Distributed computing building blocks for rational agents

Abstract

Following [4] we extend and generalize the game-theoretic model of distributed computing, identifying different utility functions that encompass different potential preferences of players in a distributed system. A good distributed algorithm in the game-theoretic context is one that prohibits the agents (processors with interests) from deviating from the protocol; any deviation would result in the agent losing, i.e., reducing its utility at the end of the algorithm. We distinguish between different utility functions in the context of distributed algorithms, e.g., utilities based on communication preference, solution preference, and output preference. Given these preferences we construct two basic building blocks for game theoretic distributed algorithms, a wake-up building block resilient to any preference and in particular to the communication preference (to which previous wake-up solutions were not resilient), and a knowledge sharing building block that is resilient to any and in particular to solution and output preferences. Using the building blocks we present several new algorithms for consensus, and renaming as well as a modular presentation of the leader election algorithm of [4].

Fair Leader Election for Rational Agents in Asynchronous Rings and Networks

“…1 Code for a normal processor i: 2 Initially,d i,i := d i 3 for j = 1 to n do 4 Wait to receive a data messaged i,i−j (mod n) and then send the previous data messagê d i,i−j+1 (mod n) . 5 if j = i then 6 wait for an incoming validation messagev i,j and forward it immediately. perform validation by sending v i and waiting to receive it.…”

“…In our model, we assume that the set of ids is known to all the processors prior to the execution of the protocol, however originally in [4,5] the ids are not given, but learned during the execution of the protocol. Their protocol includes a preceding wake-up phase, where processors exchange ids and agree upon a direction for the ring.…”

“…Out of the many challenges in proving these resilience conjectures, we address one specific issue -the adversaries might cause every honest segment to believe it contains an origin processor. In the wake up phase presented by [4,5], the processors exchange ids, and the processor with the lowest id is selected to be the origin. There cannot be two origin processors in the same honest segment because the set of ids perceived by every two processors in the same honest segment is identical.…”

“…Later, Afek et al [5] simplified A-LEAD uni , and decomposed it into useful intuitive building blocks.…”

Rational Fair Consensus in the Gossip Model

“…Perhaps, this framework has been investigated for the first time in [13], where rational behaviour is analyzed in secret-sharing problems and multiparty computations. More recently, the impact of rational adversaries has been investigated on fundamental tasks in Distributed Computing such as leader election, consensus, and wake-up problems [2,3,12,14].…”