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.comparison.seqalignment.ExtractedSequence;
+import edu.uci.iotproject.comparison.seqalignment.SequenceAlignment;
+import edu.uci.iotproject.comparison.seqalignment.SequenceExtraction;
import edu.uci.iotproject.io.TriggerTimesFileReader;
+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;
import java.io.EOFException;
+import java.io.File;
+import java.io.PrintWriter;
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
// ------------ # 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
+// 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?
+// 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 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 = "192.168.1.140"; // .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 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!)
+ // 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 = "192.168.1.142"; // .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 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
+// 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
+// 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 = "192.168.1.183"; // .246 == phone; .183 == siren
+// // 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 = "192.168.1.183"; // .246 == phone; .183 == siren
// 14) Nest thermostat August 15 experiment
// final String inputPcapFile = path + "/2018-08/nest/nest.wlan1.local.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
+
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.
+ /*
+ * 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>>> outer = ua.getType() == Type.TOGGLE_ON ? ons : offs;
hostnameToConvs.forEach((host, convs) -> {
Map<String, List<Conversation>> seqsToConvs = TcpConversationUtils.
- groupConversationsByPacketSequence(convs);
+ groupConversationsByPacketSequence(convs, verbose);
outer.merge(host, seqsToConvs, (oldMap, newMap) -> {
newMap.forEach((sequence, cs) -> oldMap.merge(sequence, cs, (list1, list2) -> {
list1.addAll(list2);
});
});
});
-
- System.out.println("");
-
- // -------------------------------------------------------------------------------------------------------------
- // -------------------------------------------------------------------------------------------------------------
+ */
+
+ // ================================================ 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());
+ 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, 5);
+ 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, 5);
+ List<Cluster<PcapPacketPair>> offClusters = offClusterer.cluster(offPairs);
+ // Output clusters
+ System.out.println("========================================");
+ System.out.println(" Clustering results for ON ");
+ System.out.println(" Number of clusters: " + onClusters.size());
+ int count = 0;
+ for (Cluster<PcapPacketPair> c : onClusters) {
+ System.out.println(String.format("<<< Cluster #%02d (%03d points) >>>", ++count, c.getPoints().size()));
+ System.out.print(PrintUtils.toSummaryString(c));
+ }
+ System.out.println("========================================");
+ System.out.println(" Clustering results for OFF ");
+ System.out.println(" Number of clusters: " + offClusters.size());
+ count = 0;
+ for (Cluster<PcapPacketPair> c : offClusters) {
+ System.out.println(String.format("<<< Cluster #%03d (%06d points) >>>", ++count, c.getPoints().size()));
+ System.out.print(PrintUtils.toSummaryString(c));
+ }
+ System.out.println("========================================");
+ // ============================================================================================================
+
+ /*
+ 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();
+ */
+
+// // ================================================================================================
+// // <<< 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("");
+// // ================================================================================================
+//
+//
+// // -------------------------------------------------------------------------------------------------------------
+// // -------------------------------------------------------------------------------------------------------------
}
}