Neural brouter for network interconnections: MS-net interconnects DQDB-nets

The performance of interconnected distributed-queue dual-bus (DQDB) networks is studied while the Manhattan street network (MSN) is being used as the backbone architecture with neural brouters as its nodes. It is interesting to note that, since the MSN topology is logically equivalent to a multichannel ring network, it may be easily adapted to operate as a token ring-of course, with multiple tokens-or token mesh network at heavy load conditions. Relevant performance measures using both an analytical method and simulation are also included.<>