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