing the MAI over the increased number of chips (we assume asynchronous transmission). Using techniques from return4o-zero (Rz) measurement (which are appropriate to optical CDMA because of the requirement to detect the auto-correlation peak of the signal marks), we find that the guard-time enhances the Q-factor of the signal and optical hard-limiting further improves this. Thus, jointly, these techniques increase the number of simultaneous, asynchronous users supported by optical CDMA. The price of the guard-time is the reduced spectral efficiency of the optical CDMA system (defined as: (number of codes possible)/(code dimension, or rows* columns of the matrix codes)).