Adding Python ML examples of clustering data points.
[pingpong.git] / Code / Projects / SmartPlugDetector / src / main / java / edu / uci / iotproject / Main.java
index f3b8a005bdb37009c0d81b4d50565b4750a567f2..72b14e0b0b075e8c322ea9f9b3fb83311a0982e8 100644 (file)
@@ -1,11 +1,18 @@
 package edu.uci.iotproject;
 
-import edu.uci.iotproject.maclayer.MacLayerFlowPattern;
-import edu.uci.iotproject.maclayer.MacLayerFlowPatternFinder;
+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.namednumber.DataLinkType;
 
 import java.io.EOFException;
 import java.net.UnknownHostException;
+import java.time.Instant;
 import java.util.*;
 import java.util.concurrent.TimeoutException;
 
@@ -24,51 +31,195 @@ public class Main {
 
     public static void main(String[] args) throws PcapNativeException, NotOpenException, EOFException, TimeoutException, UnknownHostException {
         // -------------------------------------------------------------------------------------------------------------
-        // Example/debug code for searching for a pattern at the MAC layer.
-//        String fileName = "./pcap/mac-tplink.local.pcapng";
-//        PcapHandle handle;
-//        try {
-//            handle = Pcaps.openOffline(fileName, PcapHandle.TimestampPrecision.NANO);
-//        } catch (PcapNativeException pne) {
-//            handle = Pcaps.openOffline(fileName);
-//        }
-//        Arrays.asList(1590, 1590, 1590, 1001, 337, 197, 636, 1311, 177) // Full pattern (all non-zero payload packets).
-//        MacLayerFlowPattern pattern = new MacLayerFlowPattern("TP_LINK_LOCAL_OFF_MAC", "50:c7:bf:33:1f:09", Arrays.asList(637, 1312));
-//        MacLayerFlowPatternFinder finder = new MacLayerFlowPatternFinder(handle, pattern);
-//        finder.findFlowPattern();
-        // -------------------------------------------------------------------------------------------------------------
+        // ------------ # 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
+
+        // 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 = "192.168.1.246"; // .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";
+
+        // 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 = "192.168.1.229"; // .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
 
-        //final String fileName = args.length > 0 ? args[0] : "/home/rtrimana/pcap_processing/smart_home_traffic/Code/Projects/SmartPlugDetector/pcap/wlan1.local.dns.pcap";
-        final String fileName = args.length > 0 ? args[0] : "/scratch/June-2018/TPLink/wlan1/tplink.wlan1.local.pcap";
-        //final String fileName = args.length > 0 ? args[0] : "/scratch/June-2018/DLink/wlan1/dlink.wlan1.local.pcap";
-        final String trainingFileName = "./pcap/TP_LINK_LOCAL_ON_SUBSET.pcap";
-//        final String trainingFileName = "./pcap/TP_LINK_LOCAL_ON.pcap";
-//
-//        // ====== Debug code ======
-        PcapHandle handle;
-        PcapHandle trainingPcap;
-        try {
-            handle = Pcaps.openOffline(fileName, PcapHandle.TimestampPrecision.NANO);
-            trainingPcap = Pcaps.openOffline(trainingFileName, PcapHandle.TimestampPrecision.NANO);
-        } catch (PcapNativeException pne) {
-            handle = Pcaps.openOffline(fileName);
-            trainingPcap = Pcaps.openOffline(trainingFileName);
+        // 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, 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()));
         }
-//
-//        // TODO: The followings are the way to extract multiple hostnames and their associated packet lengths lists
-//        //List<String> list = new ArrayList<>();
-//        //list.add("events.tplinkra.com");
-//        //FlowPattern fp = new FlowPattern("TP_LINK_LOCAL_ON", list, trainingPcap);
-//        //List<String> list2 = new ArrayList<>();
-//        //list2.add("devs.tplinkcloud.com");
-//        //list2.add("events.tplinkra.com");
-//        //FlowPattern fp3 = new FlowPattern("TP_LINK_REMOTE_ON", list2, trainingPcap);
-//
-        FlowPattern fp = new FlowPattern("TP_LINK_LOCAL_ON", "events.tplinkra.com", trainingPcap);
-        //FlowPattern fp = new FlowPattern("DLINK_LOCAL_ON", "rfe-us-west-1.dch.dlink.com", trainingPcap);
-        FlowPatternFinder fpf = new FlowPatternFinder(handle, fp);
-        fpf.start();
-//
-//        // ========================
+
+        // 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("");
+
+        // -------------------------------------------------------------------------------------------------------------
+        // -------------------------------------------------------------------------------------------------------------
     }
+
 }
+
+
+// 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