Multi Variant Statistical Model for Improved Botnet Detection in Dynamic Networks Using Particle Swarm Optimization

The presence of compromised hosts in any network degrades the performance in many ways. The compromised nodes form the botnet and the detection of them is most primary solution to improve the network performance. The problem of botnet detection has been approached in many ways using fl ow factors but does not produce a remarking solution. To improve the botnet detection effi ciency, a novel statistical analysis model has been discussed in this paper. The multi variant statistical analysis model considers payload, TTL, host names, hop count, transmission frequency, and common nodes to perform botnet detection. The statistical model computes stream weight, transmission delay weight, and host weight. The proposed method applies the particle swarm optimization to perform botnet detection. The selection function computes the botnet weight using the result of statistical model.