TrafficLabeler.java: provide functionality for extracting labeled traffic grouped...
[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.io.TriggerTimesFileReader;
10 import org.pcap4j.core.*;
11 import org.pcap4j.packet.namednumber.DataLinkType;
12
13 import java.io.EOFException;
14 import java.net.UnknownHostException;
15 import java.time.Instant;
16 import java.util.*;
17 import java.util.concurrent.TimeoutException;
18
19 /**
20  * This is a system that reads PCAP files to compare
21  * patterns of DNS hostnames, packet sequences, and packet
22  * lengths with training data to determine certain events
23  * or actions for smart home devices.
24  *
25  * @author Janus Varmarken
26  * @author Rahmadi Trimananda (rtrimana@uci.edu)
27  * @version 0.1
28  */
29 public class Main {
30
31
32     public static void main(String[] args) throws PcapNativeException, NotOpenException, EOFException, TimeoutException, UnknownHostException {
33         // -------------------------------------------------------------------------------------------------------------
34         // ------------ # Code for extracting traffic generated by a device within x seconds of a trigger # ------------
35         // Paths to input and output files (consider supplying these as arguments instead) and IP of the device for
36         // which traffic is to be extracted:
37
38         // D-Link July 26 experiment
39 //        final String inputPcapFile = "/Users/varmarken/temp/UCI IoT Project/experiments/2018-07/dlink/dlink.wlan1.local.pcap";
40 //        final String outputPcapFile = "/Users/varmarken/temp/UCI IoT Project/experiments/2018-07/dlink/dlink-processed.pcap";
41 //        final String triggerTimesFile = "/Users/varmarken/temp/UCI IoT Project/experiments/2018-07/dlink/dlink-july-26-2018.timestamps";
42 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .199 == dlink plug?
43
44         // TP-Link July 25 experiment
45 //        final String inputPcapFile = "/Users/varmarken/temp/UCI IoT Project/experiments/2018-07/tplink/tplink.wlan1.local.pcap";
46 //        final String outputPcapFile = "/Users/varmarken/temp/UCI IoT Project/experiments/2018-07/tplink/tplink-processed.pcap";
47 //        final String triggerTimesFile = "/Users/varmarken/temp/UCI IoT Project/experiments/2018-07/tplink/tplink-july-25-2018.timestamps";
48 //        final String deviceIp = "192.168.1.159";
49
50         // SmartThings Plug July 25 experiment
51 //        final String inputPcapFile = "/Users/varmarken/temp/UCI IoT Project/experiments/2018-07/stplug/stplug.wlan1.local.pcap";
52 //        final String outputPcapFile = "/Users/varmarken/temp/UCI IoT Project/experiments/2018-07/stplug/stplug-processed.pcap";
53 //        final String triggerTimesFile = "/Users/varmarken/temp/UCI IoT Project/experiments/2018-07/stplug/smartthings-july-25-2018.timestamps";
54 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
55
56         // Wemo July 30 experiment
57 //        final String inputPcapFile = "/Users/varmarken/temp/UCI IoT Project/experiments/2018-07/wemo/wemo.wlan1.local.pcap";
58 //        final String outputPcapFile = "/Users/varmarken/temp/UCI IoT Project/experiments/2018-07/wemo/wemo-processed.pcap";
59 //        final String triggerTimesFile = "/Users/varmarken/temp/UCI IoT Project/experiments/2018-07/wemo/wemo-july-30-2018.timestamps";
60 //        final String deviceIp = "192.168.1.145";
61
62         // Wemo Insight July 31 experiment
63 //        final String inputPcapFile = "/Users/varmarken/temp/UCI IoT Project/experiments/2018-07/wemoinsight/wemoinsight.wlan1.local.pcap";
64 //        final String outputPcapFile = "/Users/varmarken/temp/UCI IoT Project/experiments/2018-07/wemoinsight/wemoinsight-processed.pcap";
65 //        final String triggerTimesFile = "/Users/varmarken/temp/UCI IoT Project/experiments/2018-07/wemoinsight/wemo-insight-july-31-2018.timestamps";
66 //        final String deviceIp = "192.168.1.135";
67
68         // TP-Link BULB August 1 experiment
69         final String inputPcapFile = "/Users/varmarken/temp/UCI IoT Project/experiments/2018-08/tplink-bulb/tplinkbulb.wlan1.local.pcap";
70         final String outputPcapFile = "/Users/varmarken/temp/UCI IoT Project/experiments/2018-08/tplink-bulb/tplinkbulb-processed.pcap";
71         final String triggerTimesFile = "/Users/varmarken/temp/UCI IoT Project/experiments/2018-08/tplink-bulb/tplink-bulb-aug-3-2018.timestamps";
72         final String deviceIp = "192.168.1.140";
73
74         TriggerTimesFileReader ttfr = new TriggerTimesFileReader();
75         List<Instant> triggerTimes = ttfr.readTriggerTimes(triggerTimesFile, false);
76         // Tag each trigger with "ON" or "OFF", assuming that the first trigger is an "ON" and that they alternate.
77         List<UserAction> userActions = new ArrayList<>();
78         for (int i = 0; i < triggerTimes.size(); i++) {
79             userActions.add(new UserAction(i % 2 == 0 ? Type.TOGGLE_ON : Type.TOGGLE_OFF, triggerTimes.get(i)));
80         }
81         TriggerTrafficExtractor tte = new TriggerTrafficExtractor(inputPcapFile, triggerTimes, deviceIp);
82         final PcapDumper outputter = Pcaps.openDead(DataLinkType.EN10MB, 65536).dumpOpen(outputPcapFile);
83         DnsMap dnsMap = new DnsMap();
84         TcpReassembler tcpReassembler = new TcpReassembler();
85         TrafficLabeler trafficLabeler = new TrafficLabeler(userActions);
86         tte.performExtraction(pkt -> {
87             try {
88                 outputter.dump(pkt);
89             } catch (NotOpenException e) {
90                 e.printStackTrace();
91             }
92         }, dnsMap, tcpReassembler, trafficLabeler);
93         outputter.flush();
94         outputter.close();
95
96         if (tte.getPacketsIncludedCount() != trafficLabeler.getTotalPacketCount()) {
97             // Sanity/debug check
98             throw new AssertionError(String.format("mismatch between packet count in %s and %s",
99                     TriggerTrafficExtractor.class.getSimpleName(), TrafficLabeler.class.getSimpleName()));
100         }
101
102         // Extract all conversations present in the filtered trace.
103         List<Conversation> allConversations = tcpReassembler.getTcpConversations();
104         // Group conversations by hostname.
105         Map<String, List<Conversation>> convsByHostname = TcpConversationUtils.groupConversationsByHostname(allConversations, dnsMap);
106         System.out.println("Grouped conversations by hostname.");
107         // For each hostname, count the frequencies of packet lengths exchanged with that hostname.
108         final Map<String, Map<Integer, Integer>> pktLenFreqsByHostname = new HashMap<>();
109         convsByHostname.forEach((host, convs) -> pktLenFreqsByHostname.put(host, TcpConversationUtils.countPacketLengthFrequencies(convs)));
110         System.out.println("Counted frequencies of packet lengths exchanged with each hostname.");
111         // For each hostname, count the frequencies of packet sequences (i.e., count how many conversations exchange a
112         // sequence of packets of some specific lengths).
113         final Map<String, Map<String, Integer>> pktSeqFreqsByHostname = new HashMap<>();
114         convsByHostname.forEach((host, convs) -> pktSeqFreqsByHostname.put(host, TcpConversationUtils.countPacketSequenceFrequencies(convs)));
115         System.out.println("Counted frequencies of packet sequences exchanged with each hostname.");
116         // For each hostname, count frequencies of packet pairs exchanged with that hostname across all conversations
117         final Map<String, Map<String, Integer>> pktPairFreqsByHostname =
118                 TcpConversationUtils.countPacketPairFrequenciesByHostname(allConversations, dnsMap);
119         System.out.println("Counted frequencies of packet pairs per hostname");
120         // For each user action, reassemble the set of TCP connections occurring shortly after
121         final Map<UserAction, List<Conversation>> userActionToConversations = trafficLabeler.getLabeledReassembledTcpTraffic();
122         final Map<UserAction, Map<String, List<Conversation>>> userActionsToConvsByHostname = trafficLabeler.getLabeledReassembledTcpTraffic(dnsMap);
123         System.out.println("Reassembled TCP conversations occurring shortly after each user event");
124
125
126
127
128         // ons
129         Map<String, Map<String, Integer>> ons = new HashMap<>();
130         Map<String, Map<String, Integer>> offs = new HashMap<>();
131
132         userActionsToConvsByHostname.forEach((ua, hostnameToConvs) -> {
133             Map<String, Map<String, Integer>> outer = ua.getType() == Type.TOGGLE_ON ? ons : offs;
134             hostnameToConvs.forEach((host, convs) -> {
135                 Map<String, Integer> sequenceCounts = TcpConversationUtils.countPacketSequenceFrequencies(convs);
136                 outer.merge(host, sequenceCounts, (existingMap, newMap) -> {
137                     newMap.forEach((sequence, count) -> existingMap.merge(sequence, count, (i1, i2) -> i1+i2));
138                     return existingMap;
139                 });
140             });
141         });
142
143
144 //                    for (Map.Entry<String, Integer> newMapEntry : newMap.entrySet()) {
145 //                        if (existingMap.get(newMapEntry.getKey()) != null) {
146 //                            existingMap.put(newMapEntry.getKey(), existingMap.get(newMapEntry.getKey()) + newMapEntry.getValue());
147 //                        } else {
148 //                            existingMap.put(newMapEntry.getKey(), newMapEntry.getValue());
149 //                        }
150 //                    }
151 //                    return existingMap;
152
153         System.out.println("");
154
155         // -------------------------------------------------------------------------------------------------------------
156         // -------------------------------------------------------------------------------------------------------------
157     }
158
159 }
160
161
162 // TP-Link MAC 50:c7:bf:33:1f:09 and usually IP 192.168.1.159 (remember to verify per file)
163 // frame.len >= 556 && frame.len <= 558 && ip.addr == 192.168.1.159