Checking a Non-Byzantine FT Scheme Against Byzantine Faults

Process duplication for several application processes in combination with appropriate protocols provides byzantine fault tolerance. In contrast, lightweight protocols are often used among duplicated processes that are primarily not able to tolerate byzantine faults but reach fault tolerance under acceptable costs. Instead of guaranteeing byzantine fault tolerance by protocols, an overlaid system is suggested, to detect byzantine faults during normal processing and to detect byzantine faults possibly occurring within a recovery procedure. It is based on collecting and fusing digital signatures. In certain periods checks can be applied locally. These checks circumstantiate whether the current state has been reached by two duplicates in the same way and whether received messages were generated by two sender duplicates with the same content in the same order. Such a check over collected signatures is used to indicate absence of byzantine faults for certain processing periods.