computation - Identifying colluding nodes in a distributed computing environment -
i have distributed environment in computations carried out in parallel in same form of map reduce paradigm. want know if there result verification scheme such computations. have read basic method replicate computation in multiple worker nodes. in such method, hash value of computation returned worker nodes. hash value compared identify if result correct. method not prevent collusion in malicious nodes might communicate each other , return same result. there method available thwart collusion. there several collusion detection algorithms, i've not been able find algorithm identifies colluding group of nodes malicious. kindly give me insight on solution.
sounds you're trying solve byzantine generals problem.
the essence of solution problem problem (and linked paper) is:
the method of having 1 general simulate m others can used prove no solution fewer 3m + 1 generals can cope m traitors. proof similar 1 original byzantine generals problem , left reader.
so when expect m
of nodes can hacked , provide wrong answers, you'll need have @ least 3*m + 1
nodes detect faulty ones , find correct solution.
Comments
Post a Comment