Handshake Circuits for RHBD Systems

Fault-tolerant synchronous handshaking circuits are synthesized that avoid, detect and recover from radiation induced faults. This approach is suitable for commercial deepsubmicron CMOS processes. Every computation uses error detecting codes. If a computation output shows no errors, then handshaking signals are asserted allowing data and control information in preceding and following nodes to propagate. If an error is detected, then recovery is attempted by re-trying the computation. If the fault was a glitch, then the re-try is successful. If a re-try is unsuccessful, then a permanent fault is assumed and the computation is re-tried on a redundant hardware module(s). Once the computation shows no errors, then the handshaking signals are asserted as in normal operation. Since all circuits use delay insensitive handshaking circuits, no other circuits are effected by the extra time required to recover from the fault. No checkpoints or rollback re-initialization are needed since handshaking is utilized. As a result, the use of handshake circuits provides an efficient low-cost error recovery mechanism for CMOS circuits.