A fault model notation and error-control scheme for switch-to-switch buses in a network-on-chip

Managing Temperature Effects in Nanoscale Adaptive Systems

“…Hamming codes are the most widely used codes in previous research on interconnect link error protection [184][189] [190] [191][192] [194]. As mentioned earlier, the minimum distance of a Hamming code is 3, so it can correct a single error in each codeword.…”

“…Hamming codes can enhance the burst error tolerance of the larger codeword and also tolerate multiple errors (provided each error occurs in a separate group) [194], [209].…”

On reliable modular testing with vulnerable test access mechanisms

“…It is obvious that test data cannot be corrupted if the error correction scheme is capable to fully recover the error. However, the associated cost of such techniques (e.g., forward error correction [27]) in terms of area and energy-efficiency is quite high and hence retransmission scheme is the mainstream technique utilized nowadays. Also, even for the case when single error correction is employed, since the probability of a double (or higher) error within a single flit may not be insignificant due to crosstalk, typically a hybrid technique that provide both error correction and retransmission is utilized [23].…”

An adaptive system-on-chip for network applications

“…While other NoCs like [6,7,10] make use of relatively lean protocols, the NoC used in DynaCORE utilises a three-layer protocol stack. The decision for a more complex protocol was made, because processing instances should be relocatable, there should be support for prioritised traffic, and a future expansion of the DynaCORE implementation to multiple FPGAs should be taken into account.…”