New Codes and New Approaches for Code Division Multiple Access on Optical Fiber Networks
MetadataShow full item record
This thesis concerns the use of optical fiber in a multi-user communication networks. specifically, it considers the use of code-division multiple access (CDMA) techniques that permit many users to share a single optical channel through the assignment of unique "signature sequences".<P>In most optical systems - where incoherent processing means that only signal intensity is measured - there are no negative signal components, and the effect on code design is profound. This was noted by Salehi and others in their design of optical orthogonal codes (OOC's).<P>The results in this thesis extend previous work on OOC's in several areas. First, we consider codes for which the auto- and cross- correlation constraints are not equal. We observe that the effects of the two constraints on system performance are not identical, and so considering only codes for which they are identical may lead to a sub-optimal code. Bounds on such OOC's are derived and techniques for constructing them are described.<P>OOC's with unequal auto- and cross-correlation constraints may be viewed as constant-weight unequal error protection (UEP) codes; therefore we in interpret the bounds and constructions in that context and compare them with previous work on UEP codes.<P>We develop bounds on the size of "variable weight" OOC's and demonstrate techniques for building them.<P>By considering the original CDMA system as an inner code and adding a channel encoder for each user as an outer code, we demonstrate a concatenated coding scheme that can improve system performance without increasing the weight of the OOC sequences.<P>A new approach to CDMA on optical fiber networks employing pulse position modulation (PPM) is developed. We propose assigning to each user in the network two signature sequences - a synchronization sequence and a data sequence. We stipulate the required properties of such sequences, and we demonstrate that in some cases more users can be accommodated with this approach than with the assignment of single sequences that "do it all".