Evaluation of Stateful Reliability Counter in Small-World Cellular Neural Networks

In this paper, fault tolerant techniques and the evaluation results for the small world cellular network are described. The small world network is characterized by smaller number of hops necessary for the communication between nodes in the network compared with usual regular networks. So that, high speed operations are expected in image processing and the kind of field. However, the fault tolerance of the permanent fault is a severe problem, because the error propagation tends to spread quickly compared with usual networks due to the network characteristics. Moreover, it is expected that the integration limitation in the circuit technology will come in the near future, so it is also important to cope with the temporary fault caused by humidity, heat, and the radiation, and so on. In this paper, effective techniques of the fault tolerance for the temporary fault are proposed as well as a permanent fault, and the evaluation results are provided.