Using Temporal Awareness to Improve Distributed Problem Solving
Saeid SamadiDana, Roger Mailler · 2019
Many distributed problem solving algorithms are designed to work in conditions where computational power and information is distributed and cannot be centralized in a timely fashion. It is interesting to note that most distributed algorithms do not consider time as an integral component of their problem solving tactics. In this study we present a new algorithm based on theDistributed Probabilistic Protocol (DPP), called mDPP, that explicitly using time and prior knowledge to determine when to act. We examined the performance of mDPP on DCSP problems and compared its result with some distributed algorithms including DSA, DBA and MGM. We show that mDPP works very well and can be used as a basis for designing future distributed algorithms.