Adding Analysis Code

Author: nasirkamal

analyze.py

@@ -0,0 +1,81 @@
+import argparse
+from wsgiref.simple_server import WSGIRequestHandler
+from calc_funcs import *
+
+args_parser = argparse.ArgumentParser()
+args_parser.add_argument('--sender_pcap', '-sp', type=str, required=True, help='Pcap from the node sending traffic.')
+args_parser.add_argument('--receiver_pcap', '-rp', type=str, required=True, help='Pcap from the node receiving traffic.')
+args_parser.add_argument('--sender_ip', '-sip', type=str, required=True, help='IP address of the sender node.')
+args_parser.add_argument('--receiver_ip', '-rip', required=True, help='IP address of the receiver node.')
+args_parser.add_argument('--dest_port', '-dp', type=int, required=False, default=5201, help='Listening server port, in case of iperf3 5201.')
+args_parser.add_argument('--name', '-n', type=str, required=False, default='noname', help='Name of the experiment to generate graphs with that name for distiction.')
+args_parser.add_argument('--hist_filename', '-hf', type=str, required=False, help='File name to save histogram of packet delays. E.g., hist.jpg')
+args_parser.add_argument('--pdf_filename', '-pf', type=str, required=False, help='File name to save pdf graph. E.g., pdf.jpg')
+args_parser.add_argument('--cdf_filename', '-cf', type=str, required=False, help='File name to save cdf graph. E.g., cdf.jpg')
+args_parser.add_argument('--pdf_cdf_filename', '-pcf', type=str, required=False, help='File name to save pdf-cdf graph. E.g., pdf-cdf.jpg')
+args_parser.add_argument('--bins', '-bn', default='auto', required=False, help='Number of bins for drawing histogram/pdf/cdf.')
+args_parser.add_argument('--hist_yscale', '-hy', default='linear', required=False, type=str, choices=['linear', 'log', 'symlog', 'logit'], help='Histogram Y-Scale, ')
+args_parser.add_argument('--pcaps_dir', '-pd', default='pcaps', required=False, type=str, help='Directory name containing pcaps.')
+args_parser.add_argument('--graphs_dir', '-gd', default='graphs', required=False, type=str, help='Directory name to save graph plots output.')
+
+args = args_parser.parse_args()
+
+if __name__ == '__main__':
+    if args.hist_filename:
+        hist_filename = os.path.join(args.graphs_dir, args.hist_filename)
+    else:
+        hist_filename = os.path.join(args.graphs_dir, args.name, args.name + '_histogram.jpg')
+
+    if args.pdf_filename:
+        pdf_filename = os.path.join(args.graphs_dir, args.pdf_filename)
+    else:
+        pdf_filename = os.path.join(args.graphs_dir, args.name, args.name + '_pdf.jpg')
+
+    if args.cdf_filename:
+        cdf_filename = os.path.join(args.graphs_dir, args.cdf_filename)
+    else:
+        cdf_filename = os.path.join(args.graphs_dir, args.name, args.name + '_cdf.jpg')
+
+    if args.pdf_cdf_filename:
+        pdf_cdf_filename = os.path.join(args.graphs_dir, args.pdf_cdf_filename)
+    else:
+        pdf_cdf_filename = os.path.join(args.graphs_dir, args.name, args.name + '_pdf_cdf.jpg')
+
+    create_if_not_present_dir(args.graphs_dir)
+    graphs_save_dir = os.path.join(args.graphs_dir, args.name)
+    create_if_not_present_dir(graphs_save_dir)
+
+    sender_pcap_filename = os.path.join(args.pcaps_dir, args.sender_pcap)
+    receiver_pcap_filename = os.path.join(args.pcaps_dir, args.receiver_pcap)
+
+    sender_scapy_cap = read_pcap(sender_pcap_filename)
+    receiver_scapy_cap = read_pcap(receiver_pcap_filename)
+    
+    ip_filtered_sender_scapy_cap = filter_pcap(scapy_cap=sender_scapy_cap, src_ip=args.sender_ip, dest_ip=args.receiver_ip)
+    ip_filtered_receiver_scapy_cap = filter_pcap(scapy_cap=receiver_scapy_cap, src_ip=args.sender_ip, dest_ip=args.receiver_ip)
+
+    src_port = get_src_port(ip_filtered_sender_scapy_cap)
+
+    sender_filtered_scapy_cap = filter_pcap(scapy_cap=ip_filtered_sender_scapy_cap, src_port=src_port, dest_port=args.dest_port)
+    receiver_filtered_scapy_cap = filter_pcap(scapy_cap=ip_filtered_receiver_scapy_cap, src_port=src_port, dest_port=args.dest_port)
+
+    sender_seq_time_list = get_pkt_seq_n_time(sender_filtered_scapy_cap)
+    receiver_seq_time_list = get_pkt_seq_n_time(receiver_filtered_scapy_cap)
+
+    sender_df = pd.DataFrame(sender_seq_time_list, columns =['Sequence', 'Time_Transmit'])
+    receiver_df = pd.DataFrame(receiver_seq_time_list, columns =['Sequence', 'Time_Receive'])
+
+    sender_df.drop_duplicates(subset ="Sequence", keep = 'last', inplace = True)
+    receiver_df.drop_duplicates(subset ="Sequence", keep = 'last', inplace = True)
+
+    final_df = pd.merge(sender_df, receiver_df, on='Sequence', how='outer')
+    final_df = remove_NaN(final_df)
+
+    final_df['Delay'] = (final_df['Time_Receive'] - final_df['Time_Transmit']) * 1000
+    data_arr = final_df['Delay'].to_numpy()
+
+
+    draw_hist(data_arr, bins=args.bins, yscale=args.hist_yscale, hist_filename=hist_filename)
+    draw_pdf(data_arr, bins=args.bins, pdf_filename=pdf_filename)
+    draw_cdf(data_arr, bins=args.bins, cdf_filename=cdf_filename)
+    draw_pdf_cdf(data_arr, bins=args.bins, pdf_cdf_filename=pdf_cdf_filename)

calc_funcs.py

@@ -0,0 +1,157 @@
+from scapy.all import *
+import os
+import numpy as np
+import pandas as pd
+import matplotlib.pyplot as plt
+
+def read_pcap(pcap_name):
+    scapy_cap = rdpcap(pcap_name)
+    print('Read ', pcap_name, '. Total Packets: ', len(scapy_cap))
+    return scapy_cap
+
+def create_if_not_present_dir(path):
+    if not os.path.exists(path):
+        os.makedirs(path)
+
+def filter_pcap(scapy_cap=None, src_ip=None, src_port=None, dest_ip=None, dest_port=None):
+    filtered = []
+    if scapy_cap:
+        for pkt in scapy_cap:
+            if IP in pkt and TCP in pkt:
+                if pkt.len > 1400:
+                    if src_ip:
+                        if pkt[IP].src != src_ip:
+                            continue
+                    if src_port:
+                        if pkt[TCP].sport != src_port:
+                            continue
+                    if dest_ip:
+                        if pkt[IP].dst != dest_ip:
+                            continue
+                    if dest_port:
+                        if pkt[TCP].dport != dest_port:
+                            continue
+                    filtered.append(pkt)
+    print('After Filtering No. of Packets: ', len(filtered))
+    return filtered
+
+def get_pkt_seq_n_time(scapy_cap):
+    seq_time_list = []
+    for pkt in scapy_cap:
+        seq_time_list.append((pkt[TCP].seq, float(pkt.time)))
+    print('Fetched sequence number and timestamp')
+    return seq_time_list
+
+def save_pcap(pkt_list, name):
+    print('Saved as pcap: ', name)
+    wrpcap(name, pkt_list)
+
+def most_common(lst):
+    return max(set(lst), key=lst.count)
+
+def remove_NaN(pandas_df):
+    for column in pandas_df.columns:
+        pandas_df = pandas_df[pandas_df[column].notna()]
+    print('Removed NaN from columns.')
+    return pandas_df
+
+def get_src_port(scapy_cap):
+    src_port_list = []
+    for pkt in scapy_cap:
+        if TCP in pkt:
+            src_port_list.append(pkt[TCP].sport)
+    src_port = most_common(src_port_list)
+    print('Source Port is: ', src_port)
+    return src_port
+
+def draw_hist(data_arr, bins='auto', yscale='log', hist_filename='histogram.jpg'):    
+    plt_hist = plt
+    plt_hist.yscale(yscale)
+    plt_hist.xlabel('Delay(ms) (min: ' + str(round(data_arr.min(), 3)) + 'ms max: ' + str(round(data_arr.max(), 3)) + 'ms)')
+    plt_hist.ylabel('Frequency')
+    plt_hist.hist(data_arr, bins=bins)
+    plt_hist.savefig(hist_filename)
+    plt_hist.close()
+    if data_arr.max() > 20:
+        plt_hist = plt
+        plt_hist.yscale(yscale)
+        plt_hist.xlabel('Delay(ms)')
+        plt_hist.ylabel('Frequency')
+        plt_hist.xlim([data_arr.min(), data_arr.min() + 10])
+        plt_hist.hist(data_arr, bins=bins)
+        plt_hist.savefig(hist_filename.split('.')[0] + '_zoomed.' + hist_filename.split('.')[1])
+        plt_hist.close()
+
+def draw_pdf(data_arr, color='red', label='PDF', bins='auto', pdf_filename='pdf.jpg'):
+    count, bins_count = np.histogram(data_arr, bins=bins)
+    pdf = count / sum(count)
+    plt_pdf = plt
+    plt_pdf.xlabel('Delay(ms)')
+    plt_pdf.ylabel('PDF')
+    plt_pdf.axhline(y = 0.0, color = 'r', linestyle = '-')
+    plt_pdf.plot(bins_count[1:], pdf, color=color, label='PDF')
+    plt_pdf.savefig(pdf_filename)
+    plt_pdf.close()
+    if data_arr.max() > 20:
+        plt_pdf = plt
+        plt_pdf.xlabel('Delay(ms)')
+        plt_pdf.ylabel('PDF')
+        plt_pdf.xlim([data_arr.min(), data_arr.min() + 10])
+        plt_pdf.axhline(y = 0.0, color = 'r', linestyle = '-')
+        plt_pdf.plot(bins_count[1:], pdf, color=color, label='PDF')
+        plt_pdf.savefig(pdf_filename.split('.')[0] + '_zoomed.' + pdf_filename.split('.')[1])
+        plt_pdf.close()
+
+def draw_cdf(data_arr, color='blue', label='CDF', bins='auto', cdf_filename='cdf.jpg'):
+    count, bins_count = np.histogram(data_arr, bins=bins)
+    pdf = count / sum(count)
+    cdf = np.cumsum(pdf)
+    plt_cdf = plt
+    plt_cdf.xlabel('Delay(ms)')
+    plt_cdf.ylabel('CDF')
+    plt_cdf.axhline(y = 1.0, color = 'r', linestyle = '-')
+    plt_cdf.plot(bins_count[1:], cdf, color=color, label="CDF")
+    plt_cdf.savefig(cdf_filename)
+    plt_cdf.close()
+    if data_arr.max() > 20:
+        plt_cdf = plt
+        plt_cdf.xlabel('Delay(ms)')
+        plt_cdf.ylabel('CDF')
+        plt_cdf.xlim([data_arr.min(), data_arr.min() + 10])
+        plt_cdf.axhline(y = 1.0, color = 'r', linestyle = '-')
+        plt_cdf.plot(bins_count[1:], cdf, color=color, label="CDF")
+        plt_cdf.savefig(cdf_filename.split('.')[0] + '_zoomed.' + cdf_filename.split('.')[1])
+        plt_cdf.close()
+
+def draw_pdf_cdf(data_arr, pdf_color='red', cdf_color='blue', bins='auto', pdf_cdf_filename='pdf_cdf.jpg'):
+    count, bins_count = np.histogram(data_arr, bins=bins)
+    pdf = count / sum(count)
+    cdf = np.cumsum(pdf)
+    fig, ax1 = plt.subplots()
+    color = 'tab:red'
+    ax1.set_xlabel('Delay(ms)')
+    ax1.set_ylabel('PDF')
+    ax1.plot(bins_count[1:], pdf, color=pdf_color, label='PDF')
+    ax1.tick_params(axis='y', labelcolor=color)
+    ax2 = ax1.twinx()
+    color = 'tab:blue'
+    ax2.set_ylabel('CDF')
+    ax2.plot(bins_count[1:], cdf, color=cdf_color, label='CDF')
+    ax2.tick_params(axis='y', labelcolor=color)
+    plt.savefig(pdf_cdf_filename)
+    plt.close()
+    if data_arr.max() > 20:
+        fig, ax1 = plt.subplots()
+        color = 'tab:red'
+        plt.xlim([data_arr.min(), data_arr.min() + 10])
+        ax1.set_xlabel('Delay(ms)')
+        ax1.set_ylabel('PDF')
+        ax1.plot(bins_count[1:], pdf, color=pdf_color, label='PDF')
+        ax1.tick_params(axis='y', labelcolor=color)
+        ax2 = ax1.twinx()
+        color = 'tab:blue'
+        ax2.set_ylabel('CDF')
+        ax2.plot(bins_count[1:], cdf, color=cdf_color, label='CDF')
+        ax2.tick_params(axis='y', labelcolor=color)
+        plt.savefig(pdf_cdf_filename.split('.')[0] + '_zoomed.' + pdf_cdf_filename.split('.')[1])
+        plt.close()