Title : 
A new approach for byzantine agreement
         
        
            Author : 
Wang, S.C. ; Kao, S.H.
         
        
            Author_Institution : 
Dept. of Inf. Manage., Chaoyang Univ. of Technol., Taiwan
         
        
        
        
        
        
            Abstract : 
Due to the malicious attacks and software errors cause faulty nodes to exhibit arbitrary behavior, a byzantine agreement (BA) is increasingly important. A novel solution for the BA problem is presented. We propose the use of an artificial neural network (ANN), a back propagation network (BPN), to help reach a common value in a fully connected network (FCN), and as a substitute for storing messages in each processor. There are several significant advantages: (1) using ANN to store messages, the memory space required in previous approaches is no longer needed; (2) the characteristic of the learning ability can help other new applications of the BA reach an agreement; and (3) parallel processing can be achieved in each processor
         
        
            Keywords : 
backpropagation; computer network reliability; fault tolerant computing; neural nets; parallel processing; protocols; security of data; ANN; back propagation network; byzantine agreement; distributed computing system; fault tolerance; fully connected network; learning; malicious attacks; memory space; message storage; parallel processing; protocol; reliable computer system; software errors; Artificial neural networks; Chaotic communication; Computer network reliability; Distributed computing; Fault tolerant systems; Information management; Interference; Neural networks; Parallel processing; Telecommunication network reliability;
         
        
        
        
            Conference_Titel : 
Information Networking, 2001. Proceedings. 15th International Conference on
         
        
            Conference_Location : 
Beppu City, Oita
         
        
            Print_ISBN : 
0-7695-0951-7
         
        
        
            DOI : 
10.1109/ICOIN.2001.905500