While all other fault tolerance paradigms require networks of high con
nectivity to tolerate a substantial number of faults, it was shown in
Dwork et al. that the new paradigm can be achieved even on bounded deg
ree networks, as long as the number of faults is bounded by O(n/log n)
, where n is the size of the network. A major problem that was left op
en in Dwork et al. is whether almost everywhere agreement can be achie
ved on bounded degree networks in the presence of up to O(n) faulty no
des (processors). In this work, we answer this question in the affirma
tive. As in Dwork et al., our solution is based on a general technique
for simulating on a bounded degree network an algorithm designed for
the complete network. Each communication round of the complete network
protocol is simulated by a logarithmic number of communication rounds
, and with a polynomial number of messages. (C) 1994 Academic Press, I
nc.