Centralized computing is analogous to star topology in the networks where there is n entry and n-1 exit points. We see all the computation being carried out by the pivot processor and also embed in itself a decision routine for pass on the message to other linked systems. These systems are merely the mirror whereas the computing is done by the pivot processor.
Decentralized Computing is analogous to distributed node based processing where the topology for interaction among the nodes are random and hence some times there is delay because of the orientation of nodes by since the computation is disperse, it compensates in the time in processing, storage and retrieval.
You can definitely check into IBM and Amazon Research for examples for the Centralized and Decentralized computing.
