A suite of secure multi-party computation algorithms for solving distributed constraint satisfaction and optimization problems
Silaghi, Marius Cӑlin
MetadataShow full item record
Privacy requirements can be modeled within the distributed constraint satisfaction framework, where the constraints are secrets of the participants. In  we introduced a first technique allowing agents to solve distributed constraint problems (DisCSPs), without revealing anything and without trusting each other or some server. The first technique we propose now, MPC-DisCSP2, is a dm times improvement for m variables of domain size d. A second technique we propose, MPC-DisCSP3, has a similar complexity as MPD-DisCSP2, a little slower in the worst case, but guarantees that the returned solutions are picked according to a uniform distribution over the total set of solutions. A third technique, MPC-DisCSP0, is based solely on secure arithmetic circuit evaluation and the distribution for picking its solutions can be manipulated in a more flexible way. Nevertheless this last technique has a O(d **m factorial times d **m) complexity. All techniques of  can be extended to solve optimization for distributed weighted CSPs.