Abstract
A network is complete if every processor has a direct communication channel to every other processor. We consider the election problem on complete asynchronous networks when the processors are reliable while some of the channels are subject to failure. In this thesis, we have implemented Lokre's distributed election algorithm for faulty complete networks [1]. Lokre's algorithm works as long as each node has at most f faulty links incident on it, f < (n-1)/2 where n is the number of nodes in the network. The algorithm uses at most 0(n 2 + nf 2) messages to elect a unique leader of the network. Each message consists of at most 0(logITI) bits, where ITI is the cardinality of the set of node identifiers. We built a complete network, designed the random generating software modules, and implemented the algorithm. We ran the program for several values of n and for various configurations of the network. During each run we counted the number of messages passed, and the number of phases advanced before a leader was elected. For each value of n we did a best case, average case, and the worst case analysis of the number of messages passed. We compared the number of messages, and the number of phases advanced with the theoretical results. Doing this we show that the bounds proved by the algorithm are satisfied.
Krishna, P. (1993). Implementation of election algorithm for asynchronous complete networks with intermittent link failures. Master's thesis, Texas A&M University. Available electronically from
https : / /hdl .handle .net /1969 .1 /ETD -TAMU -1993 -THESIS -K924.