基于深度学习检测恶意流量识别框架（80+特征/99%识别率）

基于深度学习检测恶意流量识别框架

简要

示例

a.检测攻击类别

b.模型训练结果输出参数

c.前端检测页面

d.前端训练界面

e.前端审计界面（后续更新了）

f.前端自学习界面（自学习模式转换）

f1.自学习模式

这里解释下：这里有两个模式，开启工作模式后，确保当前流量为正常流量，系统会自动标记并在达到阈值后进行训练，从而增加泛化能力，反之。

进度条显示内容解释：当前|总进度|训练轮数|源数据

核心代码示例

a.代码结构

b.数据预处理

def __serial(self,debug=0):
        self.data['Timestamp'] = self.data['Timestamp'].apply(lambda x: self.__timestamp_to_float(x))
        self.data['Dst_IP'] = self.data['Dst_IP'].apply(self.__ip_to_float)
        self.data['Src_IP'] = self.data['Src_IP'].apply(self.__ip_to_float)
        if debug:
            self.__pull(self.data,"d1.txt")
        self.data["Label"] = self.data["Label"].apply(self.__label_to_float)
        columns_to_convert = [col for col in self.data.columns if col not in ['Timestamp', 'Dst_IP', 'Src_IP',"Label"]]
        for col_name in columns_to_convert:
            self.data[col_name] = pd.to_numeric(self.data[col_name], errors='coerce')
        self.data = self.data.apply(pd.to_numeric, errors='coerce')
        self.data = self.data.fillna(0)
        inf_values = ~np.isfinite(self.data.to_numpy())
        self.data[inf_values] = np.nan  # 替换为NaN，您也可以选择替换为其他合理值
        self.data = self.data.dropna()  # 删除包含缺失值的行
        self.features = self.data.iloc[:, :-1]
        self.labels = self.data.iloc[:, -1]  # 标签
        if debug:
            self.__pull(self.data,"d2.txt")
        self.scaler = StandardScaler()
        self.features = self.scaler.fit_transform(self.features)
    

c.抓包模块

def packet_to_dict(packet):
    packet_dict = {}
    if const.cdist[const.pkg_id] > const.cdist[const.max_pkgn]:
        const.cdist[const.pkg_id] = 0
    packet_dict["data"] = packet
    packet_dict["id"] = const.cdist[const.pkg_id]
    const.cdist[const.pkg_id] +=1
    if IP in packet:
        packet_dict["src_ip"] = packet[IP].src
        packet_dict["dst_ip"] = packet[IP].dst
    else:
        packet_dict["src_ip"] = ""
        packet_dict["dst_ip"] = ""
    return packet_dict

def write_packet_summary(filename, packet_summary):
    with open(filename, 'a') as file:
        file.write(packet_summary + '\n')

def listen(key,qkey,filename):

    # 定义回调函数来处理捕获到的数据包
    def packet_callback(packet):
        try:
            packet_info = packet_to_dict(packet)
            if packet_info != {}:
                const.cdist[qkey].put(packet_info)
        except Exception as e:
            log.Wlog(3,f"listen* {e}")
        try:
            timestamp = datetime.now().strftime('%Y-%m-%d %H:%M:%S:%f')[:-3]
            summary = packet.summary()
            packet_with_timestamp = f"[{timestamp}] >> {summary}"
            write_packet_summary(filename, packet_with_timestamp)
            maintain_packet_summary(filename, max_lines=20)
        except Exception as e:
            log.Wlog(3, f"listen* {e}")
        # return packet.summary()

    # 定义停止条件函数
    def stop_condition(packet):
        # print(const.cdist[key],key)
        return const.cdist[key]

    # 开始捕获数据包，使用 stop_filter 参数指定停止条件
    sniff(
        iface=const.cdist[const.net_interface],
        prn=packet_callback,
        stop_filter=stop_condition
    )

d.数据库操作

def data_init():
    # 连接到数据库，如果不存在则创建
    conn = sqlite3.connect(const.cdist[const.sql_dbp])

    # 创建游标对象
    cur = conn.cursor()

    # 创建数据表
    cur.execute('''CREATE TABLE IF NOT EXISTS pkg_data (
                    id INTEGER PRIMARY KEY,
                    src_ip TEXT,
                    dst_ip TEXT,
                    data TEXT,
                    time1 INTEGER,
                    label INTEGER
                    )''')
    cur.close()
    conn.close()
    
def get_sql_cur():
    # 连接到数据库，如果不存在则创建
    conn = sqlite3.connect(const.cdist[const.sql_dbp])

    # 创建游标对象
    cur = conn.cursor()
    return cur,conn
def close_sql(cur,conn):
    try:
        cur.close()
        conn.close()
    except:
        pass
# 添加数据pkg_data
def add_data(src_ip, dst_ip, data, time1, label):
    cur,conn = get_sql_cur()
    cur.execute("INSERT INTO pkg_data (src_ip, dst_ip, data, time1, label) VALUES (?, ?, ?, ?, ?)", (src_ip, dst_ip, data, time1, label))
    conn.commit()
    close_sql(cur,conn )

# 删除指定 src_ip 的数据
def delete_data(src_ip):
    cur,conn = get_sql_cur()
    cur.execute("DELETE FROM pkg_data WHERE src_ip=?", (src_ip,))
    conn.commit()
    close_sql(cur,conn )

# 查询指定时间戳范围内的域名及出现次数
def query_data_k1(start_timestamp, end_timestamp):
    cur,conn = get_sql_cur()
    cur.execute("SELECT src_ip, COUNT(*) FROM pkg_data WHERE time1 BETWEEN ? AND ? GROUP BY src_ip", (start_timestamp, end_timestamp))
    rows = cur.fetchall()
    close_sql(cur,conn )
    return rows

e.全局变量实现

# const.py
cdist = {}
def _const_key_(key, value):
    cdist[key] = value

# run.py
def init():
    odir = os.getcwd()
    signal.signal(signal.SIGINT, quit)                                
    signal.signal(signal.SIGTERM, quit)
    const._const_key_(const.log_path, f"{odir}/plug/utils.log")
    const._const_key_(const.temp_pkg, f"{odir}/plug/temp.pkg")
    const._const_key_(const.out_csv_d, f"./temp_pkg_data/csv/")
    const._const_key_(const.out_pcap_d, f"./temp_pkg_data/pcap/")
    const._const_key_(const.train_info,f"{odir}/plug/train.info")
    const._const_key_(const.sql_dbp,f"{odir}/plug/pkg_data.db")
    const._const_key_(const.out_atrain_d,f"./temp_pkg_data/atrain/")
    const._const_key_(const.Base_h5,f"{odir}/2800w-base.h5")
    const._const_key_(const.deeps,deep_s.DeepS())
    const._const_key_(const.AddTrain_Stream_Mode,{"mode":0,"args":"","key":"","label":"","csvp":"","echo":0}) # 0不进行模式，1进行正常流量训练
    const._const_key_(const.Pkg_DATA_List,[])
    const._const_key_(const.max_pkgn,2000)
    const._const_key_(const.MAX_ADDTrain_n,10241)
    const._const_key_(const.pkg_id,0)
    const._const_key_(const.log_level, 3)
    const._const_key_(const.queue1, Queue(maxsize=65535))  # 创建队列
    data.data_init()
    f= open(const.cdist[const.train_info], 'w')
    f.close()
    CronWork(100,odir)