Stability and performance of a simple distributed tracking policy for production control of manufacturing systems

The objective of distributed tracking is to operate a production system as closely as possible to an idealized regime obtained by a continuous-flow relaxation of the actual (discrete) production control problem. The author has investigated the stability and performance of a large class of distributed tracking policies called “non-idling-non-exceeding (NINE)” in an earlier paper. In this work the author focuses on the most natural trading policy in this class, and finds a tight bound on its performance for a single machine and a sufficient condition for its stability for multiple-machine systems. This condition is considerably less stringent than the one available for general NINE policies