Self-stabilizing Uniform Reliable Broadcast

Abstract

The problem of total-order (uniform reliable) broadcast is fundamental in fault-tolerant distributed computing since it abstracts a broad set of problems requiring processes to uniformly deliver messages in the same order in which they were sent. Existing solutions (that tolerate process failures) reduce the total-order broadcast problem to the one of multivalued consensus.Our study aims at the design of an even more reliable solution. We do so through the lenses of selfstabilization-a very strong notion of fault-tolerance. In addition to node and communication failures, self-stabilizing algorithms can recover after the occurrence of arbitrary transient faults; these faults represent any violation of the assumptions according to which the system was designed to operate (as long as the algorithm code stays intact).This work proposes the first (to the best of our knowledge) self-stabilizing algorithm for total-order (uniform reliable) broadcast for asynchronous message-passing systems prone to process failures and transient faults. As we show, the proposed solution facilitates the elegant construction of self-stabilizing state-machine replication using bounded memory.

FIFO and Atomic broadcast algorithms with bounded message size for dynamic systems

“…The problem of broadcasting messages has been widely studied in the message passing model [14], [15], in wireless networks [16], [17] or overlay network [18], [19] using gossip protocols.…”

“…On the other hand, during the ith broadcast of s, the message sent by s changes whenever the acknowledgment array myT S is updated (line block [15][16][17]. At every update, s increments myCnt value (line 16).…”

“…Let r = s be a process. During a round where r delivers a s-message, myT S(r)[s] changes its value (if-block of lines [15][16][17]. If i > 1 then r does not execute the line block 9-17 during t s,i according to γ + s,i−1 properties (Corollary 1) and t s,i properties (Lemma 1).…”

Self-stabilizing Byzantine-tolerant Broadcast

“…In the context of reliable broadcast, there are (non-self-stabilizing) Byzantine fault-tolerant solutions [31] and (non-Byzantine-tolerant) self-stabilizing solutions [28]. Our study focuses on the BT [13] solution to which we propose a self-stabilizing variation.…”

Self-stabilizing Multivalued Consensus in Asynchronous Crash-prone Systems

“…Our self-stabilizing solution is wait-free since termination is achieved within at most n invocations of binary consensus objects and at most one uniform reliable broadcast [32] (URB) operation per process, where n is the number of processes in the system. However, each such URB invocation needs to be repeated until the consensus object is deactivated by the invoking algorithm.…”