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         // TODO: EXPERIMENT - November 21, 2018
 130 //        final String inputPcapFile = path + "/experimental_result/standalone/wemo-insight-plug/wlan1/wemo-insight-plug.wlan1.local.pcap";
 131 //        final String outputPcapFile = path + "/experimental_result/standalone/wemo-insight-plug/wlan1/wemo-insight-plug-processed.pcap";
 132 //        final String triggerTimesFile = path + "/experimental_result/standalone/wemo-insight-plug/timestamps/wemo-insight-plug-nov-21-2018.timestamps";
 133 ////        final String deviceIp = "192.168.1.145"; // .246 == phone; .135 == WeMo Insight
 134 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .135 == WeMo Insight
 135
 136         // 6) TP-Link Bulb August 1 experiment
 137 //        final String inputPcapFile = path + "/2018-08/tplink-bulb/tplinkbulb.wlan1.local.pcap";
 138 //        final String outputPcapFile = path + "/2018-08/tplink-bulb/tplinkbulb-processed.pcap";
 139 //        final String triggerTimesFile = path + "/2018-08/tplink-bulb/tplink-bulb-aug-3-2018.timestamps";
 140 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .140 == TP-Link bulb
 141         // TODO: EXPERIMENT - November 16, 2018
 142 //        final String inputPcapFile = path + "/experimental_result/standalone/tplink-bulb/wlan1/tplink-bulb.wlan1.local.pcap";
 143 //        final String outputPcapFile = path + "/experimental_result/standalone/tplink-bulb/wlan1/tplink-bulb-processed.pcap";
 144 ////        final String inputPcapFile = path + "/experimental_result/standalone/tplink-bulb/eth0/tplink-bulb.eth1.local.pcap";
 145 ////        final String outputPcapFile = path + "/experimental_result/standalone/tplink-bulb/eth0/tplink-bulb-processed.pcap";
 146 //        final String triggerTimesFile = path + "/experimental_result/standalone/tplink-bulb/timestamps/tplink-bulb-nov-16-2018.timestamps";
 147 ////        final String deviceIp = "192.168.1.140"; // .246 == phone; .140 == TP-Link bulb
 148 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .140 == TP-Link bulb
 149
 150         // 7) Kwikset Doorlock August 6 experiment
 151 //        final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock.data.wlan1.pcap";
 152 ////        final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock.wlan1.local.pcap";
 153 //        final String outputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-processed.pcap";
 154 ////        final String triggerTimesFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-aug-6-2018.timestamps";
 155 //        final String triggerTimesFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-8hr-data-oct-11-2018.timestamps";
 156 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
 157         // TODO: EXPERIMENT - November 10, 2018
 158 //        final String inputPcapFile = path + "/experimental_result/standalone/kwikset-doorlock/wlan1/kwikset-doorlock.wlan1.local.pcap";
 159 //        final String outputPcapFile = path + "/experimental_result/standalone/kwikset-doorlock/wlan1/kwikset-doorlock-processed.pcap";
 160 ////        final String inputPcapFile = path + "/experimental_result/standalone/kwikset-doorlock/eth1/kwikset-doorlock.eth1.local.pcap";
 161 ////        final String outputPcapFile = path + "/experimental_result/standalone/kwikset-doorlock/eth1/kwikset-doorlock-processed.pcap";
 162 //        final String triggerTimesFile = path + "/experimental_result/standalone/kwikset-doorlock/timestamps/kwikset-doorlock-nov-10-2018.timestamps";
 163 ////        final String deviceIp = "192.168.1.142"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
 164 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
 165
 166         // September 12, 2018 - includes both wlan1 and eth1 interfaces
 167 //        final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset3.wlan1.local.pcap";
 168 //        //final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset3.eth1.local.pcap";
 169 //        final String outputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset3-processed.pcap";
 170 //        final String triggerTimesFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-sept-12-2018.timestamps";
 171 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
 172
 173         // 8) Hue Bulb August 7 experiment
 174 //        final String inputPcapFile = path + "/2018-08/hue-bulb/hue-bulb.wlan1.local.pcap";
 175 //        final String outputPcapFile = path + "/2018-08/hue-bulb/hue-bulb-processed.pcap";
 176 //        final String triggerTimesFile = path + "/2018-08/hue-bulb/hue-bulb-aug-7-2018.timestamps";
 177 //        final String deviceIp = "192.168.1.246";
 178         // October 30 experiment
 179 //        final String inputPcapFile = path + "/2018-10/hue-bulb/hue-bulb.wlan1.local.pcap";
 180 //        final String outputPcapFile = path + "/2018-10/hue-bulb/hue-bulb-processed.pcap";
 181 //        final String triggerTimesFile = path + "/2018-10/hue-bulb/hue-bulb-oct-30-2018.timestamps";
 182 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .100 == Hue hub
 183         // TODO: EXPERIMENT - November 19, 2018
 184 //        final String inputPcapFile = path + "/experimental_result/standalone/hue-bulb/wlan1/hue-bulb.wlan1.local.pcap";
 185 //        final String outputPcapFile = path + "/experimental_result/standalone/hue-bulb/wlan1/hue-bulb-processed.pcap";
 186 //        final String triggerTimesFile = path + "/experimental_result/standalone/hue-bulb/timestamps/hue-bulb-nov-19-2018.timestamps";
 187 //        final String deviceIp = "192.168.1.100"; // .246 == phone; .100 == Hue hub
 188 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .100 == Hue hub
 189
 190         // 9) Lifx Bulb August 8 experiment
 191 //        final String inputPcapFile = path + "/2018-08/lifx-bulb/lifx-bulb.wlan1.local.pcap";
 192 //        final String outputPcapFile = path + "/2018-08/lifx-bulb/lifx-bulb-processed.pcap";
 193 //        final String triggerTimesFile = path + "/2018-08/lifx-bulb/lifx-bulb-aug-8-2018.timestamps";
 194 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .231 == Lifx
 195         // October 18
 196 //        final String inputPcapFile = path + "/2018-10/lifx-bulb/lifx-bulb.wlan1.local.pcap";
 197 //        final String outputPcapFile = path + "/2018-10/lifx-bulb/lifx-bulb-processed.pcap";
 198 //        final String triggerTimesFile = path + "/2018-10/lifx-bulb/lifx-bulb-oct-18-2018.timestamps";
 199 //        final String deviceIp = "192.168.1.231"; // .246 == phone; .231 == Lifx
 200         // November 1
 201 //        final String inputPcapFile = path + "/2018-10/lifx-bulb/lifx-bulb.wlan1.local.pcap";
 202 //        final String outputPcapFile = path + "/2018-10/lifx-bulb/lifx-bulb-processed.pcap";
 203 //        final String triggerTimesFile = path + "/2018-10/lifx-bulb/lifx-bulb-nov-1-2018.timestamps";
 204 //        final String deviceIp = "192.168.1.231"; // .246 == phone; .231 == Lifx
 205
 206         // 10) Amcrest Camera August 9 experiment
 207 //        final String inputPcapFile = path + "/2018-08/amcrest-camera/amcrest-camera.wlan1.local.pcap";
 208 //        final String outputPcapFile = path + "/2018-08/amcrest-camera/amcrest-camera-processed.pcap";
 209 //        final String triggerTimesFile = path + "/2018-08/amcrest-camera/amcrest-camera-aug-9-2018.timestamps";
 210 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .235 == camera
 211
 212         // 11) Arlo Camera August 10 experiment
 213 //        final String inputPcapFile = path + "/2018-08/arlo-camera/arlo-camera.wlan1.local.pcap";
 214 //        final String outputPcapFile = path + "/2018-08/arlo-camera/arlo-camera-processed.pcap";
 215 //        final String triggerTimesFile = path + "/2018-08/arlo-camera/arlo-camera-aug-10-2018.timestamps";
 216 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .140 == camera
 217         // TODO: EXPERIMENT - November 13, 2018
 218 //        final String inputPcapFile = path + "/experimental_result/standalone/arlo-camera/wlan1/arlo-camera.wlan1.local.pcap";
 219 //        final String outputPcapFile = path + "/experimental_result/standalone/arlo-camera/wlan1/arlo-camera-processed.pcap";
 220 ////        final String inputPcapFile = path + "/experimental_result/standalone/arlo-camera/eth0/arlo-camera.eth1.local.pcap";
 221 ////        final String outputPcapFile = path + "/experimental_result/standalone/arlo-camera/eth0/arlo-camera-processed.pcap";
 222 //        final String triggerTimesFile = path + "/experimental_result/standalone/arlo-camera/timestamps/arlo-camera-nov-13-2018.timestamps";
 223 ////        final String deviceIp = "192.168.1.140"; // .246 == phone; .140 == camera
 224 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .140 == camera
 225
 226         // 12) Blossom sprinkler August 13 experiment
 227 //        final String inputPcapFile = path + "/2018-08/blossom/blossom.wlan1.local.pcap";
 228 //        final String outputPcapFile = path + "/2018-08/blossom/blossom-processed.pcap";
 229 //        final String triggerTimesFile = path + "/2018-08/blossom/blossom-aug-13-2018.timestamps";
 230 //        final String deviceIp = "192.168.1.229"; // .246 == phone; .229 == sprinkler
 231 //        // 2 November
 232 //        final String inputPcapFile = path + "/2018-10/blossom-sprinkler/blossom-sprinkler.wlan1.local.pcap";
 233 //        final String outputPcapFile = path + "/2018-10/blossom-sprinkler/blossom-sprinkler-processed.pcap";
 234 //        final String triggerTimesFile = path + "/2018-10/blossom-sprinkler/blossom-sprinkler-nov-2-2018.timestamps";
 235 //        final String deviceIp = "192.168.1.229"; // .246 == phone; .229 == sprinkler
 236         // January 9
 237         final String inputPcapFile = path + "/experimental_result/standalone/blossom-sprinkler/wlan1/blossom-sprinkler.wlan1.local.pcap";
 238         final String outputPcapFile = path + "/experimental_result/standalone/blossom-sprinkler/wlan1/blossom-sprinkler-processed.pcap";
 239         final String triggerTimesFile = path + "/experimental_result/standalone/blossom-sprinkler/timestamps/blossom-sprinkler-standalone-jan-9-2018.timestamps";
 240         final String deviceIp = "192.168.1.246"; // .246 == phone; .229 == sprinkler
 241
 242 //        // 13) DLink siren August 14 experiment
 243 //        final String inputPcapFile = path + "/2018-08/dlink-siren/dlink-siren.wlan1.local.pcap";
 244 //        //final String inputPcapFile = path + "/evaluation/dlink-siren/dlink-siren.data.wlan1.pcap";
 245 //        final String outputPcapFile = path + "/2018-08/dlink-siren/dlink-siren-processed.pcap";
 246 //        final String triggerTimesFile = path + "/2018-08/dlink-siren/dlink-siren-oct-12-2018.timestamps";
 247 //        //final String triggerTimesFile = path + "/2018-08/dlink-siren/dlink-siren-aug-14-2018.timestamps";
 248 //        //final String triggerTimesFile = path + "/actual/timestamps/dlink-siren-8hr-data-oct-10-2018.timestamps";
 249 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .183 == siren
 250         // TODO: EXPERIMENT - November 9, 2018
 251 //        final String inputPcapFile = path + "/experimental_result/standalone/dlink-siren/wlan1/dlink-siren.wlan1.local.pcap";
 252 //        final String outputPcapFile = path + "/experimental_result/standalone/dlink-siren/wlan1/dlink-siren-processed.pcap";
 253 //        final String triggerTimesFile = path + "/experimental_result/standalone/dlink-siren/timestamps/dlink-siren-nov-9-2018.timestamps";
 254 //        final String deviceIp = "192.168.1.183"; // .246 == phone; .183 == siren
 255 ////        final String deviceIp = "192.168.1.246"; // .246 == phone; .183 == siren
 256
 257         // 14) Nest thermostat August 15 experiment
 258 //        final String inputPcapFile = path + "/2018-08/nest/nest.wlan1.local.pcap";
 259 //        final String outputPcapFile = path + "/2018-08/nest/nest-processed.pcap";
 260 //        final String triggerTimesFile = path + "/2018-08/nest/nest-aug-15-2018.timestamps";
 261 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .127 == Nest thermostat
 262         // TODO: EXPERIMENT - November 14, 2018
 263 //        final String inputPcapFile = path + "/experimental_result/standalone/nest-thermostat/wlan1/nest-thermostat.wlan1.local.pcap";
 264 //        final String outputPcapFile = path + "/experimental_result/standalone/nest-thermostat/wlan1/nest-thermostat-processed.pcap";
 265 ////        final String inputPcapFile = path + "/experimental_result/standalone/nest-thermostat/eth0/nest-thermostat.eth1.local.pcap";
 266 ////        final String outputPcapFile = path + "/experimental_result/standalone/nest-thermostat/eth0/nest-thermostat-processed.pcap";
 267 //        final String triggerTimesFile = path + "/experimental_result/standalone/nest-thermostat/timestamps/nest-thermostat-nov-15-2018.timestamps";
 268 ////        final String deviceIp = "192.168.1.127"; // .246 == phone; .127 == Nest thermostat
 269 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .127 == Nest thermostat
 270
 271         // 15) Alexa August 16 experiment
 272 //        final String inputPcapFile = path + "/2018-08/alexa/alexa.wlan1.local.pcap";
 273 //        final String outputPcapFile = path + "/2018-08/alexa/alexa-processed.pcap";
 274 //        final String triggerTimesFile = path + "/2018-08/alexa/alexa-aug-16-2018.timestamps";
 275 //        final String deviceIp = "192.168.1.225"; // .246 == phone; .225 == Alexa
 276         // August 17
 277 //        final String inputPcapFile = path + "/2018-08/alexa/alexa2.wlan1.local.pcap";
 278 //        final String outputPcapFile = path + "/2018-08/alexa/alexa2-processed.pcap";
 279 //        final String triggerTimesFile = path + "/2018-08/alexa/alexa-aug-17-2018.timestamps";
 280 //        final String deviceIp = "192.168.1.225"; // .246 == phone; .225 == Alexa
 281
 282         // September 17
 283 //        final String inputPcapFile = path + "/2018-08/noise/noise.eth1.pcap";
 284 //        final String outputPcapFile = path + "/2018-08/noise/noise-processed.pcap";
 285 //        final String triggerTimesFile = path + "/2018-08/noise/noise-sept-17-2018.timestamps";
 286 //        final String deviceIp = "192.168.1.142"; //  .142 == SmartThings Hub; .199 == dlink plug; .183 == siren
 287         // September 26 - D-Link noise
 288 //        final String inputPcapFile = path + "/2018-08/noise/noise.dlink.wlan1.pcap";
 289 //        final String outputPcapFile = path + "/2018-08/noise/noise-processed.pcap";
 290 //        final String triggerTimesFile = path + "/2018-08/noise/dlink-noise-sept-26-2018.timestamps";
 291 //        final String deviceIp = "192.168.1.183"; //  .199 == dlink plug; .183 == siren
 292         // September 27 - Kwikset noise
 293 //        final String inputPcapFile = path + "/2018-08/noise/noise.kwikset.eth1.pcap";
 294 //        final String outputPcapFile = path + "/2018-08/noise/noise-processed.pcap";
 295 //        final String triggerTimesFile = path + "/2018-08/noise/kwikset-doorlock-noise-sept-27-2018.timestamps";
 296 //        final String deviceIp = "192.168.1.142"; //  .142 == SmartThings Hub;
 297
 298         TriggerTimesFileReader ttfr = new TriggerTimesFileReader();
 299         List<Instant> triggerTimes = ttfr.readTriggerTimes(triggerTimesFile, false);
 300         // Tag each trigger with "ON" or "OFF", assuming that the first trigger is an "ON" and that they alternate.
 301         List<UserAction> userActions = new ArrayList<>();
 302         for (int i = 0; i < triggerTimes.size(); i++) {
 303             userActions.add(new UserAction(i % 2 == 0 ? Type.TOGGLE_ON : Type.TOGGLE_OFF, triggerTimes.get(i)));
 304         }
 305         TriggerTrafficExtractor tte = new TriggerTrafficExtractor(inputPcapFile, triggerTimes, deviceIp);
 306         final PcapDumper outputter = Pcaps.openDead(DataLinkType.EN10MB, 65536).dumpOpen(outputPcapFile);
 307         DnsMap dnsMap = new DnsMap();
 308         TcpReassembler tcpReassembler = new TcpReassembler();
 309         TrafficLabeler trafficLabeler = new TrafficLabeler(userActions);
 310         tte.performExtraction(pkt -> {
 311             try {
 312                 outputter.dump(pkt);
 313             } catch (NotOpenException e) {
 314                 e.printStackTrace();
 315             }
 316         }, dnsMap, tcpReassembler, trafficLabeler);
 317         outputter.flush();
 318         outputter.close();
 319
 320         if (tte.getPacketsIncludedCount() != trafficLabeler.getTotalPacketCount()) {
 321             // Sanity/debug check
 322             throw new AssertionError(String.format("mismatch between packet count in %s and %s",
 323                     TriggerTrafficExtractor.class.getSimpleName(), TrafficLabeler.class.getSimpleName()));
 324         }
 325
 326         // Extract all conversations present in the filtered trace.
 327         List<Conversation> allConversations = tcpReassembler.getTcpConversations();
 328         // Group conversations by hostname.
 329         Map<String, List<Conversation>> convsByHostname = TcpConversationUtils.groupConversationsByHostname(allConversations, dnsMap);
 330         System.out.println("Grouped conversations by hostname.");
 331         // For each hostname, count the frequencies of packet lengths exchanged with that hostname.
 332         final Map<String, Map<Integer, Integer>> pktLenFreqsByHostname = new HashMap<>();
 333         convsByHostname.forEach((host, convs) -> pktLenFreqsByHostname.put(host, TcpConversationUtils.countPacketLengthFrequencies(convs)));
 334         System.out.println("Counted frequencies of packet lengths exchanged with each hostname.");
 335         // For each hostname, count the frequencies of packet sequences (i.e., count how many conversations exchange a
 336         // sequence of packets of some specific lengths).
 337         final Map<String, Map<String, Integer>> pktSeqFreqsByHostname = new HashMap<>();
 338         convsByHostname.forEach((host, convs) -> pktSeqFreqsByHostname.put(host, TcpConversationUtils.countPacketSequenceFrequencies(convs)));
 339         System.out.println("Counted frequencies of packet sequences exchanged with each hostname.");
 340         // For each hostname, count frequencies of packet pairs exchanged with that hostname across all conversations
 341         final Map<String, Map<String, Integer>> pktPairFreqsByHostname =
 342                 TcpConversationUtils.countPacketPairFrequenciesByHostname(allConversations, dnsMap);
 343         System.out.println("Counted frequencies of packet pairs per hostname");
 344         // For each user action, reassemble the set of TCP connections occurring shortly after
 345         final Map<UserAction, List<Conversation>> userActionToConversations = trafficLabeler.getLabeledReassembledTcpTraffic();
 346         final Map<UserAction, Map<String, List<Conversation>>> userActionsToConvsByHostname = trafficLabeler.getLabeledReassembledTcpTraffic(dnsMap);
 347         System.out.println("Reassembled TCP conversations occurring shortly after each user event");
 348
 349
 350
 351         /*
 352          * NOTE: no need to generate these more complex on/off maps that also contain mappings from hostname and
 353          * sequence identifiers as we do not care about hostnames and sequences during clustering.
 354          * We can simply use the UserAction->List<Conversation> map to generate ON/OFF groupings of conversations.
 355          */
 356
 357 //        // Contains all ON events: hostname -> sequence identifier -> list of conversations with that sequence
 358 //        Map<String, Map<String, List<Conversation>>> ons = new HashMap<>();
 359 //        // Contains all OFF events: hostname -> sequence identifier -> list of conversations with that sequence
 360 //        Map<String, Map<String, List<Conversation>>> offs = new HashMap<>();
 361 //        userActionsToConvsByHostname.forEach((ua, hostnameToConvs) -> {
 362 //            Map<String, Map<String, List<Conversation>>> outer = ua.getType() == Type.TOGGLE_ON ? ons : offs;
 363 //            hostnameToConvs.forEach((host, convs) -> {
 364 //                Map<String, List<Conversation>> seqsToConvs = TcpConversationUtils.
 365 //                        groupConversationsByPacketSequence(convs, verbose);
 366 //                outer.merge(host, seqsToConvs, (oldMap, newMap) -> {
 367 //                    newMap.forEach((sequence, cs) -> oldMap.merge(sequence, cs, (list1, list2) -> {
 368 //                        list1.addAll(list2);
 369 //                        return list1;
 370 //                    }));
 371 //                    return oldMap;
 372 //                });
 373 //            });
 374 //        });
 375 //
 376 //        System.out.println("==== ON ====");
 377 //        // Print out all the pairs into a file for ON events
 378 //        File fileOnEvents = new File(onPairsPath);
 379 //        PrintWriter pwOn = null;
 380 //        try {
 381 //            pwOn = new PrintWriter(fileOnEvents);
 382 //        } catch(Exception ex) {
 383 //            ex.printStackTrace();
 384 //        }
 385 //        for(Map.Entry<String, Map<String, List<Conversation>>> entry : ons.entrySet()) {
 386 //            Map<String, List<Conversation>> seqsToConvs = entry.getValue();
 387 //            for(Map.Entry<String, List<Conversation>> entryConv : seqsToConvs.entrySet()) {
 388 //                List<Conversation> listConv = entryConv.getValue();
 389 //                // Just get the first Conversation because all Conversations in this group
 390 //                // should have the same pairs of Application Data.
 391 //                for(Conversation conv : listConv) {
 392 //                    // Process only if it is a TLS packet
 393 //                    if (conv.isTls()) {
 394 //                        List<PcapPacketPair> tlsAppDataList = TcpConversationUtils.extractTlsAppDataPacketPairs(conv);
 395 //                        for(PcapPacketPair pair: tlsAppDataList) {
 396 //                            System.out.println(PrintUtils.toCsv(pair, dnsMap));
 397 //                            pwOn.println(PrintUtils.toCsv(pair, dnsMap));
 398 //                        }
 399 //                    } else { // Non-TLS conversations
 400 //                        List<PcapPacketPair> packetList = TcpConversationUtils.extractPacketPairs(conv);
 401 //                        for(PcapPacketPair pair: packetList) {
 402 //                            System.out.println(PrintUtils.toCsv(pair, dnsMap));
 403 //                            pwOn.println(PrintUtils.toCsv(pair, dnsMap));
 404 //                        }
 405 //                    }
 406 //                }
 407 //            }
 408 //        }
 409 //        pwOn.close();
 410 //
 411 //        System.out.println("==== OFF ====");
 412 //        // Print out all the pairs into a file for ON events
 413 //        File fileOffEvents = new File(offPairsPath);
 414 //        PrintWriter pwOff = null;
 415 //        try {
 416 //            pwOff = new PrintWriter(fileOffEvents);
 417 //        } catch(Exception ex) {
 418 //            ex.printStackTrace();
 419 //        }
 420 //        for(Map.Entry<String, Map<String, List<Conversation>>> entry : offs.entrySet()) {
 421 //            Map<String, List<Conversation>> seqsToConvs = entry.getValue();
 422 //            for(Map.Entry<String, List<Conversation>> entryConv : seqsToConvs.entrySet()) {
 423 //                List<Conversation> listConv = entryConv.getValue();
 424 //                // Just get the first Conversation because all Conversations in this group
 425 //                // should have the same pairs of Application Data.
 426 //                for(Conversation conv : listConv) {
 427 //                    // Process only if it is a TLS packet
 428 //                    if (conv.isTls()) {
 429 //                        List<PcapPacketPair> tlsAppDataList = TcpConversationUtils.extractTlsAppDataPacketPairs(conv);
 430 //                        for(PcapPacketPair pair: tlsAppDataList) {
 431 //                            System.out.println(PrintUtils.toCsv(pair, dnsMap));
 432 //                            pwOff.println(PrintUtils.toCsv(pair, dnsMap));
 433 //                        }
 434 //                    } else { // Non-TLS conversations
 435 //                        List<PcapPacketPair> packetList = TcpConversationUtils.extractPacketPairs(conv);
 436 //                        for (PcapPacketPair pair : packetList) {
 437 //                            System.out.println(PrintUtils.toCsv(pair, dnsMap));
 438 //                            pwOff.println(PrintUtils.toCsv(pair, dnsMap));
 439 //                        }
 440 //                    }
 441 //                }
 442 //            }
 443 //        }
 444 //        pwOff.close();
 445
 446
 447         // ================================================ CLUSTERING ================================================
 448         // Note: no need to use the more convoluted on/off maps; can simply use the UserAction->List<Conversation> map
 449         // when don't care about hostnames and sequences (see comment earlier).
 450         List<Conversation> onConversations = userActionToConversations.entrySet().stream().
 451                 filter(e -> e.getKey().getType() == Type.TOGGLE_ON). // drop all OFF events from stream
 452                 map(e -> e.getValue()). // no longer interested in the UserActions
 453                 flatMap(List::stream). // flatten List<List<T>> to a List<T>
 454                 collect(Collectors.toList());
 455         List<Conversation> offConversations = userActionToConversations.entrySet().stream().
 456                 filter(e -> e.getKey().getType() == Type.TOGGLE_OFF).
 457                 map(e -> e.getValue()).
 458                 flatMap(List::stream).
 459                 collect(Collectors.toList());
 460         //Collections.sort(onConversations, (c1, c2) -> c1.getPackets().)
 461
 462         List<PcapPacketPair> onPairs = onConversations.stream().
 463                 map(c -> c.isTls() ? TcpConversationUtils.extractTlsAppDataPacketPairs(c) :
 464                         TcpConversationUtils.extractPacketPairs(c)).
 465                 flatMap(List::stream). // flatten List<List<>> to List<>
 466                 collect(Collectors.toList());
 467         List<PcapPacketPair> offPairs = offConversations.stream().
 468                 map(c -> c.isTls() ? TcpConversationUtils.extractTlsAppDataPacketPairs(c) :
 469                         TcpConversationUtils.extractPacketPairs(c)).
 470                 flatMap(List::stream). // flatten List<List<>> to List<>
 471                 collect(Collectors.toList());
 472         // Note: need to update the DnsMap of all PcapPacketPairs if we want to use the IP/hostname-sensitive distance.
 473         Stream.concat(Stream.of(onPairs), Stream.of(offPairs)).flatMap(List::stream).forEach(p -> p.setDnsMap(dnsMap));
 474         // Perform clustering on conversation logged as part of all ON events.
 475         DBSCANClusterer<PcapPacketPair> onClusterer = new DBSCANClusterer<>(10.0, 45);
 476         //DBSCANClusterer<PcapPacketPair> onClusterer = new DBSCANClusterer<>(7.0, 45);
 477         //DBSCANClusterer<PcapPacketPair> onClusterer = new DBSCANClusterer<>(10.0, 10);
 478         List<Cluster<PcapPacketPair>> onClusters = onClusterer.cluster(onPairs);
 479         // Perform clustering on conversation logged as part of all OFF events.
 480         DBSCANClusterer<PcapPacketPair> offClusterer = new DBSCANClusterer<>(10.0, 45);
 481         //DBSCANClusterer<PcapPacketPair> offClusterer = new DBSCANClusterer<>(7.0, 45);
 482         //DBSCANClusterer<PcapPacketPair> offClusterer = new DBSCANClusterer<>(10.0, 10);
 483         List<Cluster<PcapPacketPair>> offClusters = offClusterer.cluster(offPairs);
 484         // Sort the conversations as reference
 485         List<Conversation> sortedAllConversation = TcpConversationUtils.sortConversationList(allConversations);
 486         // Output clusters
 487         System.out.println("========================================");
 488         System.out.println("       Clustering results for ON        ");
 489         System.out.println("       Number of clusters: " + onClusters.size());
 490         int count = 0;
 491         List<List<List<PcapPacket>>> ppListOfListReadOn = new ArrayList<>();
 492         List<List<List<PcapPacket>>> ppListOfListListOn = new ArrayList<>();
 493         for (Cluster<PcapPacketPair> c : onClusters) {
 494             System.out.println(String.format("<<< Cluster #%02d (%03d points) >>>", ++count, c.getPoints().size()));
 495             System.out.print(PrintUtils.toSummaryString(c));
 496             if(c.getPoints().size() > 45 && c.getPoints().size() < 55) {
 497             //if(c.getPoints().size() > 25) {
 498                 // Print to file
 499                 List<List<PcapPacket>> ppListOfList = PcapPacketUtils.clusterToListOfPcapPackets(c);
 500                 ppListOfListListOn.add(ppListOfList);
 501             }
 502         }
 503         // TODO: Merging test
 504         ppListOfListListOn = PcapPacketUtils.mergeSignatures(ppListOfListListOn, sortedAllConversation);
 505         ppListOfListListOn = PcapPacketUtils.sortSignatures(ppListOfListListOn);
 506         // TODO: Need to remove sequence numbers 0,1 for Blossom phone side since it is not a good signature!
 507         PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 0);
 508         PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 1);
 509         // TODO: Need to remove sequence numbers 0 for Blossom device side since it is not a good signature!
 510         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 0);
 511         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 0);
 512         // TODO: Need to remove sequence number 2 for ST plug since it is not a good signature!
 513         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 2);
 514         // TODO: Need to remove sequence number 0 for Arlo Camera since it is not a good signature!
 515         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 0);
 516         // TODO: Need to remove sequence number 0 for TP-Link plug since it is not a good signature!
 517         // TODO: This sequence actually belongs to the local communication between the plug and the phone
 518         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 0);
 519         PcapPacketUtils.printSignatures(ppListOfListListOn);
 520         //count = 0;
 521         /*for (List<List<PcapPacket>> ll : ppListOfListListOn) {
 522             PrintUtils.serializeClustersIntoFile("./onSignature" + ++count + ".sig", ll);
 523             ppListOfListReadOn.add(PrintUtils.deserializeClustersFromFile("./onSignature" + count + ".sig"));
 524         }*/
 525         PrintUtils.serializeSignatureIntoFile("./onSignature.sig", ppListOfListListOn);
 526         ppListOfListReadOn = PrintUtils.deserializeSignatureFromFile("./onSignature.sig");
 527
 528         System.out.println("========================================");
 529         System.out.println("       Clustering results for OFF       ");
 530         System.out.println("       Number of clusters: " + offClusters.size());
 531         count = 0;
 532         List<List<List<PcapPacket>>> ppListOfListReadOff = new ArrayList<>();
 533         List<List<List<PcapPacket>>> ppListOfListListOff = new ArrayList<>();
 534         for (Cluster<PcapPacketPair> c : offClusters) {
 535             System.out.println(String.format("<<< Cluster #%03d (%06d points) >>>", ++count, c.getPoints().size()));
 536             System.out.print(PrintUtils.toSummaryString(c));
 537             if(c.getPoints().size() > 45 && c.getPoints().size() < 55) {
 538             //if(c.getPoints().size() > 25) {
 539                 // Print to file
 540                 List<List<PcapPacket>> ppListOfList = PcapPacketUtils.clusterToListOfPcapPackets(c);
 541                 ppListOfListListOff.add(ppListOfList);
 542             }
 543         }
 544         // TODO: Merging test
 545         ppListOfListListOff = PcapPacketUtils.mergeSignatures(ppListOfListListOff, sortedAllConversation);
 546         //ppListOfListListOff = PcapPacketUtils.sortSignatures(ppListOfListListOff);
 547         // TODO: Need to remove sequence numbers 0,2 for Blossom device side since it is not a good signature!
 548         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 0);
 549         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 2);
 550         // TODO: Need to remove sequence number 1 for Nest Thermostat since it is not a good signature!
 551         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 1);
 552         // TODO: Need to remove sequence number 0 for Arlo Camera since it is not a good signature!
 553         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 1);
 554         // TODO: Need to remove sequence number 2 for ST plug since it is not a good signature!
 555         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 2);
 556         // TODO: Need to remove sequence number 0 for TP-Link plug since it is not a good signature!
 557         // TODO: This sequence actually belongs to the local communication between the plug and the phone
 558         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 0);
 559         PcapPacketUtils.printSignatures(ppListOfListListOff);
 560         //count = 0;
 561         /*for (List<List<PcapPacket>> ll : ppListOfListListOff) {
 562             PrintUtils.serializeClustersIntoFile("./offSignature" + ++count + ".sig", ll);
 563             ppListOfListReadOff.add(PrintUtils.deserializeClustersFromFile("./offSignature" + count + ".sig"));
 564         }*/
 565         PrintUtils.serializeSignatureIntoFile("./offSignature.sig", ppListOfListListOff);
 566         ppListOfListReadOff = PrintUtils.deserializeSignatureFromFile("./offSignature.sig");
 567         System.out.println("========================================");
 568         // ============================================================================================================
 569
 570
 571 //        // ================================================================================================
 572 //        // <<< Some work-in-progress/explorative code that extracts a "representative" sequence >>>
 573 //        //
 574 //        // Currently need to know relevant hostname in advance :(
 575 //        String hostname = "events.tplinkra.com";
 576 ////        String hostname = "rfe-us-west-1.dch.dlink.com";
 577 //        // Conversations with 'hostname' for ON events.
 578 //        List<Conversation> onsForHostname = new ArrayList<>();
 579 //        // Conversations with 'hostname' for OFF events.
 580 //        List<Conversation> offsForHostname = new ArrayList<>();
 581 //        // "Unwrap" sequence groupings in ons/offs maps.
 582 //        ons.get(hostname).forEach((k,v) -> onsForHostname.addAll(v));
 583 //        offs.get(hostname).forEach((k,v) -> offsForHostname.addAll(v));
 584 //
 585 //
 586 //        Map<String, List<Conversation>> onsForHostnameGroupedByTlsAppDataSequence = TcpConversationUtils.groupConversationsByTlsApplicationDataPacketSequence(onsForHostname);
 587 //
 588 //
 589 //        // Extract representative sequence for ON and OFF by providing the list of conversations with
 590 //        // 'hostname' observed for each event type (the training data).
 591 //        SequenceExtraction seqExtraction = new SequenceExtraction();
 592 ////        ExtractedSequence extractedSequenceForOn = seqExtraction.extract(onsForHostname);
 593 ////        ExtractedSequence extractedSequenceForOff = seqExtraction.extract(offsForHostname);
 594 //
 595 //        ExtractedSequence extractedSequenceForOn = seqExtraction.extractByTlsAppData(onsForHostname);
 596 //        ExtractedSequence extractedSequenceForOff = seqExtraction.extractByTlsAppData(offsForHostname);
 597 //
 598 //        // Let's check how many ONs align with OFFs and vice versa (that is, how many times an event is incorrectly
 599 //        // labeled).
 600 //        int onsLabeledAsOff = 0;
 601 //        Integer[] representativeOnSeq = TcpConversationUtils.getPacketLengthSequence(extractedSequenceForOn.getRepresentativeSequence());
 602 //        Integer[] representativeOffSeq = TcpConversationUtils.getPacketLengthSequence(extractedSequenceForOff.getRepresentativeSequence());
 603 //        SequenceAlignment<Integer> seqAlg = seqExtraction.getAlignmentAlgorithm();
 604 //        for (Conversation c : onsForHostname) {
 605 //            Integer[] onSeq = TcpConversationUtils.getPacketLengthSequence(c);
 606 //            if (seqAlg.calculateAlignment(representativeOffSeq, onSeq) <= extractedSequenceForOff.getMaxAlignmentCost()) {
 607 //                onsLabeledAsOff++;
 608 //            }
 609 //        }
 610 //        int offsLabeledAsOn = 0;
 611 //        for (Conversation c : offsForHostname) {
 612 //            Integer[] offSeq = TcpConversationUtils.getPacketLengthSequence(c);
 613 //            if (seqAlg.calculateAlignment(representativeOnSeq, offSeq) <= extractedSequenceForOn.getMaxAlignmentCost()) {
 614 //                offsLabeledAsOn++;
 615 //            }
 616 //        }
 617 //        System.out.println("");
 618 //        // ================================================================================================
 619 //
 620 //
 621 //        // -------------------------------------------------------------------------------------------------------------
 622 //        // -------------------------------------------------------------------------------------------------------------
 623     }
 624
 625 }
 626
 627
 628 // TP-Link MAC 50:c7:bf:33:1f:09 and usually IP 192.168.1.159 (remember to verify per file)
 629 // frame.len >= 556 && frame.len <= 558 && ip.addr == 192.168.1.159