Service oriented Multi Model Network Inference Model for Identifying Denial of Service Attack in Network Immune System

Abstract

Identifying DDOS attacks has become more important task in network management and security where the immune system has responsibility of securing services and data available in the network. We propose a service oriented multi model (SOM) network inference model for identifying DDoS attacks, which consist of several phases. The SOM network inference model provides the solution for variety of services. Unlike other methodology available, the proposed model uses various metrics at various levels and combines the earlier models at required stages of identification process. The proposed model maintains the log trace where the features of packets are kept stored which are already received. The proposed approach extracts various features like ttl, payload, hop count, addresses, time, service details. Using those logs, the proposed approach splits them into time domain values for each of the distinct services available. The model maintains the protocol of services available and based on that the service access history according to each time window is split into different numbers, from which the flow approximation is computed using previous access history. Based on the results of time orient and flow approximation, the proposed model infers the packet signature. The proposed model produces efficient results and reduces the time complexity.