Time-to-delivery queuing: a multi-purpose resource allocation and congestion control technique

Time-to-delivery (TTD) queuing is presented as a technique that can be used to provide multiple resource allocation and congestion control functions in high-speed communication networks (frame relay networks are used for illustration). TTD queuing associates a TTD value with each data unit, and shared network queues serve data units in TTD order (smallest TTDs first). With the proper initial TTD assignments and updates, TTD queuing can (1) provide delay differentiation between connections with different negotiated average rates, (2) reduce the dependence of delay on number of hops in a connection, (3) allow for data units to “make up time” following a bottleneck node, and (4) provide delay and loss penalties for connections exceeding negotiated rates (access rate policing). Details of the TTD assignments and updates are provided along with performance comparisons (via simulation) with other queuing and policing mechanisms