Abstract: In order to solve the problems of efficient variable-length name lookup, hardware-supportable storage consumption, and detection of interest flooding attack in the pending interest table(PIT) of named data networking, an cognition and indexing model(C&I) based on character convolutional neural network is proposed. C&I can support the classification and aggregation of name data, and reduce the storage consumption of name data. At the same time, a pending interest table storage structure C&I-PIT based on C&I and its data retrieval algorithm, which supports the detection of interest flooding attack, is proposed. Through the deployment of multi-level memory, the index structure and storage space are respectively deployed on static random access memory and dynamic random access memory. Experiments show that C&I-PIT has good performance in name aggregation, memory consumption and interest flooding attack detection.

Key words: named data networking, pending interest table, name lookup, character convolutional neural network, interest flooding attack