Real-Time Token-Based Mutual Exclusion Algorithms

Many token-based, distributed mutual exclusion algorithms can be generalized by a single algorithm. The algorithm’s performance is dependent upon the logical topology imposed on the nodes and the policy used to forward requests. This paper extends the generalized algorithm to support prioritized, real-time requests in a generalized fashion and presents models that can be used to analyze the performance and verify the correctness of the generalized algorithm. Both safety and liveness properties are verified. Model checking is also used to analyze performance. Using the best topology, the generalized algorithm attains the same worst-case performance as a centralized algorithm; i.e., three messages per critical section. In the average case, the generalized algorithm performs better than a centralized one when the star topology is used. Finally, requests by nodes at each priority level are processed in order, resulting in bounded, predictable worst-case response times.