In: IEEE: Proceedings of 12th Intern. Parallel Proc. Symposium & 9th Symp. on Parallel and Distr. Processing, March 30 - April 3, Orlando, Florida, pages 791-795. 1998.
Abstract: A number of solutions have been proposed for the problem of mutual exclusion in distributed systems. Some of these approaches have since been extended to a prioritized environment suitable for real-time applications but impose a higher message passing overhead than our approach. We present a new protocol for prioritized mutual exclusion in a distributed environment. Our approach uses a token-based model working on a logical tree structure, which is dynamically modified. In addition, we utilize a set of local queues whose union would resemble a single global queue. Furthermore, our algorithm is designed for out-of-order message delivery. handles messages asynchronoulsy and supports multiple requests from one node for multi-threaded nodes. The prioritized algorithm has an average overhead of O(log(n)) messages per request for mutual exclusion with a worst-case overhead of O(n), where n represents the number of nodes in the system. Thus, our prioritized algorithm matches the message complexity of the best non-prioritized algorithms while previous prioritized algorithms have a higher message complexity, to our knowledge. Our concept of local queues can be incorporated into arbitrary token-based protocols with or without priority support to reduce the amount of messages. Performance results indicate that the additional functionality of our algorithm comes at the cost of 30% longer response times within our test environment for distributed execution when compared with an unprioritized algorithm. This result suggests that the algorithm should be used when strict priority ordering is required.