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 d82e134bc9e351851873a3e2d5f51471b0d68486..6abdfc8c202608ca6c0d3efb34e7718952d06b16 100644 (file)
@@ -1,14 +1,18 @@
 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.io.TriggerTimesFileReader;
 import org.pcap4j.core.*;
-import org.pcap4j.packet.*;
-import org.pcap4j.packet.DnsPacket;
-import org.pcap4j.packet.namednumber.DnsResourceRecordType;
+import org.pcap4j.packet.namednumber.DataLinkType;
 
 import java.io.EOFException;
-import java.net.Inet4Address;
 import java.net.UnknownHostException;
+import java.time.Instant;
 import java.util.*;
 import java.util.concurrent.TimeoutException;
 
@@ -26,255 +30,129 @@ public class Main {
 
 
     public static void main(String[] args) throws PcapNativeException, NotOpenException, EOFException, TimeoutException, UnknownHostException {
-        final String fileName = "/users/varmarken/Desktop/wlan1.local.dns.pcap";
-        List<DnsPacket> dnsPackets = extractDnsAnswerPackets(fileName);
-        Map<String, Set<String>> ipToHostnameMap = constructIpToHostnameMap(dnsPackets);
-        ipToHostnameMap.forEach((k,v) -> System.out.println(String.format("%s => %s", k, v.toString())));
-    }
+        // -------------------------------------------------------------------------------------------------------------
+        // ------------ # 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) 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
+
+        // 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 = "192.168.1.246"; // .246 == phone; .199 == dlink plug?
+
+        // 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";
+
+        // 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!)
+
+        // 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";
 
-    /**
-     * Opens a PCAP file and extracts all DNS reply packets with non-empty answer sections.
-     * @param pcapFileName The name of the PCAP file.
-     * @return A list of DNS reply packets.
-     * @throws PcapNativeException
-     * @throws NotOpenException
-     * @throws TimeoutException
-     */
-    private static List<DnsPacket> extractDnsAnswerPackets(String pcapFileName) throws PcapNativeException, NotOpenException, TimeoutException {
-        PcapHandle handle;
-        try {
-            handle = Pcaps.openOffline(pcapFileName, PcapHandle.TimestampPrecision.NANO);
-        } catch (PcapNativeException pne) {
-            handle = Pcaps.openOffline(pcapFileName);
+        // 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";
+
+        // 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";
+
+        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)));
         }
-        // Apparently BPFs don't support "dns" protocol filter, so have to filter by port.
-        handle.setFilter("port 53", BpfProgram.BpfCompileMode.OPTIMIZE);
-        ArrayList<DnsPacket> result = new ArrayList<>();
-        try {
-            Packet packet;
-            while ((packet = handle.getNextPacketEx()) != null) {
-                DnsPacket dnsPacket = packet.get(DnsPacket.class);
-                // We only care about DNS answers.
-                if (dnsPacket != null && dnsPacket.getHeader().getAnswers().size() != 0) {
-                    result.add(dnsPacket);
-                }
+        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();
             }
-        } catch (EOFException eof) {
-            // (Note have to resort to EOFException as handle.getStats().getNumPacketsCaptured() only works on Windows)
-            // Clean up.
-            handle.close();
-        }
-        System.out.println(String.format("Found %d DNS answers", result.size()));
-        return result;
-    }
+        }, dnsMap, tcpReassembler, trafficLabeler);
+        outputter.flush();
+        outputter.close();
 
-    /**
-     * Based on the information found in a list of DNS replies, this method constructs a {@link Map} that maps from an
-     * IP to a {@link Set} of hostnames associated with that IP.
-     *
-     * @param dnsPackets A list of DNS <em>reply</em> packets.
-     * @return A {@link Map} that maps from an IP to a {@link Set} of hostnames associated with that IP
-     * @throws UnknownHostException If an IP found in a {@code DnsPacket} is of incorrect length.
-     */
-    private static Map<String, Set<String>> constructIpToHostnameMap(List<DnsPacket> dnsPackets) throws UnknownHostException {
-        HashMap<String, Set<String>> result = new HashMap<>();
-        for(DnsPacket dnsPacket : dnsPackets) {
-            // The hostname that this DNS reply provides answers for.
-            // TODO: safe to assume only one question?
-            String hostname = dnsPacket.getHeader().getQuestions().get(0).getQName().getName();
-            for(DnsResourceRecord answer : dnsPacket.getHeader().getAnswers()) {
-                // We only care about type A records
-                if (!answer.getDataType().equals(DnsResourceRecordType.A)) {
-                    continue;
-                }
-                // Sanity check. For some reason the hostname appears to be the empty string in the answer .
-                // We hence have to assume that all answers correspond to a single question that holds the hostname as part of its object tree.
-                // Therefore, if there are more questions in one query-reply exchange, we are in trouble.
-                if (!answer.getName().getName().equals("") && !answer.getName().getName().equals(hostname)) {
-                    throw new RuntimeException("[DNS parser] mismatch between hostname in question and hostname in answer");
-                }
-                // The IP in byte representation.
-                byte[] ipBytes = answer.getRData().getRawData();
-                // Convert to string representation.
-                String ip = Inet4Address.getByAddress(ipBytes).getHostAddress();
-                HashSet<String> hostnameSet = new HashSet<>();
-                hostnameSet.add(hostname);
-                // Update or insert depending on presence of key:
-                // Concat the existing set and the new set if ip already present as key,
-                // otherwise add an entry for ip pointing to new set.
-                result.merge(ip, hostnameSet, (v1, v2) -> { v1.addAll(v2); return v1; });
-            }
+        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()));
         }
-        return result;
+
+        // Extract all conversations present in the filtered trace.
+        List<Conversation> allConversations = tcpReassembler.getTcpConversations();
+        // Group conversations by hostname.
+        Map<String, List<Conversation>> convsByHostname = TcpConversationUtils.groupConversationsByHostname(allConversations, dnsMap);
+        System.out.println("Grouped conversations by hostname.");
+        // For each hostname, count the frequencies of packet lengths exchanged with that hostname.
+        final Map<String, Map<Integer, Integer>> pktLenFreqsByHostname = new HashMap<>();
+        convsByHostname.forEach((host, convs) -> pktLenFreqsByHostname.put(host, TcpConversationUtils.countPacketLengthFrequencies(convs)));
+        System.out.println("Counted frequencies of packet lengths exchanged with each hostname.");
+        // For each hostname, count the frequencies of packet sequences (i.e., count how many conversations exchange a
+        // sequence of packets of some specific lengths).
+        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);
+                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("");
+
+        // -------------------------------------------------------------------------------------------------------------
+        // -------------------------------------------------------------------------------------------------------------
     }
 
-//    /**
-//     * Private class properties
-//     */
-//    private Pcap pcap;
-//    private List<Pcap.Packet> listPacket;
-//    private Map<String, String> mapIPAddressToHostname;
-//
-//    /**
-//     * Private class constants
-//     */
-//    private static final int DNS_PORT = 53;
-//
-//    /**
-//     * Constructor
-//     *
-//     * @param   file    name of the analyzed PCAP file
-//     */
-//    public Main(String file) throws IOException {
-//
-//        pcap = Pcap.fromFile(file);
-//        listPacket = pcap.packets();
-//        mapIPAddressToHostname = new HashMap<String, String>();
-//    }
-//
-//
-//
-//
-//
-//    /**
-//     * Private method that maps DNS hostnames to their
-//     * respected IP addresses. This method iterates
-//     * through the List<Pcap.Packet>, gets DNS packets,
-//     * and gets the IP addresses associated with them.
-//     */
-//    private void mapHostnamesToIPAddresses() {
-//
-//        int counter = 1;
-//        for(Pcap.Packet packet : listPacket) {
-//            System.out.print("# " + counter++);
-//            // Check the packet type
-//            if (packet._root().hdr().network() == Pcap.Linktype.ETHERNET) {
-//                EthernetFrame ethFrame = (EthernetFrame) packet.body();
-//                if (ethFrame.etherType() == EthernetFrame.EtherTypeEnum.IPV4) {
-//                    Ipv4Packet ip4Packet = (Ipv4Packet) ethFrame.body();
-//
-//                    System.out.print(" - Protocol: " + ip4Packet.protocol());
-//                    if (ip4Packet.protocol() == Ipv4Packet.ProtocolEnum.UDP) {
-//                        // DNS is UDP port 53
-//                        UdpDatagram udpData = (UdpDatagram) ip4Packet.body();
-//                        System.out.print(" - Source Port: " + udpData.srcPort());
-//                        System.out.print(" - Dest Port: " + udpData.dstPort());
-//
-//                        // Source port 53 means this is DNS response
-//                        if (udpData.srcPort() == DNS_PORT) {
-//                            KaitaiStream dnsStream = new ByteBufferKaitaiStream(udpData.body());
-//                            DnsPacket dnsPacket = new DnsPacket(dnsStream);
-//                            ArrayList<DnsPacket.Query> queries = dnsPacket.queries();
-//                            ArrayList<DnsPacket.Answer> answers = dnsPacket.answers();
-//                            String strDomainName = new String();
-//                            for(DnsPacket.Query query : queries) {
-//                                System.out.print(" - Queries: ");
-//                                DnsPacket.DomainName domainName = query.name();
-//                                ArrayList<DnsPacket.Label> labels = domainName.name();
-//                                for(int i = 0; i < labels.size(); i++) {
-//                                    System.out.print(labels.get(i).name());
-//                                    strDomainName = strDomainName + labels.get(i).name();
-//                                    if(i < labels.size()-2) {
-//                                        System.out.print(".");
-//                                        strDomainName = strDomainName + ".";
-//                                    }
-//                                }
-//                                break;  // We are assuming that there is only one query
-//                            }
-//                            System.out.print(" - Answers " + answers.size());
-//                            for(DnsPacket.Answer answer : answers) {
-//                                System.out.print(" - TypeType: " + answer.type());
-//                                System.out.print(" - ClassType: " + answer.answerClass());
-//                                System.out.print("\n - Answers: ");
-//                                DnsPacket.Address address = answer.address();
-//                                if (answer.type() == DnsPacket.TypeType.A) {
-//                                    String strAnswer = new String();
-//                                    ArrayList<Integer> ipList = address.ip();
-//                                    for(int i = 0; i < ipList.size(); i++) {
-//                                        System.out.print(ipList.get(i));
-//                                        strAnswer = strAnswer + Integer.toString(ipList.get(i));
-//                                        if(i < ipList.size()-1) {
-//                                            System.out.print(".");
-//                                            strAnswer = strAnswer + ".";
-//                                        }
-//                                    }
-//                                    mapIPAddressToHostname.put(strAnswer, strDomainName);
-//                                }
-//                            }
-//                        }
-//                    }
-//                }
-//            }
-//            System.out.println();
-//        }
-////        for(Map.Entry<String, String> entry : mapIPAddressToHostname.entrySet()) {
-////            if (entry.getValue().equals("devs.tplinkcloud.com")) {
-////                System.out.println(entry.getKey() + " - " + entry.getValue());
-////            }
-////        }
-//        System.out.println("Total map size: " + mapIPAddressToHostname.size());
-//        System.out.println("Answer for 13.33.41.8: " + mapIPAddressToHostname.get("13.33.41.8"));
-//        System.out.println("Answer for 34.226.240.125: " + mapIPAddressToHostname.get("34.226.240.125"));
-//    }
-//
-//    /*private String cloudIPAddress(String hostName) {
-//        if (hostName.equals("events.tplinkra.com"))
-//            return "205.251.203.26";
-//        else
-//            return null;
-//    }*/
-//
-//    // TODO move to separate class
-//    // Add parameter that is the trace to be analyzed (most like the pcap library's representation of a flow)
-//    public String findPattern(Map<String, List<Integer>> hostnameToPacketLengths, String smartPlugIp) {
-//
-//        // No difference, output "Complete match"
-//        // If difference, output <Packet no, deviation from expected> for each packet
-//        return null;
-//    }
-//
-//    public static void main(String[] args) {
-//        System.out.println("it works");
-//
-//        //String file = "/home/rtrimana/pcap_processing/smart_home_traffic/Code/Projects/SmartPlugDetector/pcap/wlan1.local.dns.pcap";
-//        String file = "/home/rtrimana/pcap_processing/smart_home_traffic/Code/Projects/SmartPlugDetector/pcap/wlan1.remote.dns.pcap";
-//
-//        try {
-//            Main main = new Main(file);
-//            main.mapHostnamesToIPAddresses();
-//
-//            /*Pcap data = Pcap.fromFile(file);
-//            List<Pcap.Packet> listPacket = data.packets();
-//            System.out.println("Number of packets: " + listPacket.size());
-//            System.out.println("===================");
-//            for(Pcap.Packet packet : listPacket) {
-//                if (packet._root().hdr().network() == Pcap.Linktype.ETHERNET) {
-//                    EthernetFrame eFrame = (EthernetFrame) packet.body();
-//                    if (eFrame.etherType() == EthernetFrame.EtherTypeEnum.IPV4) {
-//                        Ipv4Packet ip4Packet = (Ipv4Packet) eFrame.body();
-//                        byte[] srcIp = ip4Packet.srcIpAddr();
-//                        byte[] dstIp = ip4Packet.dstIpAddr();
-//                        System.out.println("Byte length source: " + srcIp.length + " Byte length dest: " + dstIp.length);
-//                        System.out.print("Source: ");
-//                        for(int i = 0; i < srcIp.length; i++) {
-//                            System.out.print(Byte.toUnsignedInt(srcIp[i]));
-//                            if(i < srcIp.length-1)
-//                                System.out.print(".");
-//                        }
-//                        System.out.print(" - Dest: ");
-//                        for(int i = 0; i < dstIp.length; i++) {
-//                            System.out.print(Byte.toUnsignedInt(dstIp[i]));
-//                            if(i < dstIp.length-1)
-//                                System.out.print(".");
-//                            else
-//                                System.out.println("\n");
-//                        }
-//                    }
-//                }
-//            }*/
-//
-//        } catch (Exception e) {
-//            e.printStackTrace();
-//        }
-//    }
 }
+
+
+// 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