Immune Multi-objective Optimization Based Negative Selection Algorithm

As the development of human immune system (HIS), an increasing attention has been paid to some amazing characters, such as highly evolved, parallel, distributed and adaptive. In order to better solve the problem involved in related engineering applications, artificial immune system (AIS) has been proposed inspired by the information processing abilities of HIS. Among many of the mechanisms in AIS, optimal computation provides a novel approach for improving the performing of intrusion detection, all of which mechanisms are involved in the most discussed areas. Particularly, a New Sphere-dominance based Preference Immune Algorithm (SPIAⅡ) is proposed at first in this paper. And a negative selection algorithm with optional modules is proposed for intrusion detection subsequently. Finally, we successfully propose the implementing strategy about using constrained multiobjective immune algorithm (CMIA) for V-detector negative selection.Chapter two mainly proposes an improved sphere-dominance relationship combined with the concept of soft constraint. This dominance relationship can better evaluate the preference information of Decision Maker (DM), and ensure the stability of converging to the PF within the preference area. Based on this new dominance, a New Sphere-dominance based Preference Immune Algorithm (SPIAⅡ) is proposed. This algorithm can effectively improve the diversity of antibodies, according to which the stability of convergence will be enhanced.In chapter three, V-detector negative selection algorithm, which is one of the most important algorithms in the area of intrusion detection, will be mainly discussed. As we known, this algorithm includes several mechanisms, such as two different representative interpretations of self sample and two methods of estimating detector coverage. In addition, all of them hold there own merits and demerits. For better detecting complicated intrusion data, a negative selection algorithm with optional modules will be proposed. This approach can well integrate all these mechanisms into some modules for handling different test data flexibly.In chapter four, the main contribution is the successful application about improving the negative selection algorithm (NSA) with constrained multiobjective optimization method. It is means that a novel constrained multiobjective immune algorithm (CMIA) for V-detector negative selection is proposed. CMIA can produce a series of detectors with an optimal distribution, and better satisfied the common engineering goals in various NSAs. The new algorithm increases the effectiveness of sigle detector; so a larger dangerous area can be covered by fewer detectors with a more reasonable distribution. That means the detection rate will be effectively enhanced with fixed scale of detector set.This work was supported by the National High Technology Research and Development Program (863 Program) of China (Grant No.2009AA12Z210), the National Natural Science Foundation of China (Grant Nos.60703107), the Key Scientific and Technological Innovation Special Projects of Shaanxi (Grant No. 2008ZDKG-37), the Program for New Century Excellent Talents in University (Grant No. NCET-08-0811), the Program for New Scientific and Technological Star of Shaanxi Province (Grant No.2010KJXX-03), and the Fundamental Research Funds for the Central Universities (Grant No. K50510020001).