Merge branch 'master' of https://github.uci.edu/rtrimana/smart_home_traffic
[pingpong.git] / Code / Projects / SmartPlugDetector / src / main / java / edu / uci / iotproject / Main.java
1 package edu.uci.iotproject;
2
3 import edu.uci.iotproject.analysis.PcapPacketPair;
4 import edu.uci.iotproject.analysis.TcpConversationUtils;
5 import edu.uci.iotproject.analysis.TriggerTrafficExtractor;
6 import edu.uci.iotproject.io.TriggerTimesFileReader;
7 import org.pcap4j.core.*;
8 import org.pcap4j.packet.IpV4Packet;
9 import org.pcap4j.packet.namednumber.DataLinkType;
10
11 import java.io.EOFException;
12 import java.net.UnknownHostException;
13 import java.time.Instant;
14 import java.util.*;
15 import java.util.concurrent.TimeoutException;
16
17 /**
18  * This is a system that reads PCAP files to compare
19  * patterns of DNS hostnames, packet sequences, and packet
20  * lengths with training data to determine certain events
21  * or actions for smart home devices.
22  *
23  * @author Janus Varmarken
24  * @author Rahmadi Trimananda (rtrimana@uci.edu)
25  * @version 0.1
26  */
27 public class Main {
28
29
30     public static void main(String[] args) throws PcapNativeException, NotOpenException, EOFException, TimeoutException, UnknownHostException {
31         // -------------------------------------------------------------------------------------------------------------
32         // Example/debug code for searching for a pattern at the MAC layer.
33 //        String fileName = "./pcap/mac-tplink.local.pcapng";
34 //        PcapHandle handle;
35 //        try {
36 //            handle = Pcaps.openOffline(fileName, PcapHandle.TimestampPrecision.NANO);
37 //        } catch (PcapNativeException pne) {
38 //            handle = Pcaps.openOffline(fileName);
39 //        }
40 //        Arrays.asList(1590, 1590, 1590, 1001, 337, 197, 636, 1311, 177) // Full pattern (all non-zero payload packets).
41 //        MacLayerFlowPattern pattern = new MacLayerFlowPattern("TP_LINK_LOCAL_OFF_MAC", "50:c7:bf:33:1f:09", Arrays.asList(637, 1312));
42 //        MacLayerFlowPatternFinder finder = new MacLayerFlowPatternFinder(handle, pattern);
43 //        finder.findFlowPattern();
44         // -------------------------------------------------------------------------------------------------------------
45 //
46 //        //final String fileName = args.length > 0 ? args[0] : "/home/rtrimana/pcap_processing/smart_home_traffic/Code/Projects/SmartPlugDetector/pcap/wlan1.local.dns.pcap";
47 //        final String fileName = args.length > 0 ? args[0] : "/scratch/June-2018/TPLink/wlan1/tplink.wlan1.local.pcap";
48 //        //final String fileName = args.length > 0 ? args[0] : "/scratch/June-2018/DLink/wlan1/dlink.wlan1.local.pcap";
49 //        final String trainingFileName = "./pcap/TP_LINK_LOCAL_ON_SUBSET.pcap";
50 ////        final String trainingFileName = "./pcap/TP_LINK_LOCAL_ON.pcap";
51 ////
52 ////        // ====== Debug code ======
53 //        PcapHandle handle;
54 //        PcapHandle trainingPcap;
55 //        try {
56 //            handle = Pcaps.openOffline(fileName, PcapHandle.TimestampPrecision.NANO);
57 //            trainingPcap = Pcaps.openOffline(trainingFileName, PcapHandle.TimestampPrecision.NANO);
58 //        } catch (PcapNativeException pne) {
59 //            handle = Pcaps.openOffline(fileName);
60 //            trainingPcap = Pcaps.openOffline(trainingFileName);
61 //        }
62 ////
63 ////        // TODO: The followings are the way to extract multiple hostnames and their associated packet lengths lists
64 ////        //List<String> list = new ArrayList<>();
65 ////        //list.add("events.tplinkra.com");
66 ////        //FlowPattern fp = new FlowPattern("TP_LINK_LOCAL_ON", list, trainingPcap);
67 ////        //List<String> list2 = new ArrayList<>();
68 ////        //list2.add("devs.tplinkcloud.com");
69 ////        //list2.add("events.tplinkra.com");
70 ////        //FlowPattern fp3 = new FlowPattern("TP_LINK_REMOTE_ON", list2, trainingPcap);
71 ////
72 //        FlowPattern fp = new FlowPattern("TP_LINK_LOCAL_ON", "events.tplinkra.com", trainingPcap);
73 //        //FlowPattern fp = new FlowPattern("DLINK_LOCAL_ON", "rfe-us-west-1.dch.dlink.com", trainingPcap);
74 //        FlowPatternFinder fpf = new FlowPatternFinder(handle, fp);
75 //        fpf.start();
76 ////
77 ////        // ========================
78
79         /*
80         PcapReader pcapReader = new PcapReader(args[0]);
81         PcapProcessingPipeline pipeline = new PcapProcessingPipeline(pcapReader);
82         TcpReassembler tcpReassembler = new TcpReassembler();
83         pipeline.addPcapPacketConsumer(tcpReassembler);
84         pipeline.executePipeline();
85         System.out.println("Pipeline terminated");
86
87         List<List<PcapPacketPair>> pairs = new ArrayList<>();
88         for (Conversation c : tcpReassembler.getTcpConversations()) {
89             pairs.add(TcpConversationUtils.extractPacketPairs(c));
90         }
91         */
92
93         /*
94         // -------- 07-17-2018 --------
95         // Only consider packets to/from the TP-Link plug.
96         PcapReader pcapReader = new PcapReader(args[0], "ip host 192.168.1.159");
97         TcpReassembler tcpReassembler = new TcpReassembler();
98         PcapPacket packet;
99         while((packet = pcapReader.readNextPacket()) != null) {
100             tcpReassembler.consumePacket(packet);
101         }
102         // Now we have a set of reassembled TCP conversations.
103         List<Conversation> conversations = tcpReassembler.getTcpConversations();
104         for(Conversation c : conversations) {
105             List<PcapPacketPair> pairs = TcpConversationUtils.extractPacketPairs(c);
106             for (PcapPacketPair pair : pairs) {
107                 // TODO ...
108                 // 1. discard packets that are not within X seconds after trigger time
109                 // 2. conversations may be (are) with different servers - so need to plot in different plots, one per hostname?
110             }
111         }
112
113         // ----------------------------
114         */
115
116         // -------- 07-19-2018 --------
117         TriggerTimesFileReader ttfr = new TriggerTimesFileReader();
118 //        List<Instant> triggerTimes = ttfr.readTriggerTimes("/Users/varmarken/Downloads/tplink-feb-13-2018.timestamps", false);
119         List<Instant> triggerTimes = ttfr.readTriggerTimes("/Users/varmarken/temp/UCI IoT Project/June2018 experiments/tplink/tplink-june-14-2018-timestamps.txt", false);
120 //        String pcapFile = "/Users/varmarken/Development/Repositories/UCI/NetworkingGroup/smart_home_traffic/Code/Projects/SmartPlugDetector/pcap/wlan1.local.dns.pcap";
121         String pcapFile = "/Users/varmarken/temp/UCI IoT Project/June2018 experiments/tplink/tplink.wlan1.local.pcap";
122         String tpLinkPlugIp = "192.168.1.159";
123         TriggerTrafficExtractor tte = new TriggerTrafficExtractor(pcapFile, triggerTimes, tpLinkPlugIp);
124 //        final PcapDumper outputter = Pcaps.openDead(DataLinkType.EN10MB, 65536).dumpOpen("/Users/varmarken/temp/traces/output/tplink-filtered.pcap");
125         final PcapDumper outputter = Pcaps.openDead(DataLinkType.EN10MB, 65536).dumpOpen("/Users/varmarken/temp/UCI IoT Project/June2018 experiments/tplink/tplink-filtered.pcap");
126         DnsMap dnsMap = new DnsMap();
127         TcpReassembler tcpReassembler = new TcpReassembler();
128         tte.performExtraction(pkt -> {
129             try {
130                 outputter.dump(pkt);
131             } catch (NotOpenException e) {
132                 e.printStackTrace();
133             }
134         }, dnsMap, tcpReassembler);
135         outputter.flush();
136         outputter.close();
137
138         /*
139         int packets = 0;
140         for (Conversation c : tcpReassembler.getTcpConversations()) {
141             packets += c.getPackets().size();
142             packets += c.getSynPackets().size();
143             // only count the FIN packets, not the ACKs; every FinAckPair holds a FIN packet
144             packets += c.getFinAckPairs().size();
145         }
146         // Produces 271 packets for the Feb 13 experiment
147         // Applying filter: "(tcp and not tcp.len == 0 and not tcp.analysis.retransmission and not tcp.analysis.fast_retransmission)  or (tcp.flags.syn == 1) or (tcp.flags.fin == 1)"
148         // to the file gives 295 packets, but there are 24 TCP-Out-Of-Order SYN/SYNACKs which are filtered as retransmissions in Conversation, so the numbers seem to match.
149         System.out.println("number of packets: " + packets);
150         */
151
152         List<List<PcapPacketPair>> pairs = new ArrayList<>();
153         for (Conversation c : tcpReassembler.getTcpConversations()) {
154             pairs.add(TcpConversationUtils.extractPacketPairs(c));
155         }
156         /*
157         // Sort pairs according to timestamp of first packet of conversation for (debugging) convenience.
158         Collections.sort(pairs, (l1, l2) -> {
159             if (l1.get(0).getFirst().getTimestamp().isBefore(l2.get(0).getFirst().getTimestamp())) return -1;
160             else if (l2.get(0).getFirst().getTimestamp().isBefore(l1.get(0).getFirst().getTimestamp())) return 1;
161             else return 0;
162         });
163         */
164         System.out.println("list of pairs produced");
165         List<PcapPacketPair> eventstplinkraPairs = new ArrayList<>();
166         List<List<PcapPacketPair>> otherPairs = new ArrayList<>();
167         String hostname = "events.tplinkra.com";
168         int emptyLists = 0;
169         for (List<PcapPacketPair> lppp : pairs) {
170             if (lppp.size() < 1) {
171                 emptyLists++;
172                 continue;
173             }
174             IpV4Packet ipPacket = lppp.get(0).getFirst().get(IpV4Packet.class);
175             // If packets are associated with the hostname
176             if (dnsMap.isRelatedToCloudServer(ipPacket.getHeader().getSrcAddr().getHostAddress(), hostname) ||
177                     dnsMap.isRelatedToCloudServer(ipPacket.getHeader().getDstAddr().getHostAddress(), hostname)) {
178                 eventstplinkraPairs.addAll(lppp);
179             } else {
180                 // Pairs associated with different server
181                 otherPairs.add(lppp);
182             }
183         }
184         System.out.println("number of empty list of packet pairs: " + emptyLists);
185         HashMap<String, Integer> pairCount = new HashMap<>();
186         for (PcapPacketPair ppp : eventstplinkraPairs) {
187             if (pairCount.containsKey(ppp.toString())) {
188                 pairCount.put(ppp.toString(), pairCount.get(ppp.toString()) + 1);
189             } else {
190                 pairCount.put(ppp.toString(), 1);
191             }
192         }
193         System.out.println("pairCount map built");
194
195         // Build map containing frequencies of packet lengths exchanged with events.tplinkra.com as well as a map with
196         // the frequencies of specific sequences of packet lengths for the same hostname
197         HashMap<Integer, Integer> eventstplinkraPacketLengthFreqMap = new HashMap<>();
198         HashMap<String, Integer> eventstplinkraPacketSequenceFreqMap = new HashMap<>();
199         for (Conversation c : tcpReassembler.getTcpConversations()) {
200             if (c.getPackets().size() == 0) {
201                 continue;
202             }
203             PcapPacket firstPacket = c.getPackets().get(0);
204             IpV4Packet firstPacketIp = firstPacket.get(IpV4Packet.class);
205             if (!dnsMap.isRelatedToCloudServer(firstPacketIp.getHeader().getSrcAddr().getHostAddress(), hostname) &&
206                     !dnsMap.isRelatedToCloudServer(firstPacketIp.getHeader().getDstAddr().getHostAddress(), hostname)) {
207                 continue;
208             }
209             // Update the packet length freq map
210             for (PcapPacket pp : c.getPackets()) {
211                 eventstplinkraPacketLengthFreqMap.merge(pp.length(), 1, (i1, i2) -> i1 + i2);
212             }
213             // Update the packet sequence freq map
214             StringBuilder sb = new StringBuilder();
215             for (PcapPacket pp : c.getPackets()) {
216                 sb.append(pp.length() + " ");
217             }
218             eventstplinkraPacketSequenceFreqMap.merge(sb.toString(), 1, (i1, i2) -> i1+i2);
219         }
220         System.out.println("packet length frequency map created");
221
222         Map<String, List<Conversation>> hostnameConversationMap =
223                 TcpConversationUtils.groupConversationsByHostname(tcpReassembler.getTcpConversations(), dnsMap);
224         System.out.println("hostnameConversationMap created");
225         // ----------------------------
226     }
227
228 }
229
230
231 // TP-Link MAC 50:c7:bf:33:1f:09 and usually IP 192.168.1.159 (remember to verify per file)