Probe-based distributed algorithms for deadlock detection

Distributed algorithms discussed in this paper provide a better performance than other well known algorithms designed for detecting deadlock, with respect to the communication bus load and fault tolerance, while the same assumptions for the operating conditions are preserved. Several concepts are defined for a group of deadlocked distributed tasks: probe messages, propagation law of a probe message, centralized and distributed management of probe messages, and cyclic trace of a probe message. Also, a propagation law for probe messages and two theorems are formulated and tested to establish valuable conditions of deadlock detection in distributed applications.