Merge branch 'master' of https://github.uci.edu/rtrimana/smart_home_traffic
[pingpong.git] / Code / Projects / SmartPlugDetector / src / main / java / edu / uci / iotproject / Main.java
index bf6dee6c61ea47628cfcf5efa6e9ec8b89984897..81eb3b0b9a96c833ae2db2870a4cca21316938da 100644 (file)
@@ -2,11 +2,14 @@ package edu.uci.iotproject;
 
 import static edu.uci.iotproject.analysis.UserAction.Type;
 
-import edu.uci.iotproject.analysis.TcpConversationUtils;
-import edu.uci.iotproject.analysis.TrafficLabeler;
-import edu.uci.iotproject.analysis.TriggerTrafficExtractor;
-import edu.uci.iotproject.analysis.UserAction;
+import edu.uci.iotproject.analysis.*;
 import edu.uci.iotproject.io.TriggerTimesFileReader;
+import edu.uci.iotproject.trafficreassembly.layer3.Conversation;
+import edu.uci.iotproject.trafficreassembly.layer3.TcpReassembler;
+import edu.uci.iotproject.util.PcapPacketUtils;
+import edu.uci.iotproject.util.PrintUtils;
+import org.apache.commons.math3.stat.clustering.Cluster;
+import org.apache.commons.math3.stat.clustering.DBSCANClusterer;
 import org.pcap4j.core.*;
 import org.pcap4j.packet.namednumber.DataLinkType;
 
@@ -15,6 +18,8 @@ import java.net.UnknownHostException;
 import java.time.Instant;
 import java.util.*;
 import java.util.concurrent.TimeoutException;
+import java.util.stream.Collectors;
+import java.util.stream.Stream;
 
 /**
  * This is a system that reads PCAP files to compare
@@ -35,61 +40,258 @@ public class Main {
         // Paths to input and output files (consider supplying these as arguments instead) and IP of the device for
         // which traffic is to be extracted:
         String path = "/scratch/July-2018"; // Rahmadi
-        //String path = "/Users/varmarken/temp/UCI IoT Project/experiments"; // Janus
+//        String path = "/Users/varmarken/temp/UCI IoT Project/experiments"; // Janus
+        boolean verbose = true;
+        final String onPairsPath = "/scratch/July-2018/on.txt";
+        final String offPairsPath = "/scratch/July-2018/off.txt";
 
-        // D-Link July 26 experiment
+        // 1) D-Link July 26 experiment
 //        final String inputPcapFile = path + "/2018-07/dlink/dlink.wlan1.local.pcap";
 //        final String outputPcapFile = path + "/2018-07/dlink/dlink-processed.pcap";
 //        final String triggerTimesFile = path + "/2018-07/dlink/dlink-july-26-2018.timestamps";
 //        final String deviceIp = "192.168.1.199"; // .246 == phone; .199 == dlink plug?
+          // Actual training
+//        final String inputPcapFile = path + "/2018-10/dlink-plug/dlink-plug.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/2018-10/dlink-plug/dlink-plug-processed.pcap";
+//        final String triggerTimesFile = path + "/2018-10/dlink-plug/dlink-plug-oct-17-2018.timestamps";
+//        final String deviceIp = "192.168.1.199"; // .246 == phone; .199 == dlink plug?
+        // TODO: EXPERIMENT - November 7, 2018
+//        final String inputPcapFile = path + "/experimental_result/standalone/dlink-plug/wlan1/dlink-plug.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/experimental_result/standalone/dlink-plug/wlan1/dlink-plug-processed.pcap";
+//        final String triggerTimesFile = path + "/experimental_result/standalone/dlink-plug/timestamps/dlink-plug-nov-7-2018.timestamps";
+//        //final String deviceIp = "192.168.1.199"; // .246 == phone; .199 == dlink plug?
+//        final String deviceIp = "192.168.1.246"; // .246 == phone; .199 == dlink plug?
 
-        // TP-Link July 25 experiment
+        // 2) TP-Link July 25 experiment
 //        final String inputPcapFile = path + "/2018-07/tplink/tplink.wlan1.local.pcap";
 //        final String outputPcapFile = path + "/2018-07/tplink/tplink-processed.pcap";
 //        final String triggerTimesFile = path + "/2018-07/tplink/tplink-july-25-2018.timestamps";
+//        final String deviceIp = "192.168.1.159";
+          // Actual training
+//        final String inputPcapFile = path + "/2018-10/tplink-plug/tplink-plug.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/2018-10/tplink-plug/tplink-plug-processed.pcap";
+//        final String triggerTimesFile = path + "/2018-10/tplink-plug/tplink-plug-oct-17-2018.timestamps";
+//        final String deviceIp = "192.168.1.246"; // .246 == phone; .159 == tplink plug
+        // TODO: EXPERIMENT - November 8, 2018
+//        final String inputPcapFile = path + "/experimental_result/standalone/tplink-plug/wlan1/tplink-plug.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/experimental_result/standalone/tplink-plug/wlan1/tplink-plug-processed.pcap";
+//        final String triggerTimesFile = path + "/experimental_result/standalone/tplink-plug/timestamps/tplink-plug-nov-8-2018.timestamps";
+//        final String deviceIp = "192.168.1.159"; // .246 == phone; .159 == tplink plug
+////        final String deviceIp = "192.168.1.246"; // .246 == phone; .159 == tplink plug
+
+        // 2b) TP-Link July 25 experiment TRUNCATED:
+        // Only contains "true local" events, i.e., before the behavior changes to remote-like behavior.
+        // Last included event is at July 25 10:38:11; file filtered to only include packets with arrival time <= 10:38:27.
+//        final String inputPcapFile = path + "/2018-07/tplink/tplink.wlan1.local.truncated.pcap";
+//        final String outputPcapFile = path + "/2018-07/tplink/tplink-processed.truncated.pcap";
+//        final String triggerTimesFile = path + "/2018-07/tplink/tplink-july-25-2018.truncated.timestamps";
 //        final String deviceIp = "192.168.1.159";
 
-        // SmartThings Plug July 25 experiment
+        // 3) SmartThings Plug July 25 experiment
 //        final String inputPcapFile = path + "/2018-07/stplug/stplug.wlan1.local.pcap";
 //        final String outputPcapFile = path + "/2018-07/stplug/stplug-processed.pcap";
 //        final String triggerTimesFile = path + "/2018-07/stplug/smartthings-july-25-2018.timestamps";
+//        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
+        // October 18
+//        final String inputPcapFile = path + "/2018-10/st-plug/st-plug.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/2018-10/st-plug/st-plug-processed.pcap";
+//        final String triggerTimesFile = path + "/2018-10/st-plug/st-plug-oct-18-2018.timestamps";
+//        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
+        // TODO: EXPERIMENT - November 12, 2018
+//        final String inputPcapFile = path + "/experimental_result/standalone/st-plug/wlan1/st-plug.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/experimental_result/standalone/st-plug/wlan1/st-plug-processed.pcap";
+////        final String inputPcapFile = path + "/experimental_result/standalone/st-plug/eth1/st-plug.eth1.local.pcap";
+////        final String outputPcapFile = path + "/experimental_result/standalone/st-plug/eth1/st-plug-processed.pcap";
+//        final String triggerTimesFile = path + "/experimental_result/standalone/st-plug/timestamps/st-plug-nov-12-2018.timestamps";
+////        final String deviceIp = "192.168.1.142"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
 
-        // Wemo July 30 experiment
+        // 4) Wemo July 30 experiment
 //        final String inputPcapFile = path + "/2018-07/wemo/wemo.wlan1.local.pcap";
 //        final String outputPcapFile = path + "/2018-07/wemo/wemo-processed.pcap";
 //        final String triggerTimesFile = path + "/2018-07/wemo/wemo-july-30-2018.timestamps";
-//        final String deviceIp = "192.168.1.145";
+//        final String deviceIp = "192.168.1.145";  // .246 == phone; .145 == WeMo
+        // TODO: EXPERIMENT - November 20, 2018
+//        final String inputPcapFile = path + "/experimental_result/standalone/wemo-plug/wlan1/wemo-plug.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/experimental_result/standalone/wemo-plug/wlan1/wemo-plug-processed.pcap";
+//        final String triggerTimesFile = path + "/experimental_result/standalone/wemo-plug/timestamps/wemo-plug-nov-20-2018.timestamps";
+////        final String deviceIp = "192.168.1.145"; // .246 == phone; .145 == WeMo
+//        final String deviceIp = "192.168.1.246"; // .246 == phone; .145 == WeMo
 
-        // Wemo Insight July 31 experiment
+        // 5) Wemo Insight July 31 experiment
 //        final String inputPcapFile = path + "/2018-07/wemoinsight/wemoinsight.wlan1.local.pcap";
 //        final String outputPcapFile = path + "/2018-07/wemoinsight/wemoinsight-processed.pcap";
 //        final String triggerTimesFile = path + "/2018-07/wemoinsight/wemo-insight-july-31-2018.timestamps";
 //        final String deviceIp = "192.168.1.135";
+        // TODO: EXPERIMENT - November 21, 2018
+//        final String inputPcapFile = path + "/experimental_result/standalone/wemo-insight-plug/wlan1/wemo-insight-plug.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/experimental_result/standalone/wemo-insight-plug/wlan1/wemo-insight-plug-processed.pcap";
+//        final String triggerTimesFile = path + "/experimental_result/standalone/wemo-insight-plug/timestamps/wemo-insight-plug-nov-21-2018.timestamps";
+////        final String deviceIp = "192.168.1.145"; // .246 == phone; .135 == WeMo Insight
+//        final String deviceIp = "192.168.1.246"; // .246 == phone; .135 == WeMo Insight
 
-        // TP-Link BULB August 1 experiment
+        // 6) TP-Link Bulb August 1 experiment
 //        final String inputPcapFile = path + "/2018-08/tplink-bulb/tplinkbulb.wlan1.local.pcap";
 //        final String outputPcapFile = path + "/2018-08/tplink-bulb/tplinkbulb-processed.pcap";
 //        final String triggerTimesFile = path + "/2018-08/tplink-bulb/tplink-bulb-aug-3-2018.timestamps";
-//        final String deviceIp = "192.168.1.140";
+//        final String deviceIp = "192.168.1.246"; // .246 == phone; .140 == TP-Link bulb
+        // TODO: EXPERIMENT - November 16, 2018
+//        final String inputPcapFile = path + "/experimental_result/standalone/tplink-bulb/wlan1/tplink-bulb.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/experimental_result/standalone/tplink-bulb/wlan1/tplink-bulb-processed.pcap";
+////        final String inputPcapFile = path + "/experimental_result/standalone/tplink-bulb/eth0/tplink-bulb.eth1.local.pcap";
+////        final String outputPcapFile = path + "/experimental_result/standalone/tplink-bulb/eth0/tplink-bulb-processed.pcap";
+//        final String triggerTimesFile = path + "/experimental_result/standalone/tplink-bulb/timestamps/tplink-bulb-nov-16-2018.timestamps";
+////        final String deviceIp = "192.168.1.140"; // .246 == phone; .140 == TP-Link bulb
+//        final String deviceIp = "192.168.1.246"; // .246 == phone; .140 == TP-Link bulb
 
-        // Kwikset Doorlock August 6 experiment
-//        final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock.wlan1.local.pcap";
+        // 7) Kwikset Doorlock August 6 experiment
+//        final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock.data.wlan1.pcap";
+////        final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock.wlan1.local.pcap";
 //        final String outputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-processed.pcap";
-//        final String triggerTimesFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-aug-6-2018.timestamps";
+////        final String triggerTimesFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-aug-6-2018.timestamps";
+//        final String triggerTimesFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-8hr-data-oct-11-2018.timestamps";
+//        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
+        // TODO: EXPERIMENT - November 10, 2018
+//        final String inputPcapFile = path + "/experimental_result/standalone/kwikset-doorlock/wlan1/kwikset-doorlock.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/experimental_result/standalone/kwikset-doorlock/wlan1/kwikset-doorlock-processed.pcap";
+////        final String inputPcapFile = path + "/experimental_result/standalone/kwikset-doorlock/eth1/kwikset-doorlock.eth1.local.pcap";
+////        final String outputPcapFile = path + "/experimental_result/standalone/kwikset-doorlock/eth1/kwikset-doorlock-processed.pcap";
+//        final String triggerTimesFile = path + "/experimental_result/standalone/kwikset-doorlock/timestamps/kwikset-doorlock-nov-10-2018.timestamps";
+////        final String deviceIp = "192.168.1.142"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
+//        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
+
+        // September 12, 2018 - includes both wlan1 and eth1 interfaces
+//        final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset3.wlan1.local.pcap";
+//        //final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset3.eth1.local.pcap";
+//        final String outputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset3-processed.pcap";
+//        final String triggerTimesFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-sept-12-2018.timestamps";
 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
 
-        // Hue BULB August 7 experiment
+        // 8) Hue Bulb August 7 experiment
 //        final String inputPcapFile = path + "/2018-08/hue-bulb/hue-bulb.wlan1.local.pcap";
 //        final String outputPcapFile = path + "/2018-08/hue-bulb/hue-bulb-processed.pcap";
 //        final String triggerTimesFile = path + "/2018-08/hue-bulb/hue-bulb-aug-7-2018.timestamps";
 //        final String deviceIp = "192.168.1.246";
+        // October 30 experiment
+//        final String inputPcapFile = path + "/2018-10/hue-bulb/hue-bulb.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/2018-10/hue-bulb/hue-bulb-processed.pcap";
+//        final String triggerTimesFile = path + "/2018-10/hue-bulb/hue-bulb-oct-30-2018.timestamps";
+//        final String deviceIp = "192.168.1.246"; // .246 == phone; .100 == Hue hub
+        // TODO: EXPERIMENT - November 19, 2018
+//        final String inputPcapFile = path + "/experimental_result/standalone/hue-bulb/wlan1/hue-bulb.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/experimental_result/standalone/hue-bulb/wlan1/hue-bulb-processed.pcap";
+//        final String triggerTimesFile = path + "/experimental_result/standalone/hue-bulb/timestamps/hue-bulb-nov-19-2018.timestamps";
+//        final String deviceIp = "192.168.1.100"; // .246 == phone; .100 == Hue hub
+//        final String deviceIp = "192.168.1.246"; // .246 == phone; .100 == Hue hub
+
+        // 9) Lifx Bulb August 8 experiment
+//        final String inputPcapFile = path + "/2018-08/lifx-bulb/lifx-bulb.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/2018-08/lifx-bulb/lifx-bulb-processed.pcap";
+//        final String triggerTimesFile = path + "/2018-08/lifx-bulb/lifx-bulb-aug-8-2018.timestamps";
+//        final String deviceIp = "192.168.1.246"; // .246 == phone; .231 == Lifx
+        // October 18
+//        final String inputPcapFile = path + "/2018-10/lifx-bulb/lifx-bulb.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/2018-10/lifx-bulb/lifx-bulb-processed.pcap";
+//        final String triggerTimesFile = path + "/2018-10/lifx-bulb/lifx-bulb-oct-18-2018.timestamps";
+//        final String deviceIp = "192.168.1.231"; // .246 == phone; .231 == Lifx
+        // November 1
+//        final String inputPcapFile = path + "/2018-10/lifx-bulb/lifx-bulb.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/2018-10/lifx-bulb/lifx-bulb-processed.pcap";
+//        final String triggerTimesFile = path + "/2018-10/lifx-bulb/lifx-bulb-nov-1-2018.timestamps";
+//        final String deviceIp = "192.168.1.231"; // .246 == phone; .231 == Lifx
+
+        // 10) Amcrest Camera August 9 experiment
+//        final String inputPcapFile = path + "/2018-08/amcrest-camera/amcrest-camera.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/2018-08/amcrest-camera/amcrest-camera-processed.pcap";
+//        final String triggerTimesFile = path + "/2018-08/amcrest-camera/amcrest-camera-aug-9-2018.timestamps";
+//        final String deviceIp = "192.168.1.246"; // .246 == phone; .235 == camera
+
+        // 11) Arlo Camera August 10 experiment
+//        final String inputPcapFile = path + "/2018-08/arlo-camera/arlo-camera.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/2018-08/arlo-camera/arlo-camera-processed.pcap";
+//        final String triggerTimesFile = path + "/2018-08/arlo-camera/arlo-camera-aug-10-2018.timestamps";
+//        final String deviceIp = "192.168.1.246"; // .246 == phone; .140 == camera
+        // TODO: EXPERIMENT - November 13, 2018
+//        final String inputPcapFile = path + "/experimental_result/standalone/arlo-camera/wlan1/arlo-camera.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/experimental_result/standalone/arlo-camera/wlan1/arlo-camera-processed.pcap";
+////        final String inputPcapFile = path + "/experimental_result/standalone/arlo-camera/eth0/arlo-camera.eth1.local.pcap";
+////        final String outputPcapFile = path + "/experimental_result/standalone/arlo-camera/eth0/arlo-camera-processed.pcap";
+//        final String triggerTimesFile = path + "/experimental_result/standalone/arlo-camera/timestamps/arlo-camera-nov-13-2018.timestamps";
+////        final String deviceIp = "192.168.1.140"; // .246 == phone; .140 == camera
+//        final String deviceIp = "192.168.1.246"; // .246 == phone; .140 == camera
+
+        // 12) Blossom sprinkler August 13 experiment
+//        final String inputPcapFile = path + "/2018-08/blossom/blossom.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/2018-08/blossom/blossom-processed.pcap";
+//        final String triggerTimesFile = path + "/2018-08/blossom/blossom-aug-13-2018.timestamps";
+//        final String deviceIp = "192.168.1.229"; // .246 == phone; .229 == sprinkler
+//        // 2 November
+//        final String inputPcapFile = path + "/2018-10/blossom-sprinkler/blossom-sprinkler.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/2018-10/blossom-sprinkler/blossom-sprinkler-processed.pcap";
+//        final String triggerTimesFile = path + "/2018-10/blossom-sprinkler/blossom-sprinkler-nov-2-2018.timestamps";
+//        final String deviceIp = "192.168.1.229"; // .246 == phone; .229 == sprinkler
+        // January 9, 11
+        final String inputPcapFile = path + "/experimental_result/standalone/blossom-sprinkler/wlan1/blossom-sprinkler.wlan1.local.pcap";
+        final String outputPcapFile = path + "/experimental_result/standalone/blossom-sprinkler/wlan1/blossom-sprinkler-processed.pcap";
+        final String triggerTimesFile = path + "/experimental_result/standalone/blossom-sprinkler/timestamps/blossom-sprinkler-standalone-jan-9-2019.timestamps";
+//        final String triggerTimesFile = path + "/experimental_result/standalone/blossom-sprinkler/timestamps/blossom-sprinkler-standalone-jan-11-2019.timestamps";
+        final String deviceIp = "192.168.1.246"; // .246 == phone; .229 == sprinkler
+
+//        // 13) DLink siren August 14 experiment
+//        final String inputPcapFile = path + "/2018-08/dlink-siren/dlink-siren.wlan1.local.pcap";
+//        //final String inputPcapFile = path + "/evaluation/dlink-siren/dlink-siren.data.wlan1.pcap";
+//        final String outputPcapFile = path + "/2018-08/dlink-siren/dlink-siren-processed.pcap";
+//        final String triggerTimesFile = path + "/2018-08/dlink-siren/dlink-siren-oct-12-2018.timestamps";
+//        //final String triggerTimesFile = path + "/2018-08/dlink-siren/dlink-siren-aug-14-2018.timestamps";
+//        //final String triggerTimesFile = path + "/actual/timestamps/dlink-siren-8hr-data-oct-10-2018.timestamps";
+//        final String deviceIp = "192.168.1.246"; // .246 == phone; .183 == siren
+        // TODO: EXPERIMENT - November 9, 2018
+//        final String inputPcapFile = path + "/experimental_result/standalone/dlink-siren/wlan1/dlink-siren.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/experimental_result/standalone/dlink-siren/wlan1/dlink-siren-processed.pcap";
+//        final String triggerTimesFile = path + "/experimental_result/standalone/dlink-siren/timestamps/dlink-siren-nov-9-2018.timestamps";
+//        final String deviceIp = "192.168.1.183"; // .246 == phone; .183 == siren
+////        final String deviceIp = "192.168.1.246"; // .246 == phone; .183 == siren
+
+        // 14) Nest thermostat August 15 experiment
+//        final String inputPcapFile = path + "/2018-08/nest/nest.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/2018-08/nest/nest-processed.pcap";
+//        final String triggerTimesFile = path + "/2018-08/nest/nest-aug-15-2018.timestamps";
+//        final String deviceIp = "192.168.1.246"; // .246 == phone; .127 == Nest thermostat
+        // TODO: EXPERIMENT - November 14, 2018
+//        final String inputPcapFile = path + "/experimental_result/standalone/nest-thermostat/wlan1/nest-thermostat.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/experimental_result/standalone/nest-thermostat/wlan1/nest-thermostat-processed.pcap";
+////        final String inputPcapFile = path + "/experimental_result/standalone/nest-thermostat/eth0/nest-thermostat.eth1.local.pcap";
+////        final String outputPcapFile = path + "/experimental_result/standalone/nest-thermostat/eth0/nest-thermostat-processed.pcap";
+//        final String triggerTimesFile = path + "/experimental_result/standalone/nest-thermostat/timestamps/nest-thermostat-nov-15-2018.timestamps";
+////        final String deviceIp = "192.168.1.127"; // .246 == phone; .127 == Nest thermostat
+//        final String deviceIp = "192.168.1.246"; // .246 == phone; .127 == Nest thermostat
 
-        // Lifx BULB August 8 experiment
-        final String inputPcapFile = path + "/2018-08/lifx-bulb/lifx-bulb.wlan1.local.pcap";
-        final String outputPcapFile = path + "/2018-08/lifx-bulb/lifx-bulb-processed.pcap";
-        final String triggerTimesFile = path + "/2018-08/lifx-bulb/lifx-bulb-aug-8-2018.timestamps";
-        final String deviceIp = "192.168.1.246"; // .246 == phone; .231 == Lifx
+        // 15) Alexa August 16 experiment
+//        final String inputPcapFile = path + "/2018-08/alexa/alexa.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/2018-08/alexa/alexa-processed.pcap";
+//        final String triggerTimesFile = path + "/2018-08/alexa/alexa-aug-16-2018.timestamps";
+//        final String deviceIp = "192.168.1.225"; // .246 == phone; .225 == Alexa
+        // August 17
+//        final String inputPcapFile = path + "/2018-08/alexa/alexa2.wlan1.local.pcap";
+//        final String outputPcapFile = path + "/2018-08/alexa/alexa2-processed.pcap";
+//        final String triggerTimesFile = path + "/2018-08/alexa/alexa-aug-17-2018.timestamps";
+//        final String deviceIp = "192.168.1.225"; // .246 == phone; .225 == Alexa
+
+        // September 17
+//        final String inputPcapFile = path + "/2018-08/noise/noise.eth1.pcap";
+//        final String outputPcapFile = path + "/2018-08/noise/noise-processed.pcap";
+//        final String triggerTimesFile = path + "/2018-08/noise/noise-sept-17-2018.timestamps";
+//        final String deviceIp = "192.168.1.142"; //  .142 == SmartThings Hub; .199 == dlink plug; .183 == siren
+        // September 26 - D-Link noise
+//        final String inputPcapFile = path + "/2018-08/noise/noise.dlink.wlan1.pcap";
+//        final String outputPcapFile = path + "/2018-08/noise/noise-processed.pcap";
+//        final String triggerTimesFile = path + "/2018-08/noise/dlink-noise-sept-26-2018.timestamps";
+//        final String deviceIp = "192.168.1.183"; //  .199 == dlink plug; .183 == siren
+        // September 27 - Kwikset noise
+//        final String inputPcapFile = path + "/2018-08/noise/noise.kwikset.eth1.pcap";
+//        final String outputPcapFile = path + "/2018-08/noise/noise-processed.pcap";
+//        final String triggerTimesFile = path + "/2018-08/noise/kwikset-doorlock-noise-sept-27-2018.timestamps";
+//        final String deviceIp = "192.168.1.142"; //  .142 == SmartThings Hub;
 
         TriggerTimesFileReader ttfr = new TriggerTimesFileReader();
         List<Instant> triggerTimes = ttfr.readTriggerTimes(triggerTimesFile, false);
@@ -144,29 +346,278 @@ public class Main {
 
 
 
-        // Contains all ON events: hostname -> sequence identifier -> list of conversations with that sequence
-        Map<String, Map<String, List<Conversation>>> ons = new HashMap<>();
-        // Contains all OFF events: hostname -> sequence identifier -> list of conversations with that sequence
-        Map<String, Map<String, List<Conversation>>> offs = new HashMap<>();
-        userActionsToConvsByHostname.forEach((ua, hostnameToConvs) -> {
-            Map<String, Map<String, List<Conversation>>> outer = ua.getType() == Type.TOGGLE_ON ? ons : offs;
-            hostnameToConvs.forEach((host, convs) -> {
-                Map<String, List<Conversation>> seqsToConvs = TcpConversationUtils.
-                        groupConversationsByPacketSequence(convs);
-                outer.merge(host, seqsToConvs, (oldMap, newMap) -> {
-                    newMap.forEach((sequence, cs) -> oldMap.merge(sequence, cs, (list1, list2) -> {
-                        list1.addAll(list2);
-                        return list1;
-                    }));
-                    return oldMap;
-                });
-            });
-        });
-
-        System.out.println("");
+        /*
+         * NOTE: no need to generate these more complex on/off maps that also contain mappings from hostname and
+         * sequence identifiers as we do not care about hostnames and sequences during clustering.
+         * We can simply use the UserAction->List<Conversation> map to generate ON/OFF groupings of conversations.
+         */
 
-        // -------------------------------------------------------------------------------------------------------------
-        // -------------------------------------------------------------------------------------------------------------
+//        // Contains all ON events: hostname -> sequence identifier -> list of conversations with that sequence
+//        Map<String, Map<String, List<Conversation>>> ons = new HashMap<>();
+//        // Contains all OFF events: hostname -> sequence identifier -> list of conversations with that sequence
+//        Map<String, Map<String, List<Conversation>>> offs = new HashMap<>();
+//        userActionsToConvsByHostname.forEach((ua, hostnameToConvs) -> {
+//            Map<String, Map<String, List<Conversation>>> outer = ua.getType() == Type.TOGGLE_ON ? ons : offs;
+//            hostnameToConvs.forEach((host, convs) -> {
+//                Map<String, List<Conversation>> seqsToConvs = TcpConversationUtils.
+//                        groupConversationsByPacketSequence(convs, verbose);
+//                outer.merge(host, seqsToConvs, (oldMap, newMap) -> {
+//                    newMap.forEach((sequence, cs) -> oldMap.merge(sequence, cs, (list1, list2) -> {
+//                        list1.addAll(list2);
+//                        return list1;
+//                    }));
+//                    return oldMap;
+//                });
+//            });
+//        });
+//
+//        System.out.println("==== ON ====");
+//        // Print out all the pairs into a file for ON events
+//        File fileOnEvents = new File(onPairsPath);
+//        PrintWriter pwOn = null;
+//        try {
+//            pwOn = new PrintWriter(fileOnEvents);
+//        } catch(Exception ex) {
+//            ex.printStackTrace();
+//        }
+//        for(Map.Entry<String, Map<String, List<Conversation>>> entry : ons.entrySet()) {
+//            Map<String, List<Conversation>> seqsToConvs = entry.getValue();
+//            for(Map.Entry<String, List<Conversation>> entryConv : seqsToConvs.entrySet()) {
+//                List<Conversation> listConv = entryConv.getValue();
+//                // Just get the first Conversation because all Conversations in this group
+//                // should have the same pairs of Application Data.
+//                for(Conversation conv : listConv) {
+//                    // Process only if it is a TLS packet
+//                    if (conv.isTls()) {
+//                        List<PcapPacketPair> tlsAppDataList = TcpConversationUtils.extractTlsAppDataPacketPairs(conv);
+//                        for(PcapPacketPair pair: tlsAppDataList) {
+//                            System.out.println(PrintUtils.toCsv(pair, dnsMap));
+//                            pwOn.println(PrintUtils.toCsv(pair, dnsMap));
+//                        }
+//                    } else { // Non-TLS conversations
+//                        List<PcapPacketPair> packetList = TcpConversationUtils.extractPacketPairs(conv);
+//                        for(PcapPacketPair pair: packetList) {
+//                            System.out.println(PrintUtils.toCsv(pair, dnsMap));
+//                            pwOn.println(PrintUtils.toCsv(pair, dnsMap));
+//                        }
+//                    }
+//                }
+//            }
+//        }
+//        pwOn.close();
+//
+//        System.out.println("==== OFF ====");
+//        // Print out all the pairs into a file for ON events
+//        File fileOffEvents = new File(offPairsPath);
+//        PrintWriter pwOff = null;
+//        try {
+//            pwOff = new PrintWriter(fileOffEvents);
+//        } catch(Exception ex) {
+//            ex.printStackTrace();
+//        }
+//        for(Map.Entry<String, Map<String, List<Conversation>>> entry : offs.entrySet()) {
+//            Map<String, List<Conversation>> seqsToConvs = entry.getValue();
+//            for(Map.Entry<String, List<Conversation>> entryConv : seqsToConvs.entrySet()) {
+//                List<Conversation> listConv = entryConv.getValue();
+//                // Just get the first Conversation because all Conversations in this group
+//                // should have the same pairs of Application Data.
+//                for(Conversation conv : listConv) {
+//                    // Process only if it is a TLS packet
+//                    if (conv.isTls()) {
+//                        List<PcapPacketPair> tlsAppDataList = TcpConversationUtils.extractTlsAppDataPacketPairs(conv);
+//                        for(PcapPacketPair pair: tlsAppDataList) {
+//                            System.out.println(PrintUtils.toCsv(pair, dnsMap));
+//                            pwOff.println(PrintUtils.toCsv(pair, dnsMap));
+//                        }
+//                    } else { // Non-TLS conversations
+//                        List<PcapPacketPair> packetList = TcpConversationUtils.extractPacketPairs(conv);
+//                        for (PcapPacketPair pair : packetList) {
+//                            System.out.println(PrintUtils.toCsv(pair, dnsMap));
+//                            pwOff.println(PrintUtils.toCsv(pair, dnsMap));
+//                        }
+//                    }
+//                }
+//            }
+//        }
+//        pwOff.close();
+
+
+        // ================================================ CLUSTERING ================================================
+        // Note: no need to use the more convoluted on/off maps; can simply use the UserAction->List<Conversation> map
+        // when don't care about hostnames and sequences (see comment earlier).
+        List<Conversation> onConversations = userActionToConversations.entrySet().stream().
+                filter(e -> e.getKey().getType() == Type.TOGGLE_ON). // drop all OFF events from stream
+                map(e -> e.getValue()). // no longer interested in the UserActions
+                flatMap(List::stream). // flatten List<List<T>> to a List<T>
+                collect(Collectors.toList());
+        List<Conversation> offConversations = userActionToConversations.entrySet().stream().
+                filter(e -> e.getKey().getType() == Type.TOGGLE_OFF).
+                map(e -> e.getValue()).
+                flatMap(List::stream).
+                collect(Collectors.toList());
+        //Collections.sort(onConversations, (c1, c2) -> c1.getPackets().)
+
+        List<PcapPacketPair> onPairs = onConversations.stream().
+                map(c -> c.isTls() ? TcpConversationUtils.extractTlsAppDataPacketPairs(c) :
+                        TcpConversationUtils.extractPacketPairs(c)).
+                flatMap(List::stream). // flatten List<List<>> to List<>
+                collect(Collectors.toList());
+        List<PcapPacketPair> offPairs = offConversations.stream().
+                map(c -> c.isTls() ? TcpConversationUtils.extractTlsAppDataPacketPairs(c) :
+                        TcpConversationUtils.extractPacketPairs(c)).
+                flatMap(List::stream). // flatten List<List<>> to List<>
+                collect(Collectors.toList());
+        // Note: need to update the DnsMap of all PcapPacketPairs if we want to use the IP/hostname-sensitive distance.
+        Stream.concat(Stream.of(onPairs), Stream.of(offPairs)).flatMap(List::stream).forEach(p -> p.setDnsMap(dnsMap));
+        // Perform clustering on conversation logged as part of all ON events.
+        DBSCANClusterer<PcapPacketPair> onClusterer = new DBSCANClusterer<>(10.0, 45);
+        //DBSCANClusterer<PcapPacketPair> onClusterer = new DBSCANClusterer<>(7.0, 45);
+        //DBSCANClusterer<PcapPacketPair> onClusterer = new DBSCANClusterer<>(10.0, 10);
+        List<Cluster<PcapPacketPair>> onClusters = onClusterer.cluster(onPairs);
+        // Perform clustering on conversation logged as part of all OFF events.
+        DBSCANClusterer<PcapPacketPair> offClusterer = new DBSCANClusterer<>(10.0, 45);
+        //DBSCANClusterer<PcapPacketPair> offClusterer = new DBSCANClusterer<>(7.0, 45);
+        //DBSCANClusterer<PcapPacketPair> offClusterer = new DBSCANClusterer<>(10.0, 10);
+        List<Cluster<PcapPacketPair>> offClusters = offClusterer.cluster(offPairs);
+        // Sort the conversations as reference
+        List<Conversation> sortedAllConversation = TcpConversationUtils.sortConversationList(allConversations);
+        // Output clusters
+        System.out.println("========================================");
+        System.out.println("       Clustering results for ON        ");
+        System.out.println("       Number of clusters: " + onClusters.size());
+        int count = 0;
+        List<List<List<PcapPacket>>> ppListOfListReadOn = new ArrayList<>();
+        List<List<List<PcapPacket>>> ppListOfListListOn = new ArrayList<>();
+        for (Cluster<PcapPacketPair> c : onClusters) {
+            System.out.println(String.format("<<< Cluster #%02d (%03d points) >>>", ++count, c.getPoints().size()));
+            System.out.print(PrintUtils.toSummaryString(c));
+            if(c.getPoints().size() > 45 && c.getPoints().size() < 55) {
+            //if(c.getPoints().size() > 25) {
+                // Print to file
+                List<List<PcapPacket>> ppListOfList = PcapPacketUtils.clusterToListOfPcapPackets(c);
+                ppListOfListListOn.add(ppListOfList);
+            }
+        }
+        // TODO: Merging test
+        ppListOfListListOn = PcapPacketUtils.mergeSignatures(ppListOfListListOn, sortedAllConversation);
+        ppListOfListListOn = PcapPacketUtils.sortSignatures(ppListOfListListOn);
+        // TODO: Need to remove sequence numbers 0,1 for Blossom phone side since it is not a good signature!
+        PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 0);
+        PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 1);
+        // TODO: Need to remove sequence numbers 0 for Blossom device side since it is not a good signature!
+//        PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 0);
+//        PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 0);
+        // TODO: Need to remove sequence number 2 for ST plug since it is not a good signature!
+        //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 2);
+        // TODO: Need to remove sequence number 0 for Arlo Camera since it is not a good signature!
+        //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 0);
+        // TODO: Need to remove sequence number 0 for TP-Link plug since it is not a good signature!
+        // TODO: This sequence actually belongs to the local communication between the plug and the phone
+        //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 0);
+        PcapPacketUtils.printSignatures(ppListOfListListOn);
+        //count = 0;
+        /*for (List<List<PcapPacket>> ll : ppListOfListListOn) {
+            PrintUtils.serializeClustersIntoFile("./onSignature" + ++count + ".sig", ll);
+            ppListOfListReadOn.add(PrintUtils.deserializeClustersFromFile("./onSignature" + count + ".sig"));
+        }*/
+        PrintUtils.serializeSignatureIntoFile("./onSignature.sig", ppListOfListListOn);
+        ppListOfListReadOn = PrintUtils.deserializeSignatureFromFile("./onSignature.sig");
+
+        System.out.println("========================================");
+        System.out.println("       Clustering results for OFF       ");
+        System.out.println("       Number of clusters: " + offClusters.size());
+        count = 0;
+        List<List<List<PcapPacket>>> ppListOfListReadOff = new ArrayList<>();
+        List<List<List<PcapPacket>>> ppListOfListListOff = new ArrayList<>();
+        for (Cluster<PcapPacketPair> c : offClusters) {
+            System.out.println(String.format("<<< Cluster #%03d (%06d points) >>>", ++count, c.getPoints().size()));
+            System.out.print(PrintUtils.toSummaryString(c));
+            if(c.getPoints().size() > 45 && c.getPoints().size() < 55) {
+            //if(c.getPoints().size() > 25) {
+                // Print to file
+                List<List<PcapPacket>> ppListOfList = PcapPacketUtils.clusterToListOfPcapPackets(c);
+                ppListOfListListOff.add(ppListOfList);
+            }
+        }
+        // TODO: Merging test
+        ppListOfListListOff = PcapPacketUtils.mergeSignatures(ppListOfListListOff, sortedAllConversation);
+        ppListOfListListOff = PcapPacketUtils.sortSignatures(ppListOfListListOff);
+        // TODO: Need to remove sequence numbers 0,2 for Blossom device side since it is not a good signature!
+//        PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 0);
+//        PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 0);
+        // TODO: Need to remove sequence number 1 for Nest Thermostat since it is not a good signature!
+        //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 1);
+        // TODO: Need to remove sequence number 0 for Arlo Camera since it is not a good signature!
+        //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 1);
+        // TODO: Need to remove sequence number 2 for ST plug since it is not a good signature!
+        //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 2);
+        // TODO: Need to remove sequence number 0 for TP-Link plug since it is not a good signature!
+        // TODO: This sequence actually belongs to the local communication between the plug and the phone
+        //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 0);
+        PcapPacketUtils.printSignatures(ppListOfListListOff);
+        //count = 0;
+        /*for (List<List<PcapPacket>> ll : ppListOfListListOff) {
+            PrintUtils.serializeClustersIntoFile("./offSignature" + ++count + ".sig", ll);
+            ppListOfListReadOff.add(PrintUtils.deserializeClustersFromFile("./offSignature" + count + ".sig"));
+        }*/
+        PrintUtils.serializeSignatureIntoFile("./offSignature.sig", ppListOfListListOff);
+        ppListOfListReadOff = PrintUtils.deserializeSignatureFromFile("./offSignature.sig");
+        System.out.println("========================================");
+        // ============================================================================================================
+
+
+//        // ================================================================================================
+//        // <<< Some work-in-progress/explorative code that extracts a "representative" sequence >>>
+//        //
+//        // Currently need to know relevant hostname in advance :(
+//        String hostname = "events.tplinkra.com";
+////        String hostname = "rfe-us-west-1.dch.dlink.com";
+//        // Conversations with 'hostname' for ON events.
+//        List<Conversation> onsForHostname = new ArrayList<>();
+//        // Conversations with 'hostname' for OFF events.
+//        List<Conversation> offsForHostname = new ArrayList<>();
+//        // "Unwrap" sequence groupings in ons/offs maps.
+//        ons.get(hostname).forEach((k,v) -> onsForHostname.addAll(v));
+//        offs.get(hostname).forEach((k,v) -> offsForHostname.addAll(v));
+//
+//
+//        Map<String, List<Conversation>> onsForHostnameGroupedByTlsAppDataSequence = TcpConversationUtils.groupConversationsByTlsApplicationDataPacketSequence(onsForHostname);
+//
+//
+//        // Extract representative sequence for ON and OFF by providing the list of conversations with
+//        // 'hostname' observed for each event type (the training data).
+//        SequenceExtraction seqExtraction = new SequenceExtraction();
+////        ExtractedSequence extractedSequenceForOn = seqExtraction.extract(onsForHostname);
+////        ExtractedSequence extractedSequenceForOff = seqExtraction.extract(offsForHostname);
+//
+//        ExtractedSequence extractedSequenceForOn = seqExtraction.extractByTlsAppData(onsForHostname);
+//        ExtractedSequence extractedSequenceForOff = seqExtraction.extractByTlsAppData(offsForHostname);
+//
+//        // Let's check how many ONs align with OFFs and vice versa (that is, how many times an event is incorrectly
+//        // labeled).
+//        int onsLabeledAsOff = 0;
+//        Integer[] representativeOnSeq = TcpConversationUtils.getPacketLengthSequence(extractedSequenceForOn.getRepresentativeSequence());
+//        Integer[] representativeOffSeq = TcpConversationUtils.getPacketLengthSequence(extractedSequenceForOff.getRepresentativeSequence());
+//        SequenceAlignment<Integer> seqAlg = seqExtraction.getAlignmentAlgorithm();
+//        for (Conversation c : onsForHostname) {
+//            Integer[] onSeq = TcpConversationUtils.getPacketLengthSequence(c);
+//            if (seqAlg.calculateAlignment(representativeOffSeq, onSeq) <= extractedSequenceForOff.getMaxAlignmentCost()) {
+//                onsLabeledAsOff++;
+//            }
+//        }
+//        int offsLabeledAsOn = 0;
+//        for (Conversation c : offsForHostname) {
+//            Integer[] offSeq = TcpConversationUtils.getPacketLengthSequence(c);
+//            if (seqAlg.calculateAlignment(representativeOnSeq, offSeq) <= extractedSequenceForOn.getMaxAlignmentCost()) {
+//                offsLabeledAsOn++;
+//            }
+//        }
+//        System.out.println("");
+//        // ================================================================================================
+//
+//
+//        // -------------------------------------------------------------------------------------------------------------
+//        // -------------------------------------------------------------------------------------------------------------
     }
 
 }