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         // TODO: EXPERIMENT - November 20, 2018
 118         final String inputPcapFile = path + "/experimental_result/standalone/wemo-plug/wlan1/wemo-plug.wlan1.local.pcap";
 119         final String outputPcapFile = path + "/experimental_result/standalone/wemo-plug/wlan1/wemo-plug-processed.pcap";
 120         final String triggerTimesFile = path + "/experimental_result/standalone/wemo-plug/timestamps/wemo-plug-nov-20-2018.timestamps";
 121 //        final String deviceIp = "192.168.1.145"; // .246 == phone; .145 == WeMo
 122         final String deviceIp = "192.168.1.246"; // .246 == phone; .145 == WeMo
 123
 124         // 5) Wemo Insight July 31 experiment
 125 //        final String inputPcapFile = path + "/2018-07/wemoinsight/wemoinsight.wlan1.local.pcap";
 126 //        final String outputPcapFile = path + "/2018-07/wemoinsight/wemoinsight-processed.pcap";
 127 //        final String triggerTimesFile = path + "/2018-07/wemoinsight/wemo-insight-july-31-2018.timestamps";
 128 //        final String deviceIp = "192.168.1.135";
 129
 130         // 6) TP-Link Bulb August 1 experiment
 131 //        final String inputPcapFile = path + "/2018-08/tplink-bulb/tplinkbulb.wlan1.local.pcap";
 132 //        final String outputPcapFile = path + "/2018-08/tplink-bulb/tplinkbulb-processed.pcap";
 133 //        final String triggerTimesFile = path + "/2018-08/tplink-bulb/tplink-bulb-aug-3-2018.timestamps";
 134 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .140 == TP-Link bulb
 135         // TODO: EXPERIMENT - November 16, 2018
 136 //        final String inputPcapFile = path + "/experimental_result/standalone/tplink-bulb/wlan1/tplink-bulb.wlan1.local.pcap";
 137 //        final String outputPcapFile = path + "/experimental_result/standalone/tplink-bulb/wlan1/tplink-bulb-processed.pcap";
 138 ////        final String inputPcapFile = path + "/experimental_result/standalone/tplink-bulb/eth0/tplink-bulb.eth1.local.pcap";
 139 ////        final String outputPcapFile = path + "/experimental_result/standalone/tplink-bulb/eth0/tplink-bulb-processed.pcap";
 140 //        final String triggerTimesFile = path + "/experimental_result/standalone/tplink-bulb/timestamps/tplink-bulb-nov-16-2018.timestamps";
 141 ////        final String deviceIp = "192.168.1.140"; // .246 == phone; .140 == TP-Link bulb
 142 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .140 == TP-Link bulb
 143
 144         // 7) Kwikset Doorlock August 6 experiment
 145 //        final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock.data.wlan1.pcap";
 146 ////        final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock.wlan1.local.pcap";
 147 //        final String outputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-processed.pcap";
 148 ////        final String triggerTimesFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-aug-6-2018.timestamps";
 149 //        final String triggerTimesFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-8hr-data-oct-11-2018.timestamps";
 150 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
 151         // TODO: EXPERIMENT - November 10, 2018
 152 //        final String inputPcapFile = path + "/experimental_result/standalone/kwikset-doorlock/wlan1/kwikset-doorlock.wlan1.local.pcap";
 153 //        final String outputPcapFile = path + "/experimental_result/standalone/kwikset-doorlock/wlan1/kwikset-doorlock-processed.pcap";
 154 ////        final String inputPcapFile = path + "/experimental_result/standalone/kwikset-doorlock/eth1/kwikset-doorlock.eth1.local.pcap";
 155 ////        final String outputPcapFile = path + "/experimental_result/standalone/kwikset-doorlock/eth1/kwikset-doorlock-processed.pcap";
 156 //        final String triggerTimesFile = path + "/experimental_result/standalone/kwikset-doorlock/timestamps/kwikset-doorlock-nov-10-2018.timestamps";
 157 ////        final String deviceIp = "192.168.1.142"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
 158 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
 159
 160         // September 12, 2018 - includes both wlan1 and eth1 interfaces
 161 //        final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset3.wlan1.local.pcap";
 162 //        //final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset3.eth1.local.pcap";
 163 //        final String outputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset3-processed.pcap";
 164 //        final String triggerTimesFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-sept-12-2018.timestamps";
 165 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
 166
 167         // 8) Hue Bulb August 7 experiment
 168 //        final String inputPcapFile = path + "/2018-08/hue-bulb/hue-bulb.wlan1.local.pcap";
 169 //        final String outputPcapFile = path + "/2018-08/hue-bulb/hue-bulb-processed.pcap";
 170 //        final String triggerTimesFile = path + "/2018-08/hue-bulb/hue-bulb-aug-7-2018.timestamps";
 171 //        final String deviceIp = "192.168.1.246";
 172         // October 30 experiment
 173 //        final String inputPcapFile = path + "/2018-10/hue-bulb/hue-bulb.wlan1.local.pcap";
 174 //        final String outputPcapFile = path + "/2018-10/hue-bulb/hue-bulb-processed.pcap";
 175 //        final String triggerTimesFile = path + "/2018-10/hue-bulb/hue-bulb-oct-30-2018.timestamps";
 176 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .100 == Hue hub
 177         // TODO: EXPERIMENT - November 19, 2018
 178 //        final String inputPcapFile = path + "/experimental_result/standalone/hue-bulb/wlan1/hue-bulb.wlan1.local.pcap";
 179 //        final String outputPcapFile = path + "/experimental_result/standalone/hue-bulb/wlan1/hue-bulb-processed.pcap";
 180 //        final String triggerTimesFile = path + "/experimental_result/standalone/hue-bulb/timestamps/hue-bulb-nov-19-2018.timestamps";
 181 //        final String deviceIp = "192.168.1.100"; // .246 == phone; .100 == Hue hub
 182 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .100 == Hue hub
 183
 184         // 9) Lifx Bulb August 8 experiment
 185 //        final String inputPcapFile = path + "/2018-08/lifx-bulb/lifx-bulb.wlan1.local.pcap";
 186 //        final String outputPcapFile = path + "/2018-08/lifx-bulb/lifx-bulb-processed.pcap";
 187 //        final String triggerTimesFile = path + "/2018-08/lifx-bulb/lifx-bulb-aug-8-2018.timestamps";
 188 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .231 == Lifx
 189         // October 18
 190 //        final String inputPcapFile = path + "/2018-10/lifx-bulb/lifx-bulb.wlan1.local.pcap";
 191 //        final String outputPcapFile = path + "/2018-10/lifx-bulb/lifx-bulb-processed.pcap";
 192 //        final String triggerTimesFile = path + "/2018-10/lifx-bulb/lifx-bulb-oct-18-2018.timestamps";
 193 //        final String deviceIp = "192.168.1.231"; // .246 == phone; .231 == Lifx
 194         // November 1
 195 //        final String inputPcapFile = path + "/2018-10/lifx-bulb/lifx-bulb.wlan1.local.pcap";
 196 //        final String outputPcapFile = path + "/2018-10/lifx-bulb/lifx-bulb-processed.pcap";
 197 //        final String triggerTimesFile = path + "/2018-10/lifx-bulb/lifx-bulb-nov-1-2018.timestamps";
 198 //        final String deviceIp = "192.168.1.231"; // .246 == phone; .231 == Lifx
 199
 200         // 10) Amcrest Camera August 9 experiment
 201 //        final String inputPcapFile = path + "/2018-08/amcrest-camera/amcrest-camera.wlan1.local.pcap";
 202 //        final String outputPcapFile = path + "/2018-08/amcrest-camera/amcrest-camera-processed.pcap";
 203 //        final String triggerTimesFile = path + "/2018-08/amcrest-camera/amcrest-camera-aug-9-2018.timestamps";
 204 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .235 == camera
 205
 206         // 11) Arlo Camera August 10 experiment
 207 //        final String inputPcapFile = path + "/2018-08/arlo-camera/arlo-camera.wlan1.local.pcap";
 208 //        final String outputPcapFile = path + "/2018-08/arlo-camera/arlo-camera-processed.pcap";
 209 //        final String triggerTimesFile = path + "/2018-08/arlo-camera/arlo-camera-aug-10-2018.timestamps";
 210 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .140 == camera
 211         // TODO: EXPERIMENT - November 13, 2018
 212 //        final String inputPcapFile = path + "/experimental_result/standalone/arlo-camera/wlan1/arlo-camera.wlan1.local.pcap";
 213 //        final String outputPcapFile = path + "/experimental_result/standalone/arlo-camera/wlan1/arlo-camera-processed.pcap";
 214 ////        final String inputPcapFile = path + "/experimental_result/standalone/arlo-camera/eth0/arlo-camera.eth1.local.pcap";
 215 ////        final String outputPcapFile = path + "/experimental_result/standalone/arlo-camera/eth0/arlo-camera-processed.pcap";
 216 //        final String triggerTimesFile = path + "/experimental_result/standalone/arlo-camera/timestamps/arlo-camera-nov-13-2018.timestamps";
 217 ////        final String deviceIp = "192.168.1.140"; // .246 == phone; .140 == camera
 218 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .140 == camera
 219
 220         // 12) Blossom sprinkler August 13 experiment
 221 //        final String inputPcapFile = path + "/2018-08/blossom/blossom.wlan1.local.pcap";
 222 //        final String outputPcapFile = path + "/2018-08/blossom/blossom-processed.pcap";
 223 //        final String triggerTimesFile = path + "/2018-08/blossom/blossom-aug-13-2018.timestamps";
 224 //        final String deviceIp = "192.168.1.229"; // .246 == phone; .229 == sprinkler
 225         // 2 November
 226 //        final String inputPcapFile = path + "/2018-10/blossom-sprinkler/blossom-sprinkler.wlan1.local.pcap";
 227 //        final String outputPcapFile = path + "/2018-10/blossom-sprinkler/blossom-sprinkler-processed.pcap";
 228 //        final String triggerTimesFile = path + "/2018-10/blossom-sprinkler/blossom-sprinkler-nov-2-2018.timestamps";
 229 //        final String deviceIp = "192.168.1.229"; // .246 == phone; .229 == sprinkler
 230
 231 //        // 13) DLink siren August 14 experiment
 232 //        final String inputPcapFile = path + "/2018-08/dlink-siren/dlink-siren.wlan1.local.pcap";
 233 //        //final String inputPcapFile = path + "/evaluation/dlink-siren/dlink-siren.data.wlan1.pcap";
 234 //        final String outputPcapFile = path + "/2018-08/dlink-siren/dlink-siren-processed.pcap";
 235 //        final String triggerTimesFile = path + "/2018-08/dlink-siren/dlink-siren-oct-12-2018.timestamps";
 236 //        //final String triggerTimesFile = path + "/2018-08/dlink-siren/dlink-siren-aug-14-2018.timestamps";
 237 //        //final String triggerTimesFile = path + "/actual/timestamps/dlink-siren-8hr-data-oct-10-2018.timestamps";
 238 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .183 == siren
 239         // TODO: EXPERIMENT - November 9, 2018
 240 //        final String inputPcapFile = path + "/experimental_result/standalone/dlink-siren/wlan1/dlink-siren.wlan1.local.pcap";
 241 //        final String outputPcapFile = path + "/experimental_result/standalone/dlink-siren/wlan1/dlink-siren-processed.pcap";
 242 //        final String triggerTimesFile = path + "/experimental_result/standalone/dlink-siren/timestamps/dlink-siren-nov-9-2018.timestamps";
 243 //        final String deviceIp = "192.168.1.183"; // .246 == phone; .183 == siren
 244 ////        final String deviceIp = "192.168.1.246"; // .246 == phone; .183 == siren
 245
 246         // 14) Nest thermostat August 15 experiment
 247 //        final String inputPcapFile = path + "/2018-08/nest/nest.wlan1.local.pcap";
 248 //        final String outputPcapFile = path + "/2018-08/nest/nest-processed.pcap";
 249 //        final String triggerTimesFile = path + "/2018-08/nest/nest-aug-15-2018.timestamps";
 250 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .127 == Nest thermostat
 251         // TODO: EXPERIMENT - November 14, 2018
 252 //        final String inputPcapFile = path + "/experimental_result/standalone/nest-thermostat/wlan1/nest-thermostat.wlan1.local.pcap";
 253 //        final String outputPcapFile = path + "/experimental_result/standalone/nest-thermostat/wlan1/nest-thermostat-processed.pcap";
 254 ////        final String inputPcapFile = path + "/experimental_result/standalone/nest-thermostat/eth0/nest-thermostat.eth1.local.pcap";
 255 ////        final String outputPcapFile = path + "/experimental_result/standalone/nest-thermostat/eth0/nest-thermostat-processed.pcap";
 256 //        final String triggerTimesFile = path + "/experimental_result/standalone/nest-thermostat/timestamps/nest-thermostat-nov-15-2018.timestamps";
 257 ////        final String deviceIp = "192.168.1.127"; // .246 == phone; .127 == Nest thermostat
 258 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .127 == Nest thermostat
 259
 260         // 15) Alexa August 16 experiment
 261 //        final String inputPcapFile = path + "/2018-08/alexa/alexa.wlan1.local.pcap";
 262 //        final String outputPcapFile = path + "/2018-08/alexa/alexa-processed.pcap";
 263 //        final String triggerTimesFile = path + "/2018-08/alexa/alexa-aug-16-2018.timestamps";
 264 //        final String deviceIp = "192.168.1.225"; // .246 == phone; .225 == Alexa
 265         // August 17
 266 //        final String inputPcapFile = path + "/2018-08/alexa/alexa2.wlan1.local.pcap";
 267 //        final String outputPcapFile = path + "/2018-08/alexa/alexa2-processed.pcap";
 268 //        final String triggerTimesFile = path + "/2018-08/alexa/alexa-aug-17-2018.timestamps";
 269 //        final String deviceIp = "192.168.1.225"; // .246 == phone; .225 == Alexa
 270
 271         // September 17
 272 //        final String inputPcapFile = path + "/2018-08/noise/noise.eth1.pcap";
 273 //        final String outputPcapFile = path + "/2018-08/noise/noise-processed.pcap";
 274 //        final String triggerTimesFile = path + "/2018-08/noise/noise-sept-17-2018.timestamps";
 275 //        final String deviceIp = "192.168.1.142"; //  .142 == SmartThings Hub; .199 == dlink plug; .183 == siren
 276         // September 26 - D-Link noise
 277 //        final String inputPcapFile = path + "/2018-08/noise/noise.dlink.wlan1.pcap";
 278 //        final String outputPcapFile = path + "/2018-08/noise/noise-processed.pcap";
 279 //        final String triggerTimesFile = path + "/2018-08/noise/dlink-noise-sept-26-2018.timestamps";
 280 //        final String deviceIp = "192.168.1.183"; //  .199 == dlink plug; .183 == siren
 281         // September 27 - Kwikset noise
 282 //        final String inputPcapFile = path + "/2018-08/noise/noise.kwikset.eth1.pcap";
 283 //        final String outputPcapFile = path + "/2018-08/noise/noise-processed.pcap";
 284 //        final String triggerTimesFile = path + "/2018-08/noise/kwikset-doorlock-noise-sept-27-2018.timestamps";
 285 //        final String deviceIp = "192.168.1.142"; //  .142 == SmartThings Hub;
 286
 287         TriggerTimesFileReader ttfr = new TriggerTimesFileReader();
 288         List<Instant> triggerTimes = ttfr.readTriggerTimes(triggerTimesFile, false);
 289         // Tag each trigger with "ON" or "OFF", assuming that the first trigger is an "ON" and that they alternate.
 290         List<UserAction> userActions = new ArrayList<>();
 291         for (int i = 0; i < triggerTimes.size(); i++) {
 292             userActions.add(new UserAction(i % 2 == 0 ? Type.TOGGLE_ON : Type.TOGGLE_OFF, triggerTimes.get(i)));
 293         }
 294         TriggerTrafficExtractor tte = new TriggerTrafficExtractor(inputPcapFile, triggerTimes, deviceIp);
 295         final PcapDumper outputter = Pcaps.openDead(DataLinkType.EN10MB, 65536).dumpOpen(outputPcapFile);
 296         DnsMap dnsMap = new DnsMap();
 297         TcpReassembler tcpReassembler = new TcpReassembler();
 298         TrafficLabeler trafficLabeler = new TrafficLabeler(userActions);
 299         tte.performExtraction(pkt -> {
 300             try {
 301                 outputter.dump(pkt);
 302             } catch (NotOpenException e) {
 303                 e.printStackTrace();
 304             }
 305         }, dnsMap, tcpReassembler, trafficLabeler);
 306         outputter.flush();
 307         outputter.close();
 308
 309         if (tte.getPacketsIncludedCount() != trafficLabeler.getTotalPacketCount()) {
 310             // Sanity/debug check
 311             throw new AssertionError(String.format("mismatch between packet count in %s and %s",
 312                     TriggerTrafficExtractor.class.getSimpleName(), TrafficLabeler.class.getSimpleName()));
 313         }
 314
 315         // Extract all conversations present in the filtered trace.
 316         List<Conversation> allConversations = tcpReassembler.getTcpConversations();
 317         // Group conversations by hostname.
 318         Map<String, List<Conversation>> convsByHostname = TcpConversationUtils.groupConversationsByHostname(allConversations, dnsMap);
 319         System.out.println("Grouped conversations by hostname.");
 320         // For each hostname, count the frequencies of packet lengths exchanged with that hostname.
 321         final Map<String, Map<Integer, Integer>> pktLenFreqsByHostname = new HashMap<>();
 322         convsByHostname.forEach((host, convs) -> pktLenFreqsByHostname.put(host, TcpConversationUtils.countPacketLengthFrequencies(convs)));
 323         System.out.println("Counted frequencies of packet lengths exchanged with each hostname.");
 324         // For each hostname, count the frequencies of packet sequences (i.e., count how many conversations exchange a
 325         // sequence of packets of some specific lengths).
 326         final Map<String, Map<String, Integer>> pktSeqFreqsByHostname = new HashMap<>();
 327         convsByHostname.forEach((host, convs) -> pktSeqFreqsByHostname.put(host, TcpConversationUtils.countPacketSequenceFrequencies(convs)));
 328         System.out.println("Counted frequencies of packet sequences exchanged with each hostname.");
 329         // For each hostname, count frequencies of packet pairs exchanged with that hostname across all conversations
 330         final Map<String, Map<String, Integer>> pktPairFreqsByHostname =
 331                 TcpConversationUtils.countPacketPairFrequenciesByHostname(allConversations, dnsMap);
 332         System.out.println("Counted frequencies of packet pairs per hostname");
 333         // For each user action, reassemble the set of TCP connections occurring shortly after
 334         final Map<UserAction, List<Conversation>> userActionToConversations = trafficLabeler.getLabeledReassembledTcpTraffic();
 335         final Map<UserAction, Map<String, List<Conversation>>> userActionsToConvsByHostname = trafficLabeler.getLabeledReassembledTcpTraffic(dnsMap);
 336         System.out.println("Reassembled TCP conversations occurring shortly after each user event");
 337
 338
 339
 340         /*
 341          * NOTE: no need to generate these more complex on/off maps that also contain mappings from hostname and
 342          * sequence identifiers as we do not care about hostnames and sequences during clustering.
 343          * We can simply use the UserAction->List<Conversation> map to generate ON/OFF groupings of conversations.
 344          */
 345         /*
 346         // Contains all ON events: hostname -> sequence identifier -> list of conversations with that sequence
 347         Map<String, Map<String, List<Conversation>>> ons = new HashMap<>();
 348         // Contains all OFF events: hostname -> sequence identifier -> list of conversations with that sequence
 349         Map<String, Map<String, List<Conversation>>> offs = new HashMap<>();
 350         userActionsToConvsByHostname.forEach((ua, hostnameToConvs) -> {
 351             Map<String, Map<String, List<Conversation>>> outer = ua.getType() == Type.TOGGLE_ON ? ons : offs;
 352             hostnameToConvs.forEach((host, convs) -> {
 353                 Map<String, List<Conversation>> seqsToConvs = TcpConversationUtils.
 354                         groupConversationsByPacketSequence(convs, verbose);
 355                 outer.merge(host, seqsToConvs, (oldMap, newMap) -> {
 356                     newMap.forEach((sequence, cs) -> oldMap.merge(sequence, cs, (list1, list2) -> {
 357                         list1.addAll(list2);
 358                         return list1;
 359                     }));
 360                     return oldMap;
 361                 });
 362             });
 363         });
 364         */
 365
 366         // ================================================ CLUSTERING ================================================
 367         // Note: no need to use the more convoluted on/off maps; can simply use the UserAction->List<Conversation> map
 368         // when don't care about hostnames and sequences (see comment earlier).
 369         List<Conversation> onConversations = userActionToConversations.entrySet().stream().
 370                 filter(e -> e.getKey().getType() == Type.TOGGLE_ON). // drop all OFF events from stream
 371                 map(e -> e.getValue()). // no longer interested in the UserActions
 372                 flatMap(List::stream). // flatten List<List<T>> to a List<T>
 373                 collect(Collectors.toList());
 374         List<Conversation> offConversations = userActionToConversations.entrySet().stream().
 375                 filter(e -> e.getKey().getType() == Type.TOGGLE_OFF).
 376                 map(e -> e.getValue()).
 377                 flatMap(List::stream).
 378                 collect(Collectors.toList());
 379         //Collections.sort(onConversations, (c1, c2) -> c1.getPackets().)
 380
 381         List<PcapPacketPair> onPairs = onConversations.stream().
 382                 map(c -> c.isTls() ? TcpConversationUtils.extractTlsAppDataPacketPairs(c) :
 383                         TcpConversationUtils.extractPacketPairs(c)).
 384                 flatMap(List::stream). // flatten List<List<>> to List<>
 385                 collect(Collectors.toList());
 386         List<PcapPacketPair> offPairs = offConversations.stream().
 387                 map(c -> c.isTls() ? TcpConversationUtils.extractTlsAppDataPacketPairs(c) :
 388                         TcpConversationUtils.extractPacketPairs(c)).
 389                 flatMap(List::stream). // flatten List<List<>> to List<>
 390                 collect(Collectors.toList());
 391         // Note: need to update the DnsMap of all PcapPacketPairs if we want to use the IP/hostname-sensitive distance.
 392         Stream.concat(Stream.of(onPairs), Stream.of(offPairs)).flatMap(List::stream).forEach(p -> p.setDnsMap(dnsMap));
 393         // Perform clustering on conversation logged as part of all ON events.
 394         DBSCANClusterer<PcapPacketPair> onClusterer = new DBSCANClusterer<>(10.0, 45);
 395         //DBSCANClusterer<PcapPacketPair> onClusterer = new DBSCANClusterer<>(10.0, 10);
 396         List<Cluster<PcapPacketPair>> onClusters = onClusterer.cluster(onPairs);
 397         // Perform clustering on conversation logged as part of all OFF events.
 398         DBSCANClusterer<PcapPacketPair> offClusterer = new DBSCANClusterer<>(10.0, 45);
 399         //DBSCANClusterer<PcapPacketPair> offClusterer = new DBSCANClusterer<>(10.0, 10);
 400         List<Cluster<PcapPacketPair>> offClusters = offClusterer.cluster(offPairs);
 401         // Sort the conversations as reference
 402         List<Conversation> sortedAllConversation = TcpConversationUtils.sortConversationList(allConversations);
 403         // Output clusters
 404         System.out.println("========================================");
 405         System.out.println("       Clustering results for ON        ");
 406         System.out.println("       Number of clusters: " + onClusters.size());
 407         int count = 0;
 408         List<List<List<PcapPacket>>> ppListOfListReadOn = new ArrayList<>();
 409         List<List<List<PcapPacket>>> ppListOfListListOn = new ArrayList<>();
 410         for (Cluster<PcapPacketPair> c : onClusters) {
 411             System.out.println(String.format("<<< Cluster #%02d (%03d points) >>>", ++count, c.getPoints().size()));
 412             System.out.print(PrintUtils.toSummaryString(c));
 413             if(c.getPoints().size() > 45 && c.getPoints().size() < 55) {
 414             //if(c.getPoints().size() > 25) {
 415                 // Print to file
 416                 List<List<PcapPacket>> ppListOfList = PcapPacketUtils.clusterToListOfPcapPackets(c);
 417                 ppListOfListListOn.add(ppListOfList);
 418             }
 419         }
 420         // TODO: Merging test
 421         ppListOfListListOn = PcapPacketUtils.mergeSignatures(ppListOfListListOn, sortedAllConversation);
 422         ppListOfListListOn = PcapPacketUtils.sortSignatures(ppListOfListListOn);
 423         // TODO: Need to remove sequence number 2 for ST plug since it is not a good signature!
 424         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 2);
 425         // TODO: Need to remove sequence number 0 for Arlo Camera since it is not a good signature!
 426         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 0);
 427         // TODO: Need to remove sequence number 0 for TP-Link plug since it is not a good signature!
 428         // TODO: This sequence actually belongs to the local communication between the plug and the phone
 429         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 0);
 430         PcapPacketUtils.printSignatures(ppListOfListListOn);
 431         //count = 0;
 432         /*for (List<List<PcapPacket>> ll : ppListOfListListOn) {
 433             PrintUtils.serializeClustersIntoFile("./onSignature" + ++count + ".sig", ll);
 434             ppListOfListReadOn.add(PrintUtils.deserializeClustersFromFile("./onSignature" + count + ".sig"));
 435         }*/
 436         PrintUtils.serializeSignatureIntoFile("./onSignature.sig", ppListOfListListOn);
 437         ppListOfListReadOn = PrintUtils.deserializeSignatureFromFile("./onSignature.sig");
 438
 439         System.out.println("========================================");
 440         System.out.println("       Clustering results for OFF       ");
 441         System.out.println("       Number of clusters: " + offClusters.size());
 442         count = 0;
 443         List<List<List<PcapPacket>>> ppListOfListReadOff = new ArrayList<>();
 444         List<List<List<PcapPacket>>> ppListOfListListOff = new ArrayList<>();
 445         for (Cluster<PcapPacketPair> c : offClusters) {
 446             System.out.println(String.format("<<< Cluster #%03d (%06d points) >>>", ++count, c.getPoints().size()));
 447             System.out.print(PrintUtils.toSummaryString(c));
 448             if(c.getPoints().size() > 45 && c.getPoints().size() < 55) {
 449             //if(c.getPoints().size() > 25) {
 450                 // Print to file
 451                 List<List<PcapPacket>> ppListOfList = PcapPacketUtils.clusterToListOfPcapPackets(c);
 452                 ppListOfListListOff.add(ppListOfList);
 453             }
 454         }
 455         // TODO: Merging test
 456         ppListOfListListOff = PcapPacketUtils.mergeSignatures(ppListOfListListOff, sortedAllConversation);
 457         //ppListOfListListOff = PcapPacketUtils.sortSignatures(ppListOfListListOff);
 458         // TODO: Need to remove sequence number 1 for Nest Thermostat since it is not a good signature!
 459         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 1);
 460         // TODO: Need to remove sequence number 0 for Arlo Camera since it is not a good signature!
 461         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 1);
 462         // TODO: Need to remove sequence number 2 for ST plug since it is not a good signature!
 463         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 2);
 464         // TODO: Need to remove sequence number 0 for TP-Link plug since it is not a good signature!
 465         // TODO: This sequence actually belongs to the local communication between the plug and the phone
 466         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 0);
 467         PcapPacketUtils.printSignatures(ppListOfListListOff);
 468         //count = 0;
 469         /*for (List<List<PcapPacket>> ll : ppListOfListListOff) {
 470             PrintUtils.serializeClustersIntoFile("./offSignature" + ++count + ".sig", ll);
 471             ppListOfListReadOff.add(PrintUtils.deserializeClustersFromFile("./offSignature" + count + ".sig"));
 472         }*/
 473         PrintUtils.serializeSignatureIntoFile("./offSignature.sig", ppListOfListListOff);
 474         ppListOfListReadOff = PrintUtils.deserializeSignatureFromFile("./offSignature.sig");
 475         System.out.println("========================================");
 476         // ============================================================================================================
 477
 478         /*
 479         System.out.println("==== ON ====");
 480         // Print out all the pairs into a file for ON events
 481         File fileOnEvents = new File(onPairsPath);
 482         PrintWriter pwOn = null;
 483         try {
 484             pwOn = new PrintWriter(fileOnEvents);
 485         } catch(Exception ex) {
 486             ex.printStackTrace();
 487         }
 488         for(Map.Entry<String, Map<String, List<Conversation>>> entry : ons.entrySet()) {
 489             Map<String, List<Conversation>> seqsToConvs = entry.getValue();
 490             for(Map.Entry<String, List<Conversation>> entryConv : seqsToConvs.entrySet()) {
 491                 List<Conversation> listConv = entryConv.getValue();
 492                 // Just get the first Conversation because all Conversations in this group
 493                 // should have the same pairs of Application Data.
 494                 for(Conversation conv : listConv) {
 495                     // Process only if it is a TLS packet
 496                     if (conv.isTls()) {
 497                         List<PcapPacketPair> tlsAppDataList = TcpConversationUtils.extractTlsAppDataPacketPairs(conv);
 498                         for(PcapPacketPair pair: tlsAppDataList) {
 499                             System.out.println(PrintUtils.toCsv(pair, dnsMap));
 500                             pwOn.println(PrintUtils.toCsv(pair, dnsMap));
 501                         }
 502                     } else { // Non-TLS conversations
 503                         List<PcapPacketPair> packetList = TcpConversationUtils.extractPacketPairs(conv);
 504                         for(PcapPacketPair pair: packetList) {
 505                             System.out.println(PrintUtils.toCsv(pair, dnsMap));
 506                             pwOn.println(PrintUtils.toCsv(pair, dnsMap));
 507                         }
 508                     }
 509                 }
 510             }
 511         }
 512         pwOn.close();
 513
 514         System.out.println("==== OFF ====");
 515         // Print out all the pairs into a file for ON events
 516         File fileOffEvents = new File(offPairsPath);
 517         PrintWriter pwOff = null;
 518         try {
 519             pwOff = new PrintWriter(fileOffEvents);
 520         } catch(Exception ex) {
 521             ex.printStackTrace();
 522         }
 523         for(Map.Entry<String, Map<String, List<Conversation>>> entry : offs.entrySet()) {
 524             Map<String, List<Conversation>> seqsToConvs = entry.getValue();
 525             for(Map.Entry<String, List<Conversation>> entryConv : seqsToConvs.entrySet()) {
 526                 List<Conversation> listConv = entryConv.getValue();
 527                 // Just get the first Conversation because all Conversations in this group
 528                 // should have the same pairs of Application Data.
 529                 for(Conversation conv : listConv) {
 530                     // Process only if it is a TLS packet
 531                     if (conv.isTls()) {
 532                         List<PcapPacketPair> tlsAppDataList = TcpConversationUtils.extractTlsAppDataPacketPairs(conv);
 533                         for(PcapPacketPair pair: tlsAppDataList) {
 534                             System.out.println(PrintUtils.toCsv(pair, dnsMap));
 535                             pwOff.println(PrintUtils.toCsv(pair, dnsMap));
 536                         }
 537                     } else { // Non-TLS conversations
 538                         List<PcapPacketPair> packetList = TcpConversationUtils.extractPacketPairs(conv);
 539                         for (PcapPacketPair pair : packetList) {
 540                             System.out.println(PrintUtils.toCsv(pair, dnsMap));
 541                             pwOff.println(PrintUtils.toCsv(pair, dnsMap));
 542                         }
 543                     }
 544                 }
 545             }
 546         }
 547         pwOff.close();
 548         */
 549
 550 //        // ================================================================================================
 551 //        // <<< Some work-in-progress/explorative code that extracts a "representative" sequence >>>
 552 //        //
 553 //        // Currently need to know relevant hostname in advance :(
 554 //        String hostname = "events.tplinkra.com";
 555 ////        String hostname = "rfe-us-west-1.dch.dlink.com";
 556 //        // Conversations with 'hostname' for ON events.
 557 //        List<Conversation> onsForHostname = new ArrayList<>();
 558 //        // Conversations with 'hostname' for OFF events.
 559 //        List<Conversation> offsForHostname = new ArrayList<>();
 560 //        // "Unwrap" sequence groupings in ons/offs maps.
 561 //        ons.get(hostname).forEach((k,v) -> onsForHostname.addAll(v));
 562 //        offs.get(hostname).forEach((k,v) -> offsForHostname.addAll(v));
 563 //
 564 //
 565 //        Map<String, List<Conversation>> onsForHostnameGroupedByTlsAppDataSequence = TcpConversationUtils.groupConversationsByTlsApplicationDataPacketSequence(onsForHostname);
 566 //
 567 //
 568 //        // Extract representative sequence for ON and OFF by providing the list of conversations with
 569 //        // 'hostname' observed for each event type (the training data).
 570 //        SequenceExtraction seqExtraction = new SequenceExtraction();
 571 ////        ExtractedSequence extractedSequenceForOn = seqExtraction.extract(onsForHostname);
 572 ////        ExtractedSequence extractedSequenceForOff = seqExtraction.extract(offsForHostname);
 573 //
 574 //        ExtractedSequence extractedSequenceForOn = seqExtraction.extractByTlsAppData(onsForHostname);
 575 //        ExtractedSequence extractedSequenceForOff = seqExtraction.extractByTlsAppData(offsForHostname);
 576 //
 577 //        // Let's check how many ONs align with OFFs and vice versa (that is, how many times an event is incorrectly
 578 //        // labeled).
 579 //        int onsLabeledAsOff = 0;
 580 //        Integer[] representativeOnSeq = TcpConversationUtils.getPacketLengthSequence(extractedSequenceForOn.getRepresentativeSequence());
 581 //        Integer[] representativeOffSeq = TcpConversationUtils.getPacketLengthSequence(extractedSequenceForOff.getRepresentativeSequence());
 582 //        SequenceAlignment<Integer> seqAlg = seqExtraction.getAlignmentAlgorithm();
 583 //        for (Conversation c : onsForHostname) {
 584 //            Integer[] onSeq = TcpConversationUtils.getPacketLengthSequence(c);
 585 //            if (seqAlg.calculateAlignment(representativeOffSeq, onSeq) <= extractedSequenceForOff.getMaxAlignmentCost()) {
 586 //                onsLabeledAsOff++;
 587 //            }
 588 //        }
 589 //        int offsLabeledAsOn = 0;
 590 //        for (Conversation c : offsForHostname) {
 591 //            Integer[] offSeq = TcpConversationUtils.getPacketLengthSequence(c);
 592 //            if (seqAlg.calculateAlignment(representativeOnSeq, offSeq) <= extractedSequenceForOn.getMaxAlignmentCost()) {
 593 //                offsLabeledAsOn++;
 594 //            }
 595 //        }
 596 //        System.out.println("");
 597 //        // ================================================================================================
 598 //
 599 //
 600 //        // -------------------------------------------------------------------------------------------------------------
 601 //        // -------------------------------------------------------------------------------------------------------------
 602     }
 603
 604 }
 605
 606
 607 // TP-Link MAC 50:c7:bf:33:1f:09 and usually IP 192.168.1.159 (remember to verify per file)
 608 // frame.len >= 556 && frame.len <= 558 && ip.addr == 192.168.1.159