Loss probability in distributed‐type communication networks through semishortest routing–for connectionless services
Juhro Takase · Electronics and Communications in Japan (Part I Communications) · 1996
Abstract A number of logical networks with differing traffic characteristics need to be generated over a distributed physical network in order to deal effectively with future call types and service configurations that involve a range of basic situations. Also, each call must examine one connectionless‐type logical network by selecting the logical network that conforms to fundamental conditions based on the connection stipulations. The shortest route is selected through each switching node for each of the individual header data blocks (HB) of the call. The semishortest routing scheme that permits an increase in the number of relay stages is chosen, notably when a block occurs. Using this scheme, the HB loss probability between the originating and destination nodes is maintained reliably at about Nminld2 where the HB discard rates for each node are d equally and each originating and destination node is sufficiently distant from each other. Nmin is the minimum number of relay stages between the originating and destination nodes, l equals 2 in grid‐type networks, and l equals 3 in honeycomb‐type networks. The number of relays on the route between the originating and destination nodes can be a maximum of 2Nmin. The loss probability can increase by controlling the maximum number of relay stages. If the number of relays is limited to Nmin + 2, the reliable maximum value of the loss probability is approximately Nmind. The minimum reliable value of the loss probability is maintained at a level Nminld2.