Divide and conquer for distributed mutual exclusion

A simple formal model of distributed mutual exclusion algorithms is introduced and a new class of algorithms is derived using the divide and conquer paradigm. A number of metrics reflecting certain realistic concerns about the distributed mutual exclusion algorithms are proposed and several relationships among them are identified. It is shown that the divide and conquer paradigm significantly simplifies the derivation of algorithms with desired values for these metrics. The new class of algorithms is compared to the voting algorithms