ed8c1bd27a4d61f54e0ea9200a74403311394860
[pingpong.git] / Code / Projects / SmartPlugDetector / src / main / java / edu / uci / iotproject / Main.java
1 package edu.uci.iotproject;
2
3 import static edu.uci.iotproject.analysis.UserAction.Type;
4
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;
15
16 import java.io.EOFException;
17 import java.net.UnknownHostException;
18 import java.time.Instant;
19 import java.util.*;
20 import java.util.concurrent.TimeoutException;
21
22 /**
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.
27  *
28  * @author Janus Varmarken
29  * @author Rahmadi Trimananda (rtrimana@uci.edu)
30  * @version 0.1
31  */
32 public class Main {
33
34
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
43         // 1) D-Link July 26 experiment
44         final String inputPcapFile = path + "/2018-07/dlink/dlink.wlan1.local.pcap";
45         final String outputPcapFile = path + "/2018-07/dlink/dlink-processed.pcap";
46         final String triggerTimesFile = path + "/2018-07/dlink/dlink-july-26-2018.timestamps";
47         final String deviceIp = "192.168.1.246"; // .246 == phone; .199 == dlink plug?
48
49         // 2) TP-Link July 25 experiment
50 //        final String inputPcapFile = path + "/2018-07/tplink/tplink.wlan1.local.pcap";
51 //        final String outputPcapFile = path + "/2018-07/tplink/tplink-processed.pcap";
52 //        final String triggerTimesFile = path + "/2018-07/tplink/tplink-july-25-2018.timestamps";
53 //        final String deviceIp = "192.168.1.159";
54
55         // 2b) TP-Link July 25 experiment TRUNCATED:
56         // Only contains "true local" events, i.e., before the behavior changes to remote-like behavior.
57         // Last included event is at July 25 10:38:11; file filtered to only include packets with arrival time <= 10:38:27.
58         final String inputPcapFile = path + "/2018-07/tplink/tplink.wlan1.local.truncated.pcap";
59         final String outputPcapFile = path + "/2018-07/tplink/tplink-processed.truncated.pcap";
60         final String triggerTimesFile = path + "/2018-07/tplink/tplink-july-25-2018.truncated.timestamps";
61         final String deviceIp = "192.168.1.159";
62
63         // 3) SmartThings Plug July 25 experiment
64 //        final String inputPcapFile = path + "/2018-07/stplug/stplug.wlan1.local.pcap";
65 //        final String outputPcapFile = path + "/2018-07/stplug/stplug-processed.pcap";
66 //        final String triggerTimesFile = path + "/2018-07/stplug/smartthings-july-25-2018.timestamps";
67 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
68
69         // 4) Wemo July 30 experiment
70 //        final String inputPcapFile = path + "/2018-07/wemo/wemo.wlan1.local.pcap";
71 //        final String outputPcapFile = path + "/2018-07/wemo/wemo-processed.pcap";
72 //        final String triggerTimesFile = path + "/2018-07/wemo/wemo-july-30-2018.timestamps";
73 //        final String deviceIp = "192.168.1.145";
74
75         // 5) Wemo Insight July 31 experiment
76 //        final String inputPcapFile = path + "/2018-07/wemoinsight/wemoinsight.wlan1.local.pcap";
77 //        final String outputPcapFile = path + "/2018-07/wemoinsight/wemoinsight-processed.pcap";
78 //        final String triggerTimesFile = path + "/2018-07/wemoinsight/wemo-insight-july-31-2018.timestamps";
79 //        final String deviceIp = "192.168.1.135";
80
81         // 6) TP-Link Bulb August 1 experiment
82 //        final String inputPcapFile = path + "/2018-08/tplink-bulb/tplinkbulb.wlan1.local.pcap";
83 //        final String outputPcapFile = path + "/2018-08/tplink-bulb/tplinkbulb-processed.pcap";
84 //        final String triggerTimesFile = path + "/2018-08/tplink-bulb/tplink-bulb-aug-3-2018.timestamps";
85 //        final String deviceIp = "192.168.1.140";
86
87         // 7) Kwikset Doorlock August 6 experiment
88 //        final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock.wlan1.local.pcap";
89 //        final String outputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-processed.pcap";
90 //        final String triggerTimesFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-aug-6-2018.timestamps";
91 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
92
93         // 8) Hue Bulb August 7 experiment
94 //        final String inputPcapFile = path + "/2018-08/hue-bulb/hue-bulb.wlan1.local.pcap";
95 //        final String outputPcapFile = path + "/2018-08/hue-bulb/hue-bulb-processed.pcap";
96 //        final String triggerTimesFile = path + "/2018-08/hue-bulb/hue-bulb-aug-7-2018.timestamps";
97 //        final String deviceIp = "192.168.1.246";
98
99         // 9) Lifx Bulb August 8 experiment
100 //        final String inputPcapFile = path + "/2018-08/lifx-bulb/lifx-bulb.wlan1.local.pcap";
101 //        final String outputPcapFile = path + "/2018-08/lifx-bulb/lifx-bulb-processed.pcap";
102 //        final String triggerTimesFile = path + "/2018-08/lifx-bulb/lifx-bulb-aug-8-2018.timestamps";
103 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .231 == Lifx
104
105         // 10) Amcrest Camera August 9 experiment
106 //        final String inputPcapFile = path + "/2018-08/amcrest-camera/amcrest-camera.wlan1.local.pcap";
107 //        final String outputPcapFile = path + "/2018-08/amcrest-camera/amcrest-camera-processed.pcap";
108 //        final String triggerTimesFile = path + "/2018-08/amcrest-camera/amcrest-camera-aug-9-2018.timestamps";
109 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .235 == camera
110
111         // 11) Arlo Camera August 10 experiment
112 //        final String inputPcapFile = path + "/2018-08/arlo-camera/arlo-camera.wlan1.local.pcap";
113 //        final String outputPcapFile = path + "/2018-08/arlo-camera/arlo-camera-processed.pcap";
114 //        final String triggerTimesFile = path + "/2018-08/arlo-camera/arlo-camera-aug-10-2018.timestamps";
115 //        final String deviceIp = "192.168.1.140"; // .246 == phone; .140 == camera
116
117         // 12) Blossom sprinkler August 13 experiment
118 //        final String inputPcapFile = path + "/2018-08/blossom/blossom.wlan1.local.pcap";
119 //        final String outputPcapFile = path + "/2018-08/blossom/blossom-processed.pcap";
120 //        final String triggerTimesFile = path + "/2018-08/blossom/blossom-aug-13-2018.timestamps";
121 //        final String deviceIp = "192.168.1.229"; // .246 == phone; .229 == sprinkler
122
123         // 13) DLink siren August 14 experiment
124 //        final String inputPcapFile = path + "/2018-08/dlink-siren/dlink-siren.wlan1.local.pcap";
125 //        final String outputPcapFile = path + "/2018-08/dlink-siren/dlink-siren-processed.pcap";
126 //        final String triggerTimesFile = path + "/2018-08/dlink-siren/dlink-siren-aug-14-2018.timestamps";
127 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .183 == siren
128
129         // 14) Nest thermostat August 15 experiment
130 //        final String inputPcapFile = path + "/2018-08/nest/nest.wlan1.local.pcap";
131 //        final String outputPcapFile = path + "/2018-08/nest/nest-processed.pcap";
132 //        final String triggerTimesFile = path + "/2018-08/nest/nest-aug-15-2018.timestamps";
133 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .127 == Nest thermostat
134
135         // 15) Alexa August 16 experiment
136 //        final String inputPcapFile = path + "/2018-08/alexa/alexa.wlan1.local.pcap";
137 //        final String outputPcapFile = path + "/2018-08/alexa/alexa-processed.pcap";
138 //        final String triggerTimesFile = path + "/2018-08/alexa/alexa-aug-16-2018.timestamps";
139 //        final String deviceIp = "192.168.1.225"; // .246 == phone; .225 == Alexa
140         // August 17
141 //        final String inputPcapFile = path + "/2018-08/alexa/alexa2.wlan1.local.pcap";
142 //        final String outputPcapFile = path + "/2018-08/alexa/alexa2-processed.pcap";
143 //        final String triggerTimesFile = path + "/2018-08/alexa/alexa-aug-17-2018.timestamps";
144 //        final String deviceIp = "192.168.1.225"; // .246 == phone; .225 == Alexa
145
146         TriggerTimesFileReader ttfr = new TriggerTimesFileReader();
147         List<Instant> triggerTimes = ttfr.readTriggerTimes(triggerTimesFile, false);
148         // Tag each trigger with "ON" or "OFF", assuming that the first trigger is an "ON" and that they alternate.
149         List<UserAction> userActions = new ArrayList<>();
150         for (int i = 0; i < triggerTimes.size(); i++) {
151             userActions.add(new UserAction(i % 2 == 0 ? Type.TOGGLE_ON : Type.TOGGLE_OFF, triggerTimes.get(i)));
152         }
153         TriggerTrafficExtractor tte = new TriggerTrafficExtractor(inputPcapFile, triggerTimes, deviceIp);
154         final PcapDumper outputter = Pcaps.openDead(DataLinkType.EN10MB, 65536).dumpOpen(outputPcapFile);
155         DnsMap dnsMap = new DnsMap();
156         TcpReassembler tcpReassembler = new TcpReassembler();
157         TrafficLabeler trafficLabeler = new TrafficLabeler(userActions);
158         tte.performExtraction(pkt -> {
159             try {
160                 outputter.dump(pkt);
161             } catch (NotOpenException e) {
162                 e.printStackTrace();
163             }
164         }, dnsMap, tcpReassembler, trafficLabeler);
165         outputter.flush();
166         outputter.close();
167
168         if (tte.getPacketsIncludedCount() != trafficLabeler.getTotalPacketCount()) {
169             // Sanity/debug check
170             throw new AssertionError(String.format("mismatch between packet count in %s and %s",
171                     TriggerTrafficExtractor.class.getSimpleName(), TrafficLabeler.class.getSimpleName()));
172         }
173
174         // Extract all conversations present in the filtered trace.
175         List<Conversation> allConversations = tcpReassembler.getTcpConversations();
176         // Group conversations by hostname.
177         Map<String, List<Conversation>> convsByHostname = TcpConversationUtils.groupConversationsByHostname(allConversations, dnsMap);
178         System.out.println("Grouped conversations by hostname.");
179         // For each hostname, count the frequencies of packet lengths exchanged with that hostname.
180         final Map<String, Map<Integer, Integer>> pktLenFreqsByHostname = new HashMap<>();
181         convsByHostname.forEach((host, convs) -> pktLenFreqsByHostname.put(host, TcpConversationUtils.countPacketLengthFrequencies(convs)));
182         System.out.println("Counted frequencies of packet lengths exchanged with each hostname.");
183         // For each hostname, count the frequencies of packet sequences (i.e., count how many conversations exchange a
184         // sequence of packets of some specific lengths).
185         final Map<String, Map<String, Integer>> pktSeqFreqsByHostname = new HashMap<>();
186         convsByHostname.forEach((host, convs) -> pktSeqFreqsByHostname.put(host, TcpConversationUtils.countPacketSequenceFrequencies(convs)));
187         System.out.println("Counted frequencies of packet sequences exchanged with each hostname.");
188         // For each hostname, count frequencies of packet pairs exchanged with that hostname across all conversations
189         final Map<String, Map<String, Integer>> pktPairFreqsByHostname =
190                 TcpConversationUtils.countPacketPairFrequenciesByHostname(allConversations, dnsMap);
191         System.out.println("Counted frequencies of packet pairs per hostname");
192         // For each user action, reassemble the set of TCP connections occurring shortly after
193         final Map<UserAction, List<Conversation>> userActionToConversations = trafficLabeler.getLabeledReassembledTcpTraffic();
194         final Map<UserAction, Map<String, List<Conversation>>> userActionsToConvsByHostname = trafficLabeler.getLabeledReassembledTcpTraffic(dnsMap);
195         System.out.println("Reassembled TCP conversations occurring shortly after each user event");
196
197
198
199         // Contains all ON events: hostname -> sequence identifier -> list of conversations with that sequence
200         Map<String, Map<String, List<Conversation>>> ons = new HashMap<>();
201         // Contains all OFF events: hostname -> sequence identifier -> list of conversations with that sequence
202         Map<String, Map<String, List<Conversation>>> offs = new HashMap<>();
203         userActionsToConvsByHostname.forEach((ua, hostnameToConvs) -> {
204             Map<String, Map<String, List<Conversation>>> outer = ua.getType() == Type.TOGGLE_ON ? ons : offs;
205             hostnameToConvs.forEach((host, convs) -> {
206                 Map<String, List<Conversation>> seqsToConvs = TcpConversationUtils.
207                         groupConversationsByPacketSequence(convs);
208                 outer.merge(host, seqsToConvs, (oldMap, newMap) -> {
209                     newMap.forEach((sequence, cs) -> oldMap.merge(sequence, cs, (list1, list2) -> {
210                         list1.addAll(list2);
211                         return list1;
212                     }));
213                     return oldMap;
214                 });
215             });
216         });
217
218         // ================================================================================================
219         // <<< Some work-in-progress/explorative code that extracts a "representative" sequence >>>
220         //
221         // Currently need to know relevant hostname in advance :(
222         String hostname = "events.tplinkra.com";
223         // Conversations with 'hostname' for ON events.
224         List<Conversation> onsForHostname = new ArrayList<>();
225         // Conversations with 'hostname' for OFF events.
226         List<Conversation> offsForHostname = new ArrayList<>();
227         // "Unwrap" sequence groupings in ons/offs maps.
228         ons.get(hostname).forEach((k,v) -> onsForHostname.addAll(v));
229         offs.get(hostname).forEach((k,v) -> offsForHostname.addAll(v));
230         // Extract representative sequence for ON and OFF by providing the list of conversations with
231         // 'hostname' observed for each event type (the training data).
232         SequenceExtraction seqExtraction = new SequenceExtraction();
233         ExtractedSequence extractedSequenceForOn = seqExtraction.extract(onsForHostname);
234         ExtractedSequence extractedSequenceForOff = seqExtraction.extract(offsForHostname);
235         // Let's check how many ONs align with OFFs and vice versa (that is, how many times an event is incorrectly
236         // labeled).
237         int onsLabeledAsOff = 0;
238         Integer[] representativeOnSeq = TcpConversationUtils.getPacketLengthSequence(extractedSequenceForOn.getRepresentativeSequence());
239         Integer[] representativeOffSeq = TcpConversationUtils.getPacketLengthSequence(extractedSequenceForOff.getRepresentativeSequence());
240         SequenceAlignment<Integer> seqAlg = seqExtraction.getAlignmentAlgorithm();
241         for (Conversation c : onsForHostname) {
242             Integer[] onSeq = TcpConversationUtils.getPacketLengthSequence(c);
243             if (seqAlg.calculateAlignment(representativeOffSeq, onSeq) <= extractedSequenceForOff.getMaxAlignmentCost()) {
244                 onsLabeledAsOff++;
245             }
246         }
247         int offsLabeledAsOn = 0;
248         for (Conversation c : offsForHostname) {
249             Integer[] offSeq = TcpConversationUtils.getPacketLengthSequence(c);
250             if (seqAlg.calculateAlignment(representativeOnSeq, offSeq) <= extractedSequenceForOn.getMaxAlignmentCost()) {
251                 offsLabeledAsOn++;
252             }
253         }
254         System.out.println("");
255         // ================================================================================================
256
257
258         // -------------------------------------------------------------------------------------------------------------
259         // -------------------------------------------------------------------------------------------------------------
260     }
261
262 }
263
264
265 // TP-Link MAC 50:c7:bf:33:1f:09 and usually IP 192.168.1.159 (remember to verify per file)
266 // frame.len >= 556 && frame.len <= 558 && ip.addr == 192.168.1.159