Great detail and research about two algorithms for real-time communication in distributed systems are provided in this paper. These algorithms, LMR and DTA, are very specific to the types of token-ring networks on which they run, IEEE 802.5 token-ring and fiber distributed data interface (FDDI), respectively. The algorithms are adaptive to situations in which latencies in communication are being delayed beyond some threshold, adjusting in order to correct for these deficiencies. Thus these are system-level monitoring algorithms.
LMR dynamically identifies which messages between nodes in the token-ring network contribute to slower timeliness in the real-time system, and changes their message priorities in order to improve end-to-end computation time.
DTA, on the other hand, identifies which host machines are generating the trans-node messages that contribute to slower timeliness, and attempts to give those hosts a longer token-holding time so they can catch up.
The algorithms are provided in detail and compared with one another for performance under certain scaling loads. They are also compared with an existing algorithm that attempts to solve similar problems. This paper’s main contribution, then, is the introduction of two new algorithms. It provides a great deal of information about their actual performance in experimental conditions, and they are shown to perform favorably under all conditions tested.