5G networks are facing to new challenges related to the growing traffic volume and service diversity. Some of the major concerns in this new scenario are the security and privacy issues required for a full technology adoption. Traffic characterization is a compound of strategies intended to define formally the behaviour and patterns in the Internet traffic. In this work, we propose the use of statistical features of network flows to characterize some of the most common attacks in the current networks through the L-moment ratio diagrams. Our work identify the parameters that can discriminate normal from malicious traffic. Moreover, our preliminary results show that this technique enables the differentiation of anomalies and can also identify several types of attack traffic.