Most mutual exclusion algorithms require O(n) operations to enter the critical section despite how many may be actively trying to enter the critical section. This paper presents a mutual exclusion algorithm that is much more sensitive to how many processes currently want to enter the critical section. This algorithm is based on a parameter l, and assumes there is a total of n = kl processes. If only one process wants to enter the critical section, l+7 operations are sufficient. If there are t processes currently wanting to enter the critical section, then O(tlk) operations are all that is necessary. This is usually much less than the O(n) operations required by ordinary mutual exclusion algorithms. If the need is only to elect one process, the same minimum of 8 operations will hold, but the number of variables used will be O(log t), if t processes are contending to be elected. This algorithm will also by symmetric, with no distinctions between processes.
CITATION STYLE
Styer, E. (1992). Improving fast mutual exclusion. In Proceedings of the Annual ACM Symposium on Principles of Distributed Computing (pp. 159–168). Publ by ACM. https://doi.org/10.1145/135419.135453
Mendeley helps you to discover research relevant for your work.