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