Code/Projects/SmartPlugDetector/src/main/java/edu/uci/iotproject/Main.java

   1 package edu.uci.iotproject;
   2
   3 import static edu.uci.iotproject.analysis.UserAction.Type;
   4
   5 import edu.uci.iotproject.analysis.*;
   6 import edu.uci.iotproject.comparison.seqalignment.ExtractedSequence;
   7 import edu.uci.iotproject.comparison.seqalignment.SequenceAlignment;
   8 import edu.uci.iotproject.comparison.seqalignment.SequenceExtraction;
   9 import edu.uci.iotproject.io.TriggerTimesFileReader;
  10 import edu.uci.iotproject.util.PcapPacketUtils;
  11 import edu.uci.iotproject.util.PrintUtils;
  12 import org.apache.commons.math3.stat.clustering.Cluster;
  13 import org.apache.commons.math3.stat.clustering.DBSCANClusterer;
  14 import org.pcap4j.core.*;
  15 import org.pcap4j.packet.namednumber.DataLinkType;
  16
  17 import java.io.EOFException;
  18 import java.io.File;
  19 import java.io.PrintWriter;
  20 import java.net.UnknownHostException;
  21 import java.time.Instant;
  22 import java.util.*;
  23 import java.util.concurrent.TimeoutException;
  24 import java.util.stream.Collectors;
  25 import java.util.stream.Stream;
  26
  27 /**
  28  * This is a system that reads PCAP files to compare
  29  * patterns of DNS hostnames, packet sequences, and packet
  30  * lengths with training data to determine certain events
  31  * or actions for smart home devices.
  32  *
  33  * @author Janus Varmarken
  34  * @author Rahmadi Trimananda (rtrimana@uci.edu)
  35  * @version 0.1
  36  */
  37 public class Main {
  38
  39
  40     public static void main(String[] args) throws PcapNativeException, NotOpenException, EOFException, TimeoutException, UnknownHostException {
  41         // -------------------------------------------------------------------------------------------------------------
  42         // ------------ # Code for extracting traffic generated by a device within x seconds of a trigger # ------------
  43         // Paths to input and output files (consider supplying these as arguments instead) and IP of the device for
  44         // which traffic is to be extracted:
  45         String path = "/scratch/July-2018"; // Rahmadi
  46 //        String path = "/Users/varmarken/temp/UCI IoT Project/experiments"; // Janus
  47         boolean verbose = true;
  48         final String onPairsPath = "/scratch/July-2018/on.txt";
  49         final String offPairsPath = "/scratch/July-2018/off.txt";
  50
  51         // 1) D-Link July 26 experiment
  52 //        final String inputPcapFile = path + "/2018-07/dlink/dlink.wlan1.local.pcap";
  53 //        final String outputPcapFile = path + "/2018-07/dlink/dlink-processed.pcap";
  54 //        final String triggerTimesFile = path + "/2018-07/dlink/dlink-july-26-2018.timestamps";
  55 //        final String deviceIp = "192.168.1.199"; // .246 == phone; .199 == dlink plug?
  56           // Actual training
  57 //        final String inputPcapFile = path + "/2018-10/dlink-plug/dlink-plug.wlan1.local.pcap";
  58 //        final String outputPcapFile = path + "/2018-10/dlink-plug/dlink-plug-processed.pcap";
  59 //        final String triggerTimesFile = path + "/2018-10/dlink-plug/dlink-plug-oct-17-2018.timestamps";
  60 //        final String deviceIp = "192.168.1.199"; // .246 == phone; .199 == dlink plug?
  61         // TODO: EXPERIMENT - November 7, 2018
  62 //        final String inputPcapFile = path + "/experimental_result/standalone/dlink-plug/wlan1/dlink-plug.wlan1.local.pcap";
  63 //        final String outputPcapFile = path + "/experimental_result/standalone/dlink-plug/wlan1/dlink-plug-processed.pcap";
  64 //        final String triggerTimesFile = path + "/experimental_result/standalone/dlink-plug/timestamps/dlink-plug-nov-7-2018.timestamps";
  65 //        //final String deviceIp = "192.168.1.199"; // .246 == phone; .199 == dlink plug?
  66 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .199 == dlink plug?
  67
  68         // 2) TP-Link July 25 experiment
  69 //        final String inputPcapFile = path + "/2018-07/tplink/tplink.wlan1.local.pcap";
  70 //        final String outputPcapFile = path + "/2018-07/tplink/tplink-processed.pcap";
  71 //        final String triggerTimesFile = path + "/2018-07/tplink/tplink-july-25-2018.timestamps";
  72 //        final String deviceIp = "192.168.1.159";
  73           // Actual training
  74 //        final String inputPcapFile = path + "/2018-10/tplink-plug/tplink-plug.wlan1.local.pcap";
  75 //        final String outputPcapFile = path + "/2018-10/tplink-plug/tplink-plug-processed.pcap";
  76 //        final String triggerTimesFile = path + "/2018-10/tplink-plug/tplink-plug-oct-17-2018.timestamps";
  77 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .159 == tplink plug
  78         // TODO: EXPERIMENT - November 8, 2018
  79 //        final String inputPcapFile = path + "/experimental_result/standalone/tplink-plug/wlan1/tplink-plug.wlan1.local.pcap";
  80 //        final String outputPcapFile = path + "/experimental_result/standalone/tplink-plug/wlan1/tplink-plug-processed.pcap";
  81 //        final String triggerTimesFile = path + "/experimental_result/standalone/tplink-plug/timestamps/tplink-plug-nov-8-2018.timestamps";
  82 //        final String deviceIp = "192.168.1.159"; // .246 == phone; .159 == tplink plug
  83 ////        final String deviceIp = "192.168.1.246"; // .246 == phone; .159 == tplink plug
  84
  85         // 2b) TP-Link July 25 experiment TRUNCATED:
  86         // Only contains "true local" events, i.e., before the behavior changes to remote-like behavior.
  87         // Last included event is at July 25 10:38:11; file filtered to only include packets with arrival time <= 10:38:27.
  88 //        final String inputPcapFile = path + "/2018-07/tplink/tplink.wlan1.local.truncated.pcap";
  89 //        final String outputPcapFile = path + "/2018-07/tplink/tplink-processed.truncated.pcap";
  90 //        final String triggerTimesFile = path + "/2018-07/tplink/tplink-july-25-2018.truncated.timestamps";
  91 //        final String deviceIp = "192.168.1.159";
  92
  93         // 3) SmartThings Plug July 25 experiment
  94 //        final String inputPcapFile = path + "/2018-07/stplug/stplug.wlan1.local.pcap";
  95 //        final String outputPcapFile = path + "/2018-07/stplug/stplug-processed.pcap";
  96 //        final String triggerTimesFile = path + "/2018-07/stplug/smartthings-july-25-2018.timestamps";
  97 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
  98         // October 18
  99 //        final String inputPcapFile = path + "/2018-10/st-plug/st-plug.wlan1.local.pcap";
 100 //        final String outputPcapFile = path + "/2018-10/st-plug/st-plug-processed.pcap";
 101 //        final String triggerTimesFile = path + "/2018-10/st-plug/st-plug-oct-18-2018.timestamps";
 102 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
 103         // TODO: EXPERIMENT - November 12, 2018
 104 //        final String inputPcapFile = path + "/experimental_result/standalone/st-plug/wlan1/st-plug.wlan1.local.pcap";
 105 //        final String outputPcapFile = path + "/experimental_result/standalone/st-plug/wlan1/st-plug-processed.pcap";
 106 ////        final String inputPcapFile = path + "/experimental_result/standalone/st-plug/eth1/st-plug.eth1.local.pcap";
 107 ////        final String outputPcapFile = path + "/experimental_result/standalone/st-plug/eth1/st-plug-processed.pcap";
 108 //        final String triggerTimesFile = path + "/experimental_result/standalone/st-plug/timestamps/st-plug-nov-12-2018.timestamps";
 109 ////        final String deviceIp = "192.168.1.142"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
 110 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
 111
 112         // 4) Wemo July 30 experiment
 113 //        final String inputPcapFile = path + "/2018-07/wemo/wemo.wlan1.local.pcap";
 114 //        final String outputPcapFile = path + "/2018-07/wemo/wemo-processed.pcap";
 115 //        final String triggerTimesFile = path + "/2018-07/wemo/wemo-july-30-2018.timestamps";
 116 //        final String deviceIp = "192.168.1.145";  // .246 == phone; .145 == WeMo
 117
 118         // 5) Wemo Insight July 31 experiment
 119 //        final String inputPcapFile = path + "/2018-07/wemoinsight/wemoinsight.wlan1.local.pcap";
 120 //        final String outputPcapFile = path + "/2018-07/wemoinsight/wemoinsight-processed.pcap";
 121 //        final String triggerTimesFile = path + "/2018-07/wemoinsight/wemo-insight-july-31-2018.timestamps";
 122 //        final String deviceIp = "192.168.1.135";
 123
 124         // 6) TP-Link Bulb August 1 experiment
 125 //        final String inputPcapFile = path + "/2018-08/tplink-bulb/tplinkbulb.wlan1.local.pcap";
 126 //        final String outputPcapFile = path + "/2018-08/tplink-bulb/tplinkbulb-processed.pcap";
 127 //        final String triggerTimesFile = path + "/2018-08/tplink-bulb/tplink-bulb-aug-3-2018.timestamps";
 128 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .140 == TP-Link bulb
 129         // TODO: EXPERIMENT - November 16, 2018
 130         final String inputPcapFile = path + "/experimental_result/standalone/tplink-bulb/wlan1/tplink-bulb.wlan1.local.pcap";
 131         final String outputPcapFile = path + "/experimental_result/standalone/tplink-bulb/wlan1/tplink-bulb-processed.pcap";
 132 //        final String inputPcapFile = path + "/experimental_result/standalone/tplink-bulb/eth0/tplink-bulb.eth1.local.pcap";
 133 //        final String outputPcapFile = path + "/experimental_result/standalone/tplink-bulb/eth0/tplink-bulb-processed.pcap";
 134         final String triggerTimesFile = path + "/experimental_result/standalone/tplink-bulb/timestamps/tplink-bulb-nov-16-2018.timestamps";
 135 //        final String deviceIp = "192.168.1.140"; // .246 == phone; .140 == TP-Link bulb
 136         final String deviceIp = "192.168.1.246"; // .246 == phone; .140 == TP-Link bulb
 137
 138         // 7) Kwikset Doorlock August 6 experiment
 139 //        final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock.data.wlan1.pcap";
 140 ////        final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock.wlan1.local.pcap";
 141 //        final String outputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-processed.pcap";
 142 ////        final String triggerTimesFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-aug-6-2018.timestamps";
 143 //        final String triggerTimesFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-8hr-data-oct-11-2018.timestamps";
 144 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
 145         // TODO: EXPERIMENT - November 10, 2018
 146 //        final String inputPcapFile = path + "/experimental_result/standalone/kwikset-doorlock/wlan1/kwikset-doorlock.wlan1.local.pcap";
 147 //        final String outputPcapFile = path + "/experimental_result/standalone/kwikset-doorlock/wlan1/kwikset-doorlock-processed.pcap";
 148 ////        final String inputPcapFile = path + "/experimental_result/standalone/kwikset-doorlock/eth1/kwikset-doorlock.eth1.local.pcap";
 149 ////        final String outputPcapFile = path + "/experimental_result/standalone/kwikset-doorlock/eth1/kwikset-doorlock-processed.pcap";
 150 //        final String triggerTimesFile = path + "/experimental_result/standalone/kwikset-doorlock/timestamps/kwikset-doorlock-nov-10-2018.timestamps";
 151 ////        final String deviceIp = "192.168.1.142"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
 152 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
 153
 154         // September 12, 2018 - includes both wlan1 and eth1 interfaces
 155 //        final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset3.wlan1.local.pcap";
 156 //        //final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset3.eth1.local.pcap";
 157 //        final String outputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset3-processed.pcap";
 158 //        final String triggerTimesFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-sept-12-2018.timestamps";
 159 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
 160
 161         // 8) Hue Bulb August 7 experiment
 162 //        final String inputPcapFile = path + "/2018-08/hue-bulb/hue-bulb.wlan1.local.pcap";
 163 //        final String outputPcapFile = path + "/2018-08/hue-bulb/hue-bulb-processed.pcap";
 164 //        final String triggerTimesFile = path + "/2018-08/hue-bulb/hue-bulb-aug-7-2018.timestamps";
 165 //        final String deviceIp = "192.168.1.246";
 166         // October 30 experiment
 167 //        final String inputPcapFile = path + "/2018-10/hue-bulb/hue-bulb.wlan1.local.pcap";
 168 //        final String outputPcapFile = path + "/2018-10/hue-bulb/hue-bulb-processed.pcap";
 169 //        final String triggerTimesFile = path + "/2018-10/hue-bulb/hue-bulb-oct-30-2018.timestamps";
 170 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .100 == Hue hub
 171
 172         // 9) Lifx Bulb August 8 experiment
 173 //        final String inputPcapFile = path + "/2018-08/lifx-bulb/lifx-bulb.wlan1.local.pcap";
 174 //        final String outputPcapFile = path + "/2018-08/lifx-bulb/lifx-bulb-processed.pcap";
 175 //        final String triggerTimesFile = path + "/2018-08/lifx-bulb/lifx-bulb-aug-8-2018.timestamps";
 176 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .231 == Lifx
 177         // October 18
 178 //        final String inputPcapFile = path + "/2018-10/lifx-bulb/lifx-bulb.wlan1.local.pcap";
 179 //        final String outputPcapFile = path + "/2018-10/lifx-bulb/lifx-bulb-processed.pcap";
 180 //        final String triggerTimesFile = path + "/2018-10/lifx-bulb/lifx-bulb-oct-18-2018.timestamps";
 181 //        final String deviceIp = "192.168.1.231"; // .246 == phone; .231 == Lifx
 182         // November 1
 183 //        final String inputPcapFile = path + "/2018-10/lifx-bulb/lifx-bulb.wlan1.local.pcap";
 184 //        final String outputPcapFile = path + "/2018-10/lifx-bulb/lifx-bulb-processed.pcap";
 185 //        final String triggerTimesFile = path + "/2018-10/lifx-bulb/lifx-bulb-nov-1-2018.timestamps";
 186 //        final String deviceIp = "192.168.1.231"; // .246 == phone; .231 == Lifx
 187
 188         // 10) Amcrest Camera August 9 experiment
 189 //        final String inputPcapFile = path + "/2018-08/amcrest-camera/amcrest-camera.wlan1.local.pcap";
 190 //        final String outputPcapFile = path + "/2018-08/amcrest-camera/amcrest-camera-processed.pcap";
 191 //        final String triggerTimesFile = path + "/2018-08/amcrest-camera/amcrest-camera-aug-9-2018.timestamps";
 192 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .235 == camera
 193
 194         // 11) Arlo Camera August 10 experiment
 195 //        final String inputPcapFile = path + "/2018-08/arlo-camera/arlo-camera.wlan1.local.pcap";
 196 //        final String outputPcapFile = path + "/2018-08/arlo-camera/arlo-camera-processed.pcap";
 197 //        final String triggerTimesFile = path + "/2018-08/arlo-camera/arlo-camera-aug-10-2018.timestamps";
 198 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .140 == camera
 199         // TODO: EXPERIMENT - November 13, 2018
 200 //        final String inputPcapFile = path + "/experimental_result/standalone/arlo-camera/wlan1/arlo-camera.wlan1.local.pcap";
 201 //        final String outputPcapFile = path + "/experimental_result/standalone/arlo-camera/wlan1/arlo-camera-processed.pcap";
 202 ////        final String inputPcapFile = path + "/experimental_result/standalone/arlo-camera/eth0/arlo-camera.eth1.local.pcap";
 203 ////        final String outputPcapFile = path + "/experimental_result/standalone/arlo-camera/eth0/arlo-camera-processed.pcap";
 204 //        final String triggerTimesFile = path + "/experimental_result/standalone/arlo-camera/timestamps/arlo-camera-nov-13-2018.timestamps";
 205 ////        final String deviceIp = "192.168.1.140"; // .246 == phone; .140 == camera
 206 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .140 == camera
 207
 208         // 12) Blossom sprinkler August 13 experiment
 209 //        final String inputPcapFile = path + "/2018-08/blossom/blossom.wlan1.local.pcap";
 210 //        final String outputPcapFile = path + "/2018-08/blossom/blossom-processed.pcap";
 211 //        final String triggerTimesFile = path + "/2018-08/blossom/blossom-aug-13-2018.timestamps";
 212 //        final String deviceIp = "192.168.1.229"; // .246 == phone; .229 == sprinkler
 213         // 2 November
 214 //        final String inputPcapFile = path + "/2018-10/blossom-sprinkler/blossom-sprinkler.wlan1.local.pcap";
 215 //        final String outputPcapFile = path + "/2018-10/blossom-sprinkler/blossom-sprinkler-processed.pcap";
 216 //        final String triggerTimesFile = path + "/2018-10/blossom-sprinkler/blossom-sprinkler-nov-2-2018.timestamps";
 217 //        final String deviceIp = "192.168.1.229"; // .246 == phone; .229 == sprinkler
 218
 219 //        // 13) DLink siren August 14 experiment
 220 //        final String inputPcapFile = path + "/2018-08/dlink-siren/dlink-siren.wlan1.local.pcap";
 221 //        //final String inputPcapFile = path + "/evaluation/dlink-siren/dlink-siren.data.wlan1.pcap";
 222 //        final String outputPcapFile = path + "/2018-08/dlink-siren/dlink-siren-processed.pcap";
 223 //        final String triggerTimesFile = path + "/2018-08/dlink-siren/dlink-siren-oct-12-2018.timestamps";
 224 //        //final String triggerTimesFile = path + "/2018-08/dlink-siren/dlink-siren-aug-14-2018.timestamps";
 225 //        //final String triggerTimesFile = path + "/actual/timestamps/dlink-siren-8hr-data-oct-10-2018.timestamps";
 226 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .183 == siren
 227         // TODO: EXPERIMENT - November 9, 2018
 228 //        final String inputPcapFile = path + "/experimental_result/standalone/dlink-siren/wlan1/dlink-siren.wlan1.local.pcap";
 229 //        final String outputPcapFile = path + "/experimental_result/standalone/dlink-siren/wlan1/dlink-siren-processed.pcap";
 230 //        final String triggerTimesFile = path + "/experimental_result/standalone/dlink-siren/timestamps/dlink-siren-nov-9-2018.timestamps";
 231 //        final String deviceIp = "192.168.1.183"; // .246 == phone; .183 == siren
 232 ////        final String deviceIp = "192.168.1.246"; // .246 == phone; .183 == siren
 233
 234         // 14) Nest thermostat August 15 experiment
 235 //        final String inputPcapFile = path + "/2018-08/nest/nest.wlan1.local.pcap";
 236 //        final String outputPcapFile = path + "/2018-08/nest/nest-processed.pcap";
 237 //        final String triggerTimesFile = path + "/2018-08/nest/nest-aug-15-2018.timestamps";
 238 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .127 == Nest thermostat
 239         // TODO: EXPERIMENT - November 14, 2018
 240 //        final String inputPcapFile = path + "/experimental_result/standalone/nest-thermostat/wlan1/nest-thermostat.wlan1.local.pcap";
 241 //        final String outputPcapFile = path + "/experimental_result/standalone/nest-thermostat/wlan1/nest-thermostat-processed.pcap";
 242 ////        final String inputPcapFile = path + "/experimental_result/standalone/nest-thermostat/eth0/nest-thermostat.eth1.local.pcap";
 243 ////        final String outputPcapFile = path + "/experimental_result/standalone/nest-thermostat/eth0/nest-thermostat-processed.pcap";
 244 //        final String triggerTimesFile = path + "/experimental_result/standalone/nest-thermostat/timestamps/nest-thermostat-nov-15-2018.timestamps";
 245 ////        final String deviceIp = "192.168.1.127"; // .246 == phone; .127 == Nest thermostat
 246 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .127 == Nest thermostat
 247
 248         // 15) Alexa August 16 experiment
 249 //        final String inputPcapFile = path + "/2018-08/alexa/alexa.wlan1.local.pcap";
 250 //        final String outputPcapFile = path + "/2018-08/alexa/alexa-processed.pcap";
 251 //        final String triggerTimesFile = path + "/2018-08/alexa/alexa-aug-16-2018.timestamps";
 252 //        final String deviceIp = "192.168.1.225"; // .246 == phone; .225 == Alexa
 253         // August 17
 254 //        final String inputPcapFile = path + "/2018-08/alexa/alexa2.wlan1.local.pcap";
 255 //        final String outputPcapFile = path + "/2018-08/alexa/alexa2-processed.pcap";
 256 //        final String triggerTimesFile = path + "/2018-08/alexa/alexa-aug-17-2018.timestamps";
 257 //        final String deviceIp = "192.168.1.225"; // .246 == phone; .225 == Alexa
 258
 259         // September 17
 260 //        final String inputPcapFile = path + "/2018-08/noise/noise.eth1.pcap";
 261 //        final String outputPcapFile = path + "/2018-08/noise/noise-processed.pcap";
 262 //        final String triggerTimesFile = path + "/2018-08/noise/noise-sept-17-2018.timestamps";
 263 //        final String deviceIp = "192.168.1.142"; //  .142 == SmartThings Hub; .199 == dlink plug; .183 == siren
 264         // September 26 - D-Link noise
 265 //        final String inputPcapFile = path + "/2018-08/noise/noise.dlink.wlan1.pcap";
 266 //        final String outputPcapFile = path + "/2018-08/noise/noise-processed.pcap";
 267 //        final String triggerTimesFile = path + "/2018-08/noise/dlink-noise-sept-26-2018.timestamps";
 268 //        final String deviceIp = "192.168.1.183"; //  .199 == dlink plug; .183 == siren
 269         // September 27 - Kwikset noise
 270 //        final String inputPcapFile = path + "/2018-08/noise/noise.kwikset.eth1.pcap";
 271 //        final String outputPcapFile = path + "/2018-08/noise/noise-processed.pcap";
 272 //        final String triggerTimesFile = path + "/2018-08/noise/kwikset-doorlock-noise-sept-27-2018.timestamps";
 273 //        final String deviceIp = "192.168.1.142"; //  .142 == SmartThings Hub;
 274
 275         TriggerTimesFileReader ttfr = new TriggerTimesFileReader();
 276         List<Instant> triggerTimes = ttfr.readTriggerTimes(triggerTimesFile, false);
 277         // Tag each trigger with "ON" or "OFF", assuming that the first trigger is an "ON" and that they alternate.
 278         List<UserAction> userActions = new ArrayList<>();
 279         for (int i = 0; i < triggerTimes.size(); i++) {
 280             userActions.add(new UserAction(i % 2 == 0 ? Type.TOGGLE_ON : Type.TOGGLE_OFF, triggerTimes.get(i)));
 281         }
 282         TriggerTrafficExtractor tte = new TriggerTrafficExtractor(inputPcapFile, triggerTimes, deviceIp);
 283         final PcapDumper outputter = Pcaps.openDead(DataLinkType.EN10MB, 65536).dumpOpen(outputPcapFile);
 284         DnsMap dnsMap = new DnsMap();
 285         TcpReassembler tcpReassembler = new TcpReassembler();
 286         TrafficLabeler trafficLabeler = new TrafficLabeler(userActions);
 287         tte.performExtraction(pkt -> {
 288             try {
 289                 outputter.dump(pkt);
 290             } catch (NotOpenException e) {
 291                 e.printStackTrace();
 292             }
 293         }, dnsMap, tcpReassembler, trafficLabeler);
 294         outputter.flush();
 295         outputter.close();
 296
 297         if (tte.getPacketsIncludedCount() != trafficLabeler.getTotalPacketCount()) {
 298             // Sanity/debug check
 299             throw new AssertionError(String.format("mismatch between packet count in %s and %s",
 300                     TriggerTrafficExtractor.class.getSimpleName(), TrafficLabeler.class.getSimpleName()));
 301         }
 302
 303         // Extract all conversations present in the filtered trace.
 304         List<Conversation> allConversations = tcpReassembler.getTcpConversations();
 305         // Group conversations by hostname.
 306         Map<String, List<Conversation>> convsByHostname = TcpConversationUtils.groupConversationsByHostname(allConversations, dnsMap);
 307         System.out.println("Grouped conversations by hostname.");
 308         // For each hostname, count the frequencies of packet lengths exchanged with that hostname.
 309         final Map<String, Map<Integer, Integer>> pktLenFreqsByHostname = new HashMap<>();
 310         convsByHostname.forEach((host, convs) -> pktLenFreqsByHostname.put(host, TcpConversationUtils.countPacketLengthFrequencies(convs)));
 311         System.out.println("Counted frequencies of packet lengths exchanged with each hostname.");
 312         // For each hostname, count the frequencies of packet sequences (i.e., count how many conversations exchange a
 313         // sequence of packets of some specific lengths).
 314         final Map<String, Map<String, Integer>> pktSeqFreqsByHostname = new HashMap<>();
 315         convsByHostname.forEach((host, convs) -> pktSeqFreqsByHostname.put(host, TcpConversationUtils.countPacketSequenceFrequencies(convs)));
 316         System.out.println("Counted frequencies of packet sequences exchanged with each hostname.");
 317         // For each hostname, count frequencies of packet pairs exchanged with that hostname across all conversations
 318         final Map<String, Map<String, Integer>> pktPairFreqsByHostname =
 319                 TcpConversationUtils.countPacketPairFrequenciesByHostname(allConversations, dnsMap);
 320         System.out.println("Counted frequencies of packet pairs per hostname");
 321         // For each user action, reassemble the set of TCP connections occurring shortly after
 322         final Map<UserAction, List<Conversation>> userActionToConversations = trafficLabeler.getLabeledReassembledTcpTraffic();
 323         final Map<UserAction, Map<String, List<Conversation>>> userActionsToConvsByHostname = trafficLabeler.getLabeledReassembledTcpTraffic(dnsMap);
 324         System.out.println("Reassembled TCP conversations occurring shortly after each user event");
 325
 326
 327
 328         /*
 329          * NOTE: no need to generate these more complex on/off maps that also contain mappings from hostname and
 330          * sequence identifiers as we do not care about hostnames and sequences during clustering.
 331          * We can simply use the UserAction->List<Conversation> map to generate ON/OFF groupings of conversations.
 332          */
 333         /*
 334         // Contains all ON events: hostname -> sequence identifier -> list of conversations with that sequence
 335         Map<String, Map<String, List<Conversation>>> ons = new HashMap<>();
 336         // Contains all OFF events: hostname -> sequence identifier -> list of conversations with that sequence
 337         Map<String, Map<String, List<Conversation>>> offs = new HashMap<>();
 338         userActionsToConvsByHostname.forEach((ua, hostnameToConvs) -> {
 339             Map<String, Map<String, List<Conversation>>> outer = ua.getType() == Type.TOGGLE_ON ? ons : offs;
 340             hostnameToConvs.forEach((host, convs) -> {
 341                 Map<String, List<Conversation>> seqsToConvs = TcpConversationUtils.
 342                         groupConversationsByPacketSequence(convs, verbose);
 343                 outer.merge(host, seqsToConvs, (oldMap, newMap) -> {
 344                     newMap.forEach((sequence, cs) -> oldMap.merge(sequence, cs, (list1, list2) -> {
 345                         list1.addAll(list2);
 346                         return list1;
 347                     }));
 348                     return oldMap;
 349                 });
 350             });
 351         });
 352         */
 353
 354         // ================================================ CLUSTERING ================================================
 355         // Note: no need to use the more convoluted on/off maps; can simply use the UserAction->List<Conversation> map
 356         // when don't care about hostnames and sequences (see comment earlier).
 357         List<Conversation> onConversations = userActionToConversations.entrySet().stream().
 358                 filter(e -> e.getKey().getType() == Type.TOGGLE_ON). // drop all OFF events from stream
 359                 map(e -> e.getValue()). // no longer interested in the UserActions
 360                 flatMap(List::stream). // flatten List<List<T>> to a List<T>
 361                 collect(Collectors.toList());
 362         List<Conversation> offConversations = userActionToConversations.entrySet().stream().
 363                 filter(e -> e.getKey().getType() == Type.TOGGLE_OFF).
 364                 map(e -> e.getValue()).
 365                 flatMap(List::stream).
 366                 collect(Collectors.toList());
 367         //Collections.sort(onConversations, (c1, c2) -> c1.getPackets().)
 368
 369         List<PcapPacketPair> onPairs = onConversations.stream().
 370                 map(c -> c.isTls() ? TcpConversationUtils.extractTlsAppDataPacketPairs(c) :
 371                         TcpConversationUtils.extractPacketPairs(c)).
 372                 flatMap(List::stream). // flatten List<List<>> to List<>
 373                 collect(Collectors.toList());
 374         List<PcapPacketPair> offPairs = offConversations.stream().
 375                 map(c -> c.isTls() ? TcpConversationUtils.extractTlsAppDataPacketPairs(c) :
 376                         TcpConversationUtils.extractPacketPairs(c)).
 377                 flatMap(List::stream). // flatten List<List<>> to List<>
 378                 collect(Collectors.toList());
 379         // Note: need to update the DnsMap of all PcapPacketPairs if we want to use the IP/hostname-sensitive distance.
 380         Stream.concat(Stream.of(onPairs), Stream.of(offPairs)).flatMap(List::stream).forEach(p -> p.setDnsMap(dnsMap));
 381         // Perform clustering on conversation logged as part of all ON events.
 382         DBSCANClusterer<PcapPacketPair> onClusterer = new DBSCANClusterer<>(10.0, 45);
 383         //DBSCANClusterer<PcapPacketPair> onClusterer = new DBSCANClusterer<>(10.0, 10);
 384         List<Cluster<PcapPacketPair>> onClusters = onClusterer.cluster(onPairs);
 385         // Perform clustering on conversation logged as part of all OFF events.
 386         DBSCANClusterer<PcapPacketPair> offClusterer = new DBSCANClusterer<>(10.0, 45);
 387         //DBSCANClusterer<PcapPacketPair> offClusterer = new DBSCANClusterer<>(10.0, 10);
 388         List<Cluster<PcapPacketPair>> offClusters = offClusterer.cluster(offPairs);
 389         // Sort the conversations as reference
 390         List<Conversation> sortedAllConversation = TcpConversationUtils.sortConversationList(allConversations);
 391         // Output clusters
 392         System.out.println("========================================");
 393         System.out.println("       Clustering results for ON        ");
 394         System.out.println("       Number of clusters: " + onClusters.size());
 395         int count = 0;
 396         List<List<List<PcapPacket>>> ppListOfListReadOn = new ArrayList<>();
 397         List<List<List<PcapPacket>>> ppListOfListListOn = new ArrayList<>();
 398         for (Cluster<PcapPacketPair> c : onClusters) {
 399             System.out.println(String.format("<<< Cluster #%02d (%03d points) >>>", ++count, c.getPoints().size()));
 400             System.out.print(PrintUtils.toSummaryString(c));
 401             if(c.getPoints().size() > 45 && c.getPoints().size() < 55) {
 402             //if(c.getPoints().size() > 25) {
 403                 // Print to file
 404                 List<List<PcapPacket>> ppListOfList = PcapPacketUtils.clusterToListOfPcapPackets(c);
 405                 ppListOfListListOn.add(ppListOfList);
 406             }
 407         }
 408         // TODO: Merging test
 409         ppListOfListListOn = PcapPacketUtils.mergeSignatures(ppListOfListListOn, sortedAllConversation);
 410         ppListOfListListOn = PcapPacketUtils.sortSignatures(ppListOfListListOn);
 411         // TODO: Need to remove sequence number 2 for ST plug since it is not a good signature!
 412         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 2);
 413         // TODO: Need to remove sequence number 0 for Arlo Camera since it is not a good signature!
 414         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 0);
 415         // TODO: Need to remove sequence number 0 for TP-Link plug since it is not a good signature!
 416         // TODO: This sequence actually belongs to the local communication between the plug and the phone
 417         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 0);
 418         PcapPacketUtils.printSignatures(ppListOfListListOn);
 419         //count = 0;
 420         /*for (List<List<PcapPacket>> ll : ppListOfListListOn) {
 421             PrintUtils.serializeClustersIntoFile("./onSignature" + ++count + ".sig", ll);
 422             ppListOfListReadOn.add(PrintUtils.deserializeClustersFromFile("./onSignature" + count + ".sig"));
 423         }*/
 424         PrintUtils.serializeSignatureIntoFile("./onSignature.sig", ppListOfListListOn);
 425         ppListOfListReadOn = PrintUtils.deserializeSignatureFromFile("./onSignature.sig");
 426
 427         System.out.println("========================================");
 428         System.out.println("       Clustering results for OFF       ");
 429         System.out.println("       Number of clusters: " + offClusters.size());
 430         count = 0;
 431         List<List<List<PcapPacket>>> ppListOfListReadOff = new ArrayList<>();
 432         List<List<List<PcapPacket>>> ppListOfListListOff = new ArrayList<>();
 433         for (Cluster<PcapPacketPair> c : offClusters) {
 434             System.out.println(String.format("<<< Cluster #%03d (%06d points) >>>", ++count, c.getPoints().size()));
 435             System.out.print(PrintUtils.toSummaryString(c));
 436             if(c.getPoints().size() > 45 && c.getPoints().size() < 55) {
 437             //if(c.getPoints().size() > 25) {
 438                 // Print to file
 439                 List<List<PcapPacket>> ppListOfList = PcapPacketUtils.clusterToListOfPcapPackets(c);
 440                 ppListOfListListOff.add(ppListOfList);
 441             }
 442         }
 443         // TODO: Merging test
 444         ppListOfListListOff = PcapPacketUtils.mergeSignatures(ppListOfListListOff, sortedAllConversation);
 445         //ppListOfListListOff = PcapPacketUtils.sortSignatures(ppListOfListListOff);
 446         // TODO: Need to remove sequence number 1 for Nest Thermostat since it is not a good signature!
 447         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 1);
 448         // TODO: Need to remove sequence number 0 for Arlo Camera since it is not a good signature!
 449         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 1);
 450         // TODO: Need to remove sequence number 2 for ST plug since it is not a good signature!
 451         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 2);
 452         // TODO: Need to remove sequence number 0 for TP-Link plug since it is not a good signature!
 453         // TODO: This sequence actually belongs to the local communication between the plug and the phone
 454         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 0);
 455         PcapPacketUtils.printSignatures(ppListOfListListOff);
 456         //count = 0;
 457         /*for (List<List<PcapPacket>> ll : ppListOfListListOff) {
 458             PrintUtils.serializeClustersIntoFile("./offSignature" + ++count + ".sig", ll);
 459             ppListOfListReadOff.add(PrintUtils.deserializeClustersFromFile("./offSignature" + count + ".sig"));
 460         }*/
 461         PrintUtils.serializeSignatureIntoFile("./offSignature.sig", ppListOfListListOff);
 462         ppListOfListReadOff = PrintUtils.deserializeSignatureFromFile("./offSignature.sig");
 463         System.out.println("========================================");
 464         // ============================================================================================================
 465
 466         /*
 467         System.out.println("==== ON ====");
 468         // Print out all the pairs into a file for ON events
 469         File fileOnEvents = new File(onPairsPath);
 470         PrintWriter pwOn = null;
 471         try {
 472             pwOn = new PrintWriter(fileOnEvents);
 473         } catch(Exception ex) {
 474             ex.printStackTrace();
 475         }
 476         for(Map.Entry<String, Map<String, List<Conversation>>> entry : ons.entrySet()) {
 477             Map<String, List<Conversation>> seqsToConvs = entry.getValue();
 478             for(Map.Entry<String, List<Conversation>> entryConv : seqsToConvs.entrySet()) {
 479                 List<Conversation> listConv = entryConv.getValue();
 480                 // Just get the first Conversation because all Conversations in this group
 481                 // should have the same pairs of Application Data.
 482                 for(Conversation conv : listConv) {
 483                     // Process only if it is a TLS packet
 484                     if (conv.isTls()) {
 485                         List<PcapPacketPair> tlsAppDataList = TcpConversationUtils.extractTlsAppDataPacketPairs(conv);
 486                         for(PcapPacketPair pair: tlsAppDataList) {
 487                             System.out.println(PrintUtils.toCsv(pair, dnsMap));
 488                             pwOn.println(PrintUtils.toCsv(pair, dnsMap));
 489                         }
 490                     } else { // Non-TLS conversations
 491                         List<PcapPacketPair> packetList = TcpConversationUtils.extractPacketPairs(conv);
 492                         for(PcapPacketPair pair: packetList) {
 493                             System.out.println(PrintUtils.toCsv(pair, dnsMap));
 494                             pwOn.println(PrintUtils.toCsv(pair, dnsMap));
 495                         }
 496                     }
 497                 }
 498             }
 499         }
 500         pwOn.close();
 501
 502         System.out.println("==== OFF ====");
 503         // Print out all the pairs into a file for ON events
 504         File fileOffEvents = new File(offPairsPath);
 505         PrintWriter pwOff = null;
 506         try {
 507             pwOff = new PrintWriter(fileOffEvents);
 508         } catch(Exception ex) {
 509             ex.printStackTrace();
 510         }
 511         for(Map.Entry<String, Map<String, List<Conversation>>> entry : offs.entrySet()) {
 512             Map<String, List<Conversation>> seqsToConvs = entry.getValue();
 513             for(Map.Entry<String, List<Conversation>> entryConv : seqsToConvs.entrySet()) {
 514                 List<Conversation> listConv = entryConv.getValue();
 515                 // Just get the first Conversation because all Conversations in this group
 516                 // should have the same pairs of Application Data.
 517                 for(Conversation conv : listConv) {
 518                     // Process only if it is a TLS packet
 519                     if (conv.isTls()) {
 520                         List<PcapPacketPair> tlsAppDataList = TcpConversationUtils.extractTlsAppDataPacketPairs(conv);
 521                         for(PcapPacketPair pair: tlsAppDataList) {
 522                             System.out.println(PrintUtils.toCsv(pair, dnsMap));
 523                             pwOff.println(PrintUtils.toCsv(pair, dnsMap));
 524                         }
 525                     } else { // Non-TLS conversations
 526                         List<PcapPacketPair> packetList = TcpConversationUtils.extractPacketPairs(conv);
 527                         for (PcapPacketPair pair : packetList) {
 528                             System.out.println(PrintUtils.toCsv(pair, dnsMap));
 529                             pwOff.println(PrintUtils.toCsv(pair, dnsMap));
 530                         }
 531                     }
 532                 }
 533             }
 534         }
 535         pwOff.close();
 536         */
 537
 538 //        // ================================================================================================
 539 //        // <<< Some work-in-progress/explorative code that extracts a "representative" sequence >>>
 540 //        //
 541 //        // Currently need to know relevant hostname in advance :(
 542 //        String hostname = "events.tplinkra.com";
 543 ////        String hostname = "rfe-us-west-1.dch.dlink.com";
 544 //        // Conversations with 'hostname' for ON events.
 545 //        List<Conversation> onsForHostname = new ArrayList<>();
 546 //        // Conversations with 'hostname' for OFF events.
 547 //        List<Conversation> offsForHostname = new ArrayList<>();
 548 //        // "Unwrap" sequence groupings in ons/offs maps.
 549 //        ons.get(hostname).forEach((k,v) -> onsForHostname.addAll(v));
 550 //        offs.get(hostname).forEach((k,v) -> offsForHostname.addAll(v));
 551 //
 552 //
 553 //        Map<String, List<Conversation>> onsForHostnameGroupedByTlsAppDataSequence = TcpConversationUtils.groupConversationsByTlsApplicationDataPacketSequence(onsForHostname);
 554 //
 555 //
 556 //        // Extract representative sequence for ON and OFF by providing the list of conversations with
 557 //        // 'hostname' observed for each event type (the training data).
 558 //        SequenceExtraction seqExtraction = new SequenceExtraction();
 559 ////        ExtractedSequence extractedSequenceForOn = seqExtraction.extract(onsForHostname);
 560 ////        ExtractedSequence extractedSequenceForOff = seqExtraction.extract(offsForHostname);
 561 //
 562 //        ExtractedSequence extractedSequenceForOn = seqExtraction.extractByTlsAppData(onsForHostname);
 563 //        ExtractedSequence extractedSequenceForOff = seqExtraction.extractByTlsAppData(offsForHostname);
 564 //
 565 //        // Let's check how many ONs align with OFFs and vice versa (that is, how many times an event is incorrectly
 566 //        // labeled).
 567 //        int onsLabeledAsOff = 0;
 568 //        Integer[] representativeOnSeq = TcpConversationUtils.getPacketLengthSequence(extractedSequenceForOn.getRepresentativeSequence());
 569 //        Integer[] representativeOffSeq = TcpConversationUtils.getPacketLengthSequence(extractedSequenceForOff.getRepresentativeSequence());
 570 //        SequenceAlignment<Integer> seqAlg = seqExtraction.getAlignmentAlgorithm();
 571 //        for (Conversation c : onsForHostname) {
 572 //            Integer[] onSeq = TcpConversationUtils.getPacketLengthSequence(c);
 573 //            if (seqAlg.calculateAlignment(representativeOffSeq, onSeq) <= extractedSequenceForOff.getMaxAlignmentCost()) {
 574 //                onsLabeledAsOff++;
 575 //            }
 576 //        }
 577 //        int offsLabeledAsOn = 0;
 578 //        for (Conversation c : offsForHostname) {
 579 //            Integer[] offSeq = TcpConversationUtils.getPacketLengthSequence(c);
 580 //            if (seqAlg.calculateAlignment(representativeOnSeq, offSeq) <= extractedSequenceForOn.getMaxAlignmentCost()) {
 581 //                offsLabeledAsOn++;
 582 //            }
 583 //        }
 584 //        System.out.println("");
 585 //        // ================================================================================================
 586 //
 587 //
 588 //        // -------------------------------------------------------------------------------------------------------------
 589 //        // -------------------------------------------------------------------------------------------------------------
 590     }
 591
 592 }
 593
 594
 595 // TP-Link MAC 50:c7:bf:33:1f:09 and usually IP 192.168.1.159 (remember to verify per file)
 596 // frame.len >= 556 && frame.len <= 558 && ip.addr == 192.168.1.159