1 package edu.uci.iotproject;
3 import static edu.uci.iotproject.analysis.UserAction.Type;
5 import edu.uci.iotproject.analysis.TcpConversationUtils;
6 import edu.uci.iotproject.analysis.TrafficLabeler;
7 import edu.uci.iotproject.analysis.TriggerTrafficExtractor;
8 import edu.uci.iotproject.analysis.UserAction;
9 import edu.uci.iotproject.comparison.seqalignment.ExtractedSequence;
10 import edu.uci.iotproject.comparison.seqalignment.SequenceAlignment;
11 import edu.uci.iotproject.comparison.seqalignment.SequenceExtraction;
12 import edu.uci.iotproject.io.TriggerTimesFileReader;
13 import org.pcap4j.core.*;
14 import org.pcap4j.packet.namednumber.DataLinkType;
16 import java.io.EOFException;
17 import java.net.UnknownHostException;
18 import java.time.Instant;
20 import java.util.concurrent.TimeoutException;
23 * This is a system that reads PCAP files to compare
24 * patterns of DNS hostnames, packet sequences, and packet
25 * lengths with training data to determine certain events
26 * or actions for smart home devices.
28 * @author Janus Varmarken
29 * @author Rahmadi Trimananda (rtrimana@uci.edu)
35 public static void main(String[] args) throws PcapNativeException, NotOpenException, EOFException, TimeoutException, UnknownHostException {
36 // -------------------------------------------------------------------------------------------------------------
37 // ------------ # Code for extracting traffic generated by a device within x seconds of a trigger # ------------
38 // Paths to input and output files (consider supplying these as arguments instead) and IP of the device for
39 // which traffic is to be extracted:
40 String path = "/scratch/July-2018"; // Rahmadi
41 // String path = "/Users/varmarken/temp/UCI IoT Project/experiments"; // Janus
42 boolean verbose = true;
44 // 1) D-Link July 26 experiment
45 // final String inputPcapFile = path + "/2018-07/dlink/dlink.wlan1.local.pcap";
46 // final String outputPcapFile = path + "/2018-07/dlink/dlink-processed.pcap";
47 // final String triggerTimesFile = path + "/2018-07/dlink/dlink-july-26-2018.timestamps";
48 // final String deviceIp = "192.168.1.246"; // .246 == phone; .199 == dlink plug?
50 // 2) TP-Link July 25 experiment
51 // final String inputPcapFile = path + "/2018-07/tplink/tplink.wlan1.local.pcap";
52 // final String outputPcapFile = path + "/2018-07/tplink/tplink-processed.pcap";
53 // final String triggerTimesFile = path + "/2018-07/tplink/tplink-july-25-2018.timestamps";
54 // final String deviceIp = "192.168.1.159";
56 // 2b) TP-Link July 25 experiment TRUNCATED:
57 // Only contains "true local" events, i.e., before the behavior changes to remote-like behavior.
58 // Last included event is at July 25 10:38:11; file filtered to only include packets with arrival time <= 10:38:27.
59 // final String inputPcapFile = path + "/2018-07/tplink/tplink.wlan1.local.truncated.pcap";
60 // final String outputPcapFile = path + "/2018-07/tplink/tplink-processed.truncated.pcap";
61 // final String triggerTimesFile = path + "/2018-07/tplink/tplink-july-25-2018.truncated.timestamps";
62 // final String deviceIp = "192.168.1.159";
64 // 3) SmartThings Plug July 25 experiment
65 // final String inputPcapFile = path + "/2018-07/stplug/stplug.wlan1.local.pcap";
66 // final String outputPcapFile = path + "/2018-07/stplug/stplug-processed.pcap";
67 // final String triggerTimesFile = path + "/2018-07/stplug/smartthings-july-25-2018.timestamps";
68 // final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
70 // 4) Wemo July 30 experiment
71 final String inputPcapFile = path + "/2018-07/wemo/wemo.wlan1.local.pcap";
72 final String outputPcapFile = path + "/2018-07/wemo/wemo-processed.pcap";
73 final String triggerTimesFile = path + "/2018-07/wemo/wemo-july-30-2018.timestamps";
74 final String deviceIp = "192.168.1.145";
76 // 5) Wemo Insight July 31 experiment
77 // final String inputPcapFile = path + "/2018-07/wemoinsight/wemoinsight.wlan1.local.pcap";
78 // final String outputPcapFile = path + "/2018-07/wemoinsight/wemoinsight-processed.pcap";
79 // final String triggerTimesFile = path + "/2018-07/wemoinsight/wemo-insight-july-31-2018.timestamps";
80 // final String deviceIp = "192.168.1.135";
82 // 6) TP-Link Bulb August 1 experiment
83 // final String inputPcapFile = path + "/2018-08/tplink-bulb/tplinkbulb.wlan1.local.pcap";
84 // final String outputPcapFile = path + "/2018-08/tplink-bulb/tplinkbulb-processed.pcap";
85 // final String triggerTimesFile = path + "/2018-08/tplink-bulb/tplink-bulb-aug-3-2018.timestamps";
86 // final String deviceIp = "192.168.1.140";
88 // 7) Kwikset Doorlock August 6 experiment
89 // final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock.wlan1.local.pcap";
90 // final String outputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-processed.pcap";
91 // final String triggerTimesFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-aug-6-2018.timestamps";
92 // final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
94 // 8) Hue Bulb August 7 experiment
95 // final String inputPcapFile = path + "/2018-08/hue-bulb/hue-bulb.wlan1.local.pcap";
96 // final String outputPcapFile = path + "/2018-08/hue-bulb/hue-bulb-processed.pcap";
97 // final String triggerTimesFile = path + "/2018-08/hue-bulb/hue-bulb-aug-7-2018.timestamps";
98 // final String deviceIp = "192.168.1.246";
100 // 9) Lifx Bulb August 8 experiment
101 // final String inputPcapFile = path + "/2018-08/lifx-bulb/lifx-bulb.wlan1.local.pcap";
102 // final String outputPcapFile = path + "/2018-08/lifx-bulb/lifx-bulb-processed.pcap";
103 // final String triggerTimesFile = path + "/2018-08/lifx-bulb/lifx-bulb-aug-8-2018.timestamps";
104 // final String deviceIp = "192.168.1.246"; // .246 == phone; .231 == Lifx
106 // 10) Amcrest Camera August 9 experiment
107 // final String inputPcapFile = path + "/2018-08/amcrest-camera/amcrest-camera.wlan1.local.pcap";
108 // final String outputPcapFile = path + "/2018-08/amcrest-camera/amcrest-camera-processed.pcap";
109 // final String triggerTimesFile = path + "/2018-08/amcrest-camera/amcrest-camera-aug-9-2018.timestamps";
110 // final String deviceIp = "192.168.1.246"; // .246 == phone; .235 == camera
112 // 11) Arlo Camera August 10 experiment
113 // final String inputPcapFile = path + "/2018-08/arlo-camera/arlo-camera.wlan1.local.pcap";
114 // final String outputPcapFile = path + "/2018-08/arlo-camera/arlo-camera-processed.pcap";
115 // final String triggerTimesFile = path + "/2018-08/arlo-camera/arlo-camera-aug-10-2018.timestamps";
116 // final String deviceIp = "192.168.1.140"; // .246 == phone; .140 == camera
118 // 12) Blossom sprinkler August 13 experiment
119 // final String inputPcapFile = path + "/2018-08/blossom/blossom.wlan1.local.pcap";
120 // final String outputPcapFile = path + "/2018-08/blossom/blossom-processed.pcap";
121 // final String triggerTimesFile = path + "/2018-08/blossom/blossom-aug-13-2018.timestamps";
122 // final String deviceIp = "192.168.1.229"; // .246 == phone; .229 == sprinkler
124 // 13) DLink siren August 14 experiment
125 // final String inputPcapFile = path + "/2018-08/dlink-siren/dlink-siren.wlan1.local.pcap";
126 // final String outputPcapFile = path + "/2018-08/dlink-siren/dlink-siren-processed.pcap";
127 // final String triggerTimesFile = path + "/2018-08/dlink-siren/dlink-siren-aug-14-2018.timestamps";
128 // final String deviceIp = "192.168.1.246"; // .246 == phone; .183 == siren
130 // 14) Nest thermostat August 15 experiment
131 // final String inputPcapFile = path + "/2018-08/nest/nest.wlan1.local.pcap";
132 // final String outputPcapFile = path + "/2018-08/nest/nest-processed.pcap";
133 // final String triggerTimesFile = path + "/2018-08/nest/nest-aug-15-2018.timestamps";
134 // final String deviceIp = "192.168.1.246"; // .246 == phone; .127 == Nest thermostat
136 // 15) Alexa August 16 experiment
137 // final String inputPcapFile = path + "/2018-08/alexa/alexa.wlan1.local.pcap";
138 // final String outputPcapFile = path + "/2018-08/alexa/alexa-processed.pcap";
139 // final String triggerTimesFile = path + "/2018-08/alexa/alexa-aug-16-2018.timestamps";
140 // final String deviceIp = "192.168.1.225"; // .246 == phone; .225 == Alexa
142 // final String inputPcapFile = path + "/2018-08/alexa/alexa2.wlan1.local.pcap";
143 // final String outputPcapFile = path + "/2018-08/alexa/alexa2-processed.pcap";
144 // final String triggerTimesFile = path + "/2018-08/alexa/alexa-aug-17-2018.timestamps";
145 // final String deviceIp = "192.168.1.225"; // .246 == phone; .225 == Alexa
147 TriggerTimesFileReader ttfr = new TriggerTimesFileReader();
148 List<Instant> triggerTimes = ttfr.readTriggerTimes(triggerTimesFile, false);
149 // Tag each trigger with "ON" or "OFF", assuming that the first trigger is an "ON" and that they alternate.
150 List<UserAction> userActions = new ArrayList<>();
151 for (int i = 0; i < triggerTimes.size(); i++) {
152 userActions.add(new UserAction(i % 2 == 0 ? Type.TOGGLE_ON : Type.TOGGLE_OFF, triggerTimes.get(i)));
154 TriggerTrafficExtractor tte = new TriggerTrafficExtractor(inputPcapFile, triggerTimes, deviceIp);
155 final PcapDumper outputter = Pcaps.openDead(DataLinkType.EN10MB, 65536).dumpOpen(outputPcapFile);
156 DnsMap dnsMap = new DnsMap();
157 TcpReassembler tcpReassembler = new TcpReassembler();
158 TrafficLabeler trafficLabeler = new TrafficLabeler(userActions);
159 tte.performExtraction(pkt -> {
162 } catch (NotOpenException e) {
165 }, dnsMap, tcpReassembler, trafficLabeler);
169 if (tte.getPacketsIncludedCount() != trafficLabeler.getTotalPacketCount()) {
170 // Sanity/debug check
171 throw new AssertionError(String.format("mismatch between packet count in %s and %s",
172 TriggerTrafficExtractor.class.getSimpleName(), TrafficLabeler.class.getSimpleName()));
175 // Extract all conversations present in the filtered trace.
176 List<Conversation> allConversations = tcpReassembler.getTcpConversations();
177 // Group conversations by hostname.
178 Map<String, List<Conversation>> convsByHostname = TcpConversationUtils.groupConversationsByHostname(allConversations, dnsMap);
179 System.out.println("Grouped conversations by hostname.");
180 // For each hostname, count the frequencies of packet lengths exchanged with that hostname.
181 final Map<String, Map<Integer, Integer>> pktLenFreqsByHostname = new HashMap<>();
182 convsByHostname.forEach((host, convs) -> pktLenFreqsByHostname.put(host, TcpConversationUtils.countPacketLengthFrequencies(convs)));
183 System.out.println("Counted frequencies of packet lengths exchanged with each hostname.");
184 // For each hostname, count the frequencies of packet sequences (i.e., count how many conversations exchange a
185 // sequence of packets of some specific lengths).
186 final Map<String, Map<String, Integer>> pktSeqFreqsByHostname = new HashMap<>();
187 convsByHostname.forEach((host, convs) -> pktSeqFreqsByHostname.put(host, TcpConversationUtils.countPacketSequenceFrequencies(convs)));
188 System.out.println("Counted frequencies of packet sequences exchanged with each hostname.");
189 // For each hostname, count frequencies of packet pairs exchanged with that hostname across all conversations
190 final Map<String, Map<String, Integer>> pktPairFreqsByHostname =
191 TcpConversationUtils.countPacketPairFrequenciesByHostname(allConversations, dnsMap);
192 System.out.println("Counted frequencies of packet pairs per hostname");
193 // For each user action, reassemble the set of TCP connections occurring shortly after
194 final Map<UserAction, List<Conversation>> userActionToConversations = trafficLabeler.getLabeledReassembledTcpTraffic();
195 final Map<UserAction, Map<String, List<Conversation>>> userActionsToConvsByHostname = trafficLabeler.getLabeledReassembledTcpTraffic(dnsMap);
196 System.out.println("Reassembled TCP conversations occurring shortly after each user event");
200 // Contains all ON events: hostname -> sequence identifier -> list of conversations with that sequence
201 Map<String, Map<String, List<Conversation>>> ons = new HashMap<>();
202 // Contains all OFF events: hostname -> sequence identifier -> list of conversations with that sequence
203 Map<String, Map<String, List<Conversation>>> offs = new HashMap<>();
206 userActionsToConvsByHostname.forEach((ua, hostnameToConvs) -> {
207 Map<String, Map<String, List<Conversation>>> outer = ua.getType() == Type.TOGGLE_ON ? ons : offs;
208 hostnameToConvs.forEach((host, convs) -> {
209 Map<String, List<Conversation>> seqsToConvs = TcpConversationUtils.
210 groupConversationsByPacketSequenceVerbose(convs);
211 outer.merge(host, seqsToConvs, (oldMap, newMap) -> {
212 newMap.forEach((sequence, cs) -> oldMap.merge(sequence, cs, (list1, list2) -> {
221 userActionsToConvsByHostname.forEach((ua, hostnameToConvs) -> {
222 Map<String, Map<String, List<Conversation>>> outer = ua.getType() == Type.TOGGLE_ON ? ons : offs;
223 hostnameToConvs.forEach((host, convs) -> {
224 Map<String, List<Conversation>> seqsToConvs = TcpConversationUtils.
225 groupConversationsByPacketSequence(convs);
226 outer.merge(host, seqsToConvs, (oldMap, newMap) -> {
227 newMap.forEach((sequence, cs) -> oldMap.merge(sequence, cs, (list1, list2) -> {
237 // ================================================================================================
238 // <<< Some work-in-progress/explorative code that extracts a "representative" sequence >>>
240 // Currently need to know relevant hostname in advance :(
241 String hostname = "events.tplinkra.com";
242 // Conversations with 'hostname' for ON events.
243 List<Conversation> onsForHostname = new ArrayList<>();
244 // Conversations with 'hostname' for OFF events.
245 List<Conversation> offsForHostname = new ArrayList<>();
246 // "Unwrap" sequence groupings in ons/offs maps.
247 ons.get(hostname).forEach((k,v) -> onsForHostname.addAll(v));
248 offs.get(hostname).forEach((k,v) -> offsForHostname.addAll(v));
249 // Extract representative sequence for ON and OFF by providing the list of conversations with
250 // 'hostname' observed for each event type (the training data).
251 SequenceExtraction seqExtraction = new SequenceExtraction();
252 ExtractedSequence extractedSequenceForOn = seqExtraction.extract(onsForHostname);
253 ExtractedSequence extractedSequenceForOff = seqExtraction.extract(offsForHostname);
254 // Let's check how many ONs align with OFFs and vice versa (that is, how many times an event is incorrectly
256 int onsLabeledAsOff = 0;
257 Integer[] representativeOnSeq = TcpConversationUtils.getPacketLengthSequence(extractedSequenceForOn.getRepresentativeSequence());
258 Integer[] representativeOffSeq = TcpConversationUtils.getPacketLengthSequence(extractedSequenceForOff.getRepresentativeSequence());
259 SequenceAlignment<Integer> seqAlg = seqExtraction.getAlignmentAlgorithm();
260 for (Conversation c : onsForHostname) {
261 Integer[] onSeq = TcpConversationUtils.getPacketLengthSequence(c);
262 if (seqAlg.calculateAlignment(representativeOffSeq, onSeq) <= extractedSequenceForOff.getMaxAlignmentCost()) {
266 int offsLabeledAsOn = 0;
267 for (Conversation c : offsForHostname) {
268 Integer[] offSeq = TcpConversationUtils.getPacketLengthSequence(c);
269 if (seqAlg.calculateAlignment(representativeOnSeq, offSeq) <= extractedSequenceForOn.getMaxAlignmentCost()) {
273 System.out.println("");
274 // ================================================================================================
277 // -------------------------------------------------------------------------------------------------------------
278 // -------------------------------------------------------------------------------------------------------------
284 // TP-Link MAC 50:c7:bf:33:1f:09 and usually IP 192.168.1.159 (remember to verify per file)
285 // frame.len >= 556 && frame.len <= 558 && ip.addr == 192.168.1.159