Adding info about TP-Link bulb device IP.
[pingpong.git] / Code / Projects / SmartPlugDetector / src / main / java / edu / uci / iotproject / Main.java
index 0ddae8de57b209b1c04521486e08caf07201ca72..a855f473d92e0c703e2393f2250e1862f7830404 100644 (file)
@@ -1,12 +1,21 @@
 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.comparison.seqalignment.ExtractedSequence;
+import edu.uci.iotproject.comparison.seqalignment.SequenceAlignment;
+import edu.uci.iotproject.comparison.seqalignment.SequenceExtraction;
 import edu.uci.iotproject.io.TriggerTimesFileReader;
 import org.pcap4j.core.*;
 import org.pcap4j.packet.namednumber.DataLinkType;
 
 import java.io.EOFException;
+import java.io.File;
+import java.io.PrintWriter;
 import java.net.UnknownHostException;
 import java.time.Instant;
 import java.util.*;
@@ -28,29 +37,145 @@ public class Main {
     public static void main(String[] args) throws PcapNativeException, NotOpenException, EOFException, TimeoutException, UnknownHostException {
         // -------------------------------------------------------------------------------------------------------------
         // ------------ # Code for extracting traffic generated by a device within x seconds of a trigger # ------------
-        // Paths to input and output files (consider supplying these as arguments instead)
-        final String inputPcapFile = "/Users/varmarken/temp/UCI IoT Project/June2018 experiments/tplink/tplink.wlan1.local.pcap";
-        final String outputPcapFile = "/Users/varmarken/temp/UCI IoT Project/June2018 experiments/tplink/tplink-filtered.pcap";
-        final String triggerTimesFile = "/Users/varmarken/temp/UCI IoT Project/June2018 experiments/tplink/tplink-june-14-2018-timestamps.txt";
-        // IP of the device for which traffic is to be extracted
-        final String deviceIp = "192.168.1.159";
+        // 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
+        boolean verbose = true;
+        final String onPairsPath = "/scratch/July-2018/on.txt";
+        final String offPairsPath = "/scratch/July-2018/off.txt";
+
+        // 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?
+
+        // 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";
+
+        // 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";
+
+        // 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!)
+
+        // 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";
+
+        // 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";
+
+        // 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"; // .246 == phone; .140 == TP-Link bulb
+
+        // 7) Kwikset Doorlock August 6 experiment
+//        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 deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
+
+        // 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";
+
+        // 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
+
+        // 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 = "192.168.1.140"; // .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 = "192.168.1.246"; // .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 outputPcapFile = path + "/2018-08/dlink-siren/dlink-siren-processed.pcap";
+//        final String triggerTimesFile = path + "/2018-08/dlink-siren/dlink-siren-aug-14-2018.timestamps";
+//        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
+
+        // 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
 
         TriggerTimesFileReader ttfr = new TriggerTimesFileReader();
         List<Instant> triggerTimes = ttfr.readTriggerTimes(triggerTimesFile, false);
+        // Tag each trigger with "ON" or "OFF", assuming that the first trigger is an "ON" and that they alternate.
+        List<UserAction> userActions = new ArrayList<>();
+        for (int i = 0; i < triggerTimes.size(); i++) {
+            userActions.add(new UserAction(i % 2 == 0 ? Type.TOGGLE_ON : Type.TOGGLE_OFF, triggerTimes.get(i)));
+        }
         TriggerTrafficExtractor tte = new TriggerTrafficExtractor(inputPcapFile, triggerTimes, deviceIp);
         final PcapDumper outputter = Pcaps.openDead(DataLinkType.EN10MB, 65536).dumpOpen(outputPcapFile);
         DnsMap dnsMap = new DnsMap();
         TcpReassembler tcpReassembler = new TcpReassembler();
+        TrafficLabeler trafficLabeler = new TrafficLabeler(userActions);
         tte.performExtraction(pkt -> {
             try {
                 outputter.dump(pkt);
             } catch (NotOpenException e) {
                 e.printStackTrace();
             }
-        }, dnsMap, tcpReassembler);
+        }, dnsMap, tcpReassembler, trafficLabeler);
         outputter.flush();
         outputter.close();
 
+        if (tte.getPacketsIncludedCount() != trafficLabeler.getTotalPacketCount()) {
+            // Sanity/debug check
+            throw new AssertionError(String.format("mismatch between packet count in %s and %s",
+                    TriggerTrafficExtractor.class.getSimpleName(), TrafficLabeler.class.getSimpleName()));
+        }
+
         // Extract all conversations present in the filtered trace.
         List<Conversation> allConversations = tcpReassembler.getTcpConversations();
         // Group conversations by hostname.
@@ -65,14 +190,247 @@ public class Main {
         final Map<String, Map<String, Integer>> pktSeqFreqsByHostname = new HashMap<>();
         convsByHostname.forEach((host, convs) -> pktSeqFreqsByHostname.put(host, TcpConversationUtils.countPacketSequenceFrequencies(convs)));
         System.out.println("Counted frequencies of packet sequences exchanged with each hostname.");
+        // For each hostname, count frequencies of packet pairs exchanged with that hostname across all conversations
+        final Map<String, Map<String, Integer>> pktPairFreqsByHostname =
+                TcpConversationUtils.countPacketPairFrequenciesByHostname(allConversations, dnsMap);
+        System.out.println("Counted frequencies of packet pairs per hostname");
+        // For each user action, reassemble the set of TCP connections occurring shortly after
+        final Map<UserAction, List<Conversation>> userActionToConversations = trafficLabeler.getLabeledReassembledTcpTraffic();
+        final Map<UserAction, Map<String, List<Conversation>>> userActionsToConvsByHostname = trafficLabeler.getLabeledReassembledTcpTraffic(dnsMap);
+        System.out.println("Reassembled TCP conversations occurring shortly after each user event");
 
 
 
-        // -------------------------------------------------------------------------------------------------------------
-        // -------------------------------------------------------------------------------------------------------------
+        // 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;
+                });
+            });
+        });
+
+
+        // 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<PcapPacket> tlsAppDataList = conv.getTlsApplicationDataPackets();
+                        // Loop and print out packets
+                        int count = 0;
+                        // The direction of the first packet
+                        Conversation.Direction firstDir = null;
+                        // The length of the first packet
+                        int firstLen = 0;
+                        for (PcapPacket pcap : tlsAppDataList) {
+                            boolean isPair = false;
+                            if (count % 2 == 0) {
+                                firstDir = conv.getDirection(pcap);
+                                firstLen = pcap.length();
+                            } else {// count%2 == 1
+                                if(conv.getDirection(pcap) != firstDir) {
+                                    isPair = true;
+                                    pwOn.println(firstLen + ", " + pcap.length());
+                                    //System.out.println(firstDir + ", " + conv.getDirection(pcap));
+                                    //System.out.println(firstLen + ", " + pcap.length());
+                                }
+                            }
+                            count++;
+                            // If we can't create a pair then just pad it with 0
+                            if (count == tlsAppDataList.size() && !isPair) {
+                                pwOn.println(firstLen + ", 0");
+                            }
+                        }
+                    } else { // Non-TLS conversations
+                        List<PcapPacket> packetList = conv.getPackets();
+                        // Loop and print out packets
+                        int count = 0;
+                        // The direction of the first packet
+                        Conversation.Direction firstDir = null;
+                        // The length of the first packet
+                        int firstLen = 0;
+                        for (PcapPacket pcap : packetList) {
+                            boolean isPair = false;
+                            if (count % 2 == 0) {
+                                firstDir = conv.getDirection(pcap);
+                                firstLen = pcap.length();
+                            } else {// count%2 == 1
+                                if(conv.getDirection(pcap) != firstDir) {
+                                    isPair = true;
+                                    pwOn.println(firstLen + ", " + pcap.length());
+                                    System.out.println(firstDir + ", " + conv.getDirection(pcap));
+                                    System.out.println(firstLen + ", " + pcap.length());
+                                }
+                            }
+                            count++;
+                            // If we can't create a pair then just pad it with 0
+                            if (count == packetList.size() && !isPair) {
+                                pwOn.println(firstLen + ", 0");
+                            }
+                        }
+                    }
+                }
+            }
+        }
+        pwOn.close();
+
+        // 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<PcapPacket> tlsAppDataList = conv.getTlsApplicationDataPackets();
+                        // Loop and print out packets
+                        int count = 0;
+                        // The direction of the first packet
+                        Conversation.Direction firstDir = null;
+                        // The length of the first packet
+                        int firstLen = 0;
+                        for (PcapPacket pcap : tlsAppDataList) {
+                            boolean isPair = false;
+                            if (count % 2 == 0) {
+                                firstDir = conv.getDirection(pcap);
+                                firstLen = pcap.length();
+                            } else {// count%2 == 1
+                                if(conv.getDirection(pcap) != firstDir) {
+                                    isPair = true;
+                                    pwOff.println(firstLen + ", " + pcap.length());
+                                    System.out.println(firstDir + ", " + conv.getDirection(pcap));
+                                    System.out.println(firstLen + ", " + pcap.length());
+                                }
+                            }
+                            count++;
+                            // If we can't create a pair then just pad it with 0
+                            if (count == tlsAppDataList.size() && !isPair) {
+                                pwOff.println(firstLen + ", 0");
+                            }
+
+                        }
+                    } else { // Non-TLS conversations
+                        List<PcapPacket> packetList = conv.getPackets();
+                        // Loop and print out packets
+                        int count = 0;
+                        // The direction of the first packet
+                        Conversation.Direction firstDir = null;
+                        // The length of the first packet
+                        int firstLen = 0;
+                        for (PcapPacket pcap : packetList) {
+                            boolean isPair = false;
+                            if (count % 2 == 0) {
+                                firstDir = conv.getDirection(pcap);
+                                firstLen = pcap.length();
+                            } else {// count%2 == 1
+                                if(conv.getDirection(pcap) != firstDir) {
+                                    isPair = true;
+                                    pwOff.println(firstLen + ", " + pcap.length());
+                                    System.out.println(firstDir + ", " + conv.getDirection(pcap));
+                                    System.out.println(firstLen + ", " + pcap.length());
+                                }
+                            }
+                            count++;
+                            // If we can't create a pair then just pad it with 0
+                            if (count == packetList.size() && !isPair) {
+                                pwOff.println(firstLen + ", 0");
+                            }
+                        }
+                    }
+                }
+            }
+        }
+        pwOff.close();
+
+
+//        // ================================================================================================
+//        // <<< 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("");
+//        // ================================================================================================
+//
+//
+//        // -------------------------------------------------------------------------------------------------------------
+//        // -------------------------------------------------------------------------------------------------------------
     }
 
 }
 
 
-// TP-Link MAC 50:c7:bf:33:1f:09 and usually IP 192.168.1.159 (remember to verify per file)
\ No newline at end of file
+// TP-Link MAC 50:c7:bf:33:1f:09 and usually IP 192.168.1.159 (remember to verify per file)
+// frame.len >= 556 && frame.len <= 558 && ip.addr == 192.168.1.159
\ No newline at end of file