Improving the normalized proportional allocation scheme for time-critical communication in FDDI networks

The normalized proportional allocation (NPA) scheme has been recommended for use in time-critical communication due to its near-optimal performance in terms of guaranteeing the transmission of synchronous messages before their deadlines in a token ring network (such as FDDI) where the timed token medium access control (MAC) protocol is employed. However, the use of this scheme in an FDDI network may lead to an over-allocation of synchronous bandwidth to a network node, which may in turn cause the deadline of a synchronous message that can actually be guaranteed in practice to be misjudged otherwise by the existing theoretical method. In this paper an example is given to demonstrate the misjudgement problem caused by the over-allocated synchronous bandwidth. We also suggest a solution to the above problem by proposing an improved NPA scheme which does not over-allocate synchronous bandwidths to nodes and therefore can effectively avoid the misjudgement problem with the NPA scheme.