There are two basic approaches for establishing a connection in a reconfigurable switched optical network, whose links are multipiexed with virtual channels (e.g., wavelengths or time slots). One is called path multiplexing (PM), in which the same virtual channel has to be used on each link along a path, and the other is link multiplexing (LM), in which different virtual channels may be used. We focus on the problem of off-line permutation embedding and scheduling as a part of the comparative study of the multiplexing approaches. Specifically, we determine the minimum number of virtual channels per link needed for a given network to be rearrangeably nonblocking in PM and LM, respectively. We also examine the schedule length of a permutation in PM and LM when the network is blocking as a result of having an insufficient number of virtual channels per link. We found that PM and LM are equally effective in linear arrays, and LM is slightly more effective than PM in rings, meshes (grids), tori, and hypercubes.

Path multiplexing (PM) and link multiplexing (LM) are two approaches for establishing connections (or lightpaths) in optical networks. This paper describes distributed control protocols for establishing lightpaths in WDM networks using LM and PM. We propose and evaluate the performance of two classes of protocols, namely source initiated reservation (SIR) and destination initiated reservation (DIR). It is found that DIR protocols generally perform better than SIR protocols. However, the impacts of DIR protocols on the performance of a network using LM and PM are different.

There are two approaches for establishing a connection in a network whose links are multiplexed with virtual channels. One is called Path Multiplexing (PM), in which the same channel has to be used on each link along a path, and the other is Link Multiplexing (LM), in which different channels may be used. We study the problem of embedding permutations in PM and LM, and in particular, determine the threshold (minimal) multiplexing degree needed for a network to be rearrangeably nonblocking. We found that PM and LM are equally effective in linear arrays, and LM is slightly more effective than PM in rings, meshes, tori and hypercubes. Our results suggest that PM may be more cost-effective in implementing networks with multiplexed optical interconnects.