Abstract: A method of deep neural network (DNN) modularization for optimal loading includes receiving, by an electronic device, a DNN model for execution, obtaining, by the electronic device, a plurality of parameters associated with the electronic device and a plurality of parameters associated with the DNN model, determining, by the electronic device, a number of sub-models of the DNN model and a splitting index, based on the obtained plurality of parameters associated with the electronic device and the obtained plurality of parameters associated with the DNN model, and splitting, by the electronic device, the received DNN model into a plurality of sub-models, based on the determined number of sub-models of the DNN model and the determined splitting index.

Abstract: The present invention provides an integrated prevention of header, state, rate and content anomalies along with network policy enforcement for domain name service (DNS). A hardware-based apparatus helps identifying DNS rate-thresholds through continuous and adaptive learning. The apparatus can determine DNS header and DNS state anomalies and drop packets containing those anomalies. DNS queries and responses are inspected for known malicious contents using a Content Inspection Engine. The apparatus integrates advantageous solutions to prevent anomalous packets and enables a policy based packet filter for DNS.

Abstract: The present invention provides an integrated prevention of header, state, rate and content anomalies along with network policy enforcement. A hardware based apparatus classifies layers 2, 3, 4 and 7 network data and maintains rate-thresholds through continuous and adaptive learning. In the process of classifying the packets, the apparatus can determine header and state anomalies and drop packets containing those anomalies. Accurate detection and prevention of layer 7 content anomalies is achieved using fragment assembly, TCP reorder and retransmission removal components, which also identify anomalies in those areas. Content inspection is achieved at high speed through a Content Inspection Engine. The apparatus integrates advantageous solutions to prevent anomalous packets and enables a policy based packet filter.

Abstract: The present invention provides a method and apparatus for searching multiple strings within a packet data using deterministic finite automata. The apparatus includes means for updating memory tables stored in a layered memory architecture comprising a BRAM, an SRAM and a DRAM; a mechanism to strategically store the relevant data structure in the three memories based on the characteristics of data, size/capacity of the data structure, and frequency of access. The apparatus intelligently and efficiently places the associated data in different memories based on the observed fact that density of most rule-sets is around 10% for common data in typical network intrusion prevention systems. The methodology and layered memory architecture enable the apparatus implementing the present invention to achieve data processing line rates over 2 Gbps.