加密流量分类torch实践4:TrafficClassificationPandemonium项目更新
更新日志
代码已经推送开源至露露云的github,如果能帮助你,就给鼠鼠点一个star吧!!!
3/10号更新
流量预处理更新
-
增加了基于
splitCap.exe
分流预处理,并且除了提取负载与包长序列后,支持提取统计特征(26维度)。26维度统计分别为
"Avg_syn_flag", "Avg_urg_flag", "Avg_fin_flag", "Avg_ack_flag", "Avg_psh_flag", "Avg_rst_flag", "Avg_DNS_pkt", "Avg_TCP_pkt","Avg_UDP_pkt", "Avg_ICMP_pkt", "Duration_window_flow", "Avg_delta_time", "Min_delta_time", "Max_delta_time", "StDev_delta_time","Avg_pkts_lenght", "Min_pkts_lenght", "Max_pkts_lenght", "StDev_pkts_lenght", "Avg_small_payload_pkt", "Avg_payload", "Min_payload","Max_payload", "StDev_payload", "Avg_DNS_over_TCP", "Num_pkts"
从
entry.pcap2npy/1_preprocess_with_splitCap_1.py
进入配置文件preprocess下路径要为windows格式
运行完的预览图,可以看到有statistic.npy
的统计特征文件
- 增加了基于
cic-meterflower
工具对pcap的处理,将pcap处理为csv格式文件
使用
entry.pcap2csv/1_preprocess_with_cic.py
,参考博客流量预处理-3:利用cic-flowmeter工具提取流量特征修改相应的路径变量注意:pcap路径与名称在使用该方式处理时不能出现中文,否则报错。
运行完的预览图,可以看到已经对中文进行改名,出现各个标签的csv文件