A DRDoS Detection and Defense Method Based on Deep Forest in the Big Data Environment

摘要: Distributed denial-of-service (DDoS) has developed multiple variants, one of which is distributed reflective (DRDoS). Within the increasing number Internet-of-Things (IoT) devices, threat DRDoS attack growing, and damage a more destructive than other types. Many existing methods for cannot generalize early detection, leads to heavy load or degradation service when deployed at final point. In this paper, we propose detection defense method based on deep forest model (DDDF), then integrate differentiated into filter out flow. Firstly, from statistics perspective different stages flow in big data environment, extract host-based index (HDTI) network Secondly, using HDTI feature build forest, consists 5 estimators each layer. Lastly, procedure applies result DDDF drop identified points. Theoretical analysis experiments show that proposed can effectively identify with higher rate lower false alarm rate, also shows distinguishing ability eliminate flow, dramatically reduce attack.