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