d342508a44f6a945ab3b6d3091e9ae096441efc6
[pingpong.git] / Code / Projects / PacketLevelSignatureExtractor / src / main / java / edu / uci / iotproject / SignatureGenerator.java
1 package edu.uci.iotproject;
2
3 import static edu.uci.iotproject.analysis.UserAction.Type;
4
5 import edu.uci.iotproject.analysis.*;
6 import edu.uci.iotproject.io.TriggerTimesFileReader;
7 import edu.uci.iotproject.trafficreassembly.layer3.Conversation;
8 import edu.uci.iotproject.trafficreassembly.layer3.TcpReassembler;
9 import edu.uci.iotproject.util.PcapPacketUtils;
10 import edu.uci.iotproject.util.PrintUtils;
11 import org.apache.commons.math3.stat.clustering.Cluster;
12 import org.apache.commons.math3.stat.clustering.DBSCANClusterer;
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.Duration;
19 import java.time.Instant;
20 import java.util.*;
21 import java.util.concurrent.TimeoutException;
22 import java.util.stream.Collectors;
23 import java.util.stream.Stream;
24
25 /**
26  * This is a system that reads PCAP files to compare
27  * patterns of DNS hostnames, packet sequences, and packet
28  * lengths with training data to determine certain events
29  * or actions for smart home devices.
30  *
31  * @author Janus Varmarken
32  * @author Rahmadi Trimananda (rtrimana@uci.edu)
33  * @version 0.1
34  */
35 public class SignatureGenerator {
36
37
38     public static void main(String[] args) throws PcapNativeException, NotOpenException, EOFException, TimeoutException, UnknownHostException {
39         // -------------------------------------------------------------------------------------------------------------
40         // ------------ # Code for extracting traffic generated by a device within x seconds of a trigger # ------------
41         // Paths to input and output files (consider supplying these as arguments instead) and IP of the device for
42         // which traffic is to be extracted:
43         String path = "/scratch/July-2018"; // Rahmadi
44 //        String path = "/Users/varmarken/temp/UCI IoT Project/experiments"; // Janus
45         boolean verbose = true;
46         final String onPairsPath = "/scratch/July-2018/on.txt";
47         final String offPairsPath = "/scratch/July-2018/off.txt";
48
49         // 1) TODO: D-LINK PLUG July 26 experiment
50 //        final String inputPcapFile = path + "/2018-07/dlink/dlink.wlan1.local.pcap";
51 //        final String outputPcapFile = path + "/2018-07/dlink/dlink-processed.pcap";
52 //        final String triggerTimesFile = path + "/2018-07/dlink/dlink-july-26-2018.timestamps";
53 //        final String deviceIp = "192.168.1.199"; // .246 == phone; .199 == dlink plug?
54           // Actual training
55 //        final String inputPcapFile = path + "/2018-10/dlink-plug/dlink-plug.wlan1.local.pcap";
56 //        final String outputPcapFile = path + "/2018-10/dlink-plug/dlink-plug-processed.pcap";
57 //        final String triggerTimesFile = path + "/2018-10/dlink-plug/dlink-plug-oct-17-2018.timestamps";
58 //        final String deviceIp = "192.168.1.199"; // .246 == phone; .199 == dlink plug?
59         // TODO: EXPERIMENT - November 7, 2018
60 //        final String inputPcapFile = path + "/experimental_result/standalone/dlink-plug/wlan1/dlink-plug.wlan1.local.pcap";
61 //        final String outputPcapFile = path + "/experimental_result/standalone/dlink-plug/wlan1/dlink-plug-processed.pcap";
62 //        final String triggerTimesFile = path + "/experimental_result/standalone/dlink-plug/timestamps/dlink-plug-nov-7-2018.timestamps";
63 ////        final String deviceIp = "192.168.1.199"; // .246 == phone; .199 == dlink plug?
64 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .199 == dlink plug?
65
66         // 2) TODO: TP-LINK PLUG July 25 experiment
67 //        final String inputPcapFile = path + "/2018-07/tplink/tplink.wlan1.local.pcap";
68 //        final String outputPcapFile = path + "/2018-07/tplink/tplink-processed.pcap";
69 //        final String triggerTimesFile = path + "/2018-07/tplink/tplink-july-25-2018.timestamps";
70 //        final String deviceIp = "192.168.1.159";
71           // Actual training
72 //        final String inputPcapFile = path + "/2018-10/tplink-plug/tplink-plug.wlan1.local.pcap";
73 //        final String outputPcapFile = path + "/2018-10/tplink-plug/tplink-plug-processed.pcap";
74 //        final String triggerTimesFile = path + "/2018-10/tplink-plug/tplink-plug-oct-17-2018.timestamps";
75 //        final String deviceIp = "192.168.1.159"; // .246 == phone; .159 == tplink plug
76         // TODO: EXPERIMENT - November 8, 2018
77 //        final String inputPcapFile = path + "/experimental_result/standalone/tplink-plug/wlan1/tplink-plug.wlan1.local.pcap";
78 //        final String outputPcapFile = path + "/experimental_result/standalone/tplink-plug/wlan1/tplink-plug-processed.pcap";
79 //        final String triggerTimesFile = path + "/experimental_result/standalone/tplink-plug/timestamps/tplink-plug-nov-8-2018.timestamps";
80 //        final String deviceIp = "192.168.1.159"; // .246 == phone; .159 == tplink plug
81 ////        final String deviceIp = "192.168.1.246"; // .246 == phone; .159 == tplink plug
82
83         // 2b) TP-Link July 25 experiment TRUNCATED:
84         // Only contains "true local" events, i.e., before the behavior changes to remote-like behavior.
85         // Last included event is at July 25 10:38:11; file filtered to only include packets with arrival time <= 10:38:27.
86 //        final String inputPcapFile = path + "/2018-07/tplink/tplink.wlan1.local.truncated.pcap";
87 //        final String outputPcapFile = path + "/2018-07/tplink/tplink-processed.truncated.pcap";
88 //        final String triggerTimesFile = path + "/2018-07/tplink/tplink-july-25-2018.truncated.timestamps";
89 //        final String deviceIp = "192.168.1.159";
90
91         // 3) TODO: SMARTTHINGS PLUG July 25 experiment
92 //        final String inputPcapFile = path + "/2018-07/stplug/stplug.wlan1.local.pcap";
93 //        final String outputPcapFile = path + "/2018-07/stplug/stplug-processed.pcap";
94 //        final String triggerTimesFile = path + "/2018-07/stplug/smartthings-july-25-2018.timestamps";
95 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
96         // October 18
97 //        final String inputPcapFile = path + "/2018-10/st-plug/st-plug.wlan1.local.pcap";
98 //        final String outputPcapFile = path + "/2018-10/st-plug/st-plug-processed.pcap";
99 //        final String triggerTimesFile = path + "/2018-10/st-plug/st-plug-oct-18-2018.timestamps";
100 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
101         // TODO: EXPERIMENT - November 12, 2018
102 //        final String inputPcapFile = path + "/experimental_result/standalone/st-plug/wlan1/st-plug.wlan1.local.pcap";
103 //        final String outputPcapFile = path + "/experimental_result/standalone/st-plug/wlan1/st-plug-processed.pcap";
104 ////        final String inputPcapFile = path + "/experimental_result/standalone/st-plug/eth1/st-plug.eth1.local.pcap";
105 ////        final String outputPcapFile = path + "/experimental_result/standalone/st-plug/eth1/st-plug-processed.pcap";
106 //        final String triggerTimesFile = path + "/experimental_result/standalone/st-plug/timestamps/st-plug-nov-12-2018.timestamps";
107 ////        final String deviceIp = "192.168.1.142"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
108 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
109
110         // 4) TODO: WEMO PLUG July 30 experiment
111 //        final String inputPcapFile = path + "/2018-07/wemo/wemo.wlan1.local.pcap";
112 //        final String outputPcapFile = path + "/2018-07/wemo/wemo-processed.pcap";
113 //        final String triggerTimesFile = path + "/2018-07/wemo/wemo-july-30-2018.timestamps";
114 //        final String deviceIp = "192.168.1.145";  // .246 == phone; .145 == WeMo
115         // TODO: EXPERIMENT - November 20, 2018
116 //        final String inputPcapFile = path + "/experimental_result/standalone/wemo-plug/wlan1/wemo-plug.wlan1.local.pcap";
117 //        final String outputPcapFile = path + "/experimental_result/standalone/wemo-plug/wlan1/wemo-plug-processed.pcap";
118 //        final String triggerTimesFile = path + "/experimental_result/standalone/wemo-plug/timestamps/wemo-plug-nov-20-2018.timestamps";
119 ////        final String deviceIp = "192.168.1.145"; // .246 == phone; .145 == WeMo
120 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .145 == WeMo
121
122         // 5) TODO: WEMO INSIGHT July 31 experiment
123 //        final String inputPcapFile = path + "/2018-07/wemoinsight/wemoinsight.wlan1.local.pcap";
124 //        final String outputPcapFile = path + "/2018-07/wemoinsight/wemoinsight-processed.pcap";
125 //        final String triggerTimesFile = path + "/2018-07/wemoinsight/wemo-insight-july-31-2018.timestamps";
126 //        final String deviceIp = "192.168.1.135";
127         // TODO: EXPERIMENT - November 21, 2018
128         final String inputPcapFile = path + "/experimental_result/standalone/wemo-insight-plug/wlan1/wemo-insight-plug.wlan1.local.pcap";
129         final String outputPcapFile = path + "/experimental_result/standalone/wemo-insight-plug/wlan1/wemo-insight-plug-processed.pcap";
130         final String triggerTimesFile = path + "/experimental_result/standalone/wemo-insight-plug/timestamps/wemo-insight-plug-nov-21-2018.timestamps";
131 //        final String deviceIp = "192.168.1.145"; // .246 == phone; .135 == WeMo Insight
132         final String deviceIp = "192.168.1.246"; // .246 == phone; .135 == WeMo Insight
133
134         // 6) TODO: TP-LINK BULB August 1 experiment
135 //        final String inputPcapFile = path + "/2018-08/tplink-bulb/tplinkbulb.wlan1.local.pcap";
136 //        final String outputPcapFile = path + "/2018-08/tplink-bulb/tplinkbulb-processed.pcap";
137 //        final String triggerTimesFile = path + "/2018-08/tplink-bulb/tplink-bulb-aug-3-2018.timestamps";
138 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .140 == TP-Link bulb
139         // TODO: EXPERIMENT - November 16, 2018
140 //        final String inputPcapFile = path + "/experimental_result/standalone/tplink-bulb/wlan1/tplink-bulb.wlan1.local.pcap";
141 //        final String outputPcapFile = path + "/experimental_result/standalone/tplink-bulb/wlan1/tplink-bulb-processed.pcap";
142 ////        final String inputPcapFile = path + "/experimental_result/standalone/tplink-bulb/eth0/tplink-bulb.eth1.local.pcap";
143 ////        final String outputPcapFile = path + "/experimental_result/standalone/tplink-bulb/eth0/tplink-bulb-processed.pcap";
144 //        final String triggerTimesFile = path + "/experimental_result/standalone/tplink-bulb/timestamps/tplink-bulb-nov-16-2018.timestamps";
145 ////        final String deviceIp = "192.168.1.140"; // .246 == phone; .140 == TP-Link bulb
146 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .140 == TP-Link bulb
147
148         // 7) TODO: KWIKSET DOORLOCK August 6 experiment
149 //        final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock.data.wlan1.pcap";
150 ////        final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock.wlan1.local.pcap";
151 //        final String outputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-processed.pcap";
152 ////        final String triggerTimesFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-aug-6-2018.timestamps";
153 //        final String triggerTimesFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-8hr-data-oct-11-2018.timestamps";
154 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
155         // TODO: EXPERIMENT - November 10, 2018
156 //        final String inputPcapFile = path + "/experimental_result/standalone/kwikset-doorlock/wlan1/kwikset-doorlock.wlan1.local.pcap";
157 //        final String outputPcapFile = path + "/experimental_result/standalone/kwikset-doorlock/wlan1/kwikset-doorlock-processed.pcap";
158 ////        final String inputPcapFile = path + "/experimental_result/standalone/kwikset-doorlock/eth1/kwikset-doorlock.eth1.local.pcap";
159 ////        final String outputPcapFile = path + "/experimental_result/standalone/kwikset-doorlock/eth1/kwikset-doorlock-processed.pcap";
160 //        final String triggerTimesFile = path + "/experimental_result/standalone/kwikset-doorlock/timestamps/kwikset-doorlock-nov-10-2018.timestamps";
161 ////        final String deviceIp = "192.168.1.142"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
162 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
163
164         // September 12, 2018 - includes both wlan1 and eth1 interfaces
165 //        final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset3.wlan1.local.pcap";
166 //        //final String inputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset3.eth1.local.pcap";
167 //        final String outputPcapFile = path + "/2018-08/kwikset-doorlock/kwikset3-processed.pcap";
168 //        final String triggerTimesFile = path + "/2018-08/kwikset-doorlock/kwikset-doorlock-sept-12-2018.timestamps";
169 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .142 == SmartThings Hub (note: use eth0 capture for this!)
170
171         // 8) TODO: HUE BULB August 7 experiment
172 //        final String inputPcapFile = path + "/2018-08/hue-bulb/hue-bulb.wlan1.local.pcap";
173 //        final String outputPcapFile = path + "/2018-08/hue-bulb/hue-bulb-processed.pcap";
174 //        final String triggerTimesFile = path + "/2018-08/hue-bulb/hue-bulb-aug-7-2018.timestamps";
175 //        final String deviceIp = "192.168.1.246";
176         // October 30 experiment
177 //        final String inputPcapFile = path + "/2018-10/hue-bulb/hue-bulb.wlan1.local.pcap";
178 //        final String outputPcapFile = path + "/2018-10/hue-bulb/hue-bulb-processed.pcap";
179 //        final String triggerTimesFile = path + "/2018-10/hue-bulb/hue-bulb-oct-30-2018.timestamps";
180 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .100 == Hue hub
181         // TODO: EXPERIMENT - November 19, 2018
182 //        final String inputPcapFile = path + "/experimental_result/standalone/hue-bulb/wlan1/hue-bulb.wlan1.local.pcap";
183 //        final String outputPcapFile = path + "/experimental_result/standalone/hue-bulb/wlan1/hue-bulb-processed.pcap";
184 //        final String triggerTimesFile = path + "/experimental_result/standalone/hue-bulb/timestamps/hue-bulb-nov-19-2018.timestamps";
185 ////        final String deviceIp = "192.168.1.100"; // .246 == phone; .100 == Hue hub
186 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .100 == Hue hub
187
188         // 9) TODO: LIFX BULB August 8 experiment
189 //        final String inputPcapFile = path + "/2018-08/lifx-bulb/lifx-bulb.wlan1.local.pcap";
190 //        final String outputPcapFile = path + "/2018-08/lifx-bulb/lifx-bulb-processed.pcap";
191 //        final String triggerTimesFile = path + "/2018-08/lifx-bulb/lifx-bulb-aug-8-2018.timestamps";
192 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .231 == Lifx
193         // October 18
194 //        final String inputPcapFile = path + "/2018-10/lifx-bulb/lifx-bulb.wlan1.local.pcap";
195 //        final String outputPcapFile = path + "/2018-10/lifx-bulb/lifx-bulb-processed.pcap";
196 //        final String triggerTimesFile = path + "/2018-10/lifx-bulb/lifx-bulb-oct-18-2018.timestamps";
197 //        final String deviceIp = "192.168.1.231"; // .246 == phone; .231 == Lifx
198         // November 1
199 //        final String inputPcapFile = path + "/2018-10/lifx-bulb/lifx-bulb.wlan1.local.pcap";
200 //        final String outputPcapFile = path + "/2018-10/lifx-bulb/lifx-bulb-processed.pcap";
201 //        final String triggerTimesFile = path + "/2018-10/lifx-bulb/lifx-bulb-nov-1-2018.timestamps";
202 //        final String deviceIp = "192.168.1.231"; // .246 == phone; .231 == Lifx
203
204         // 10) TODO: AMCREST CAMERA August 9 experiment
205 //        final String inputPcapFile = path + "/2018-08/amcrest-camera/amcrest-camera.wlan1.local.pcap";
206 //        final String outputPcapFile = path + "/2018-08/amcrest-camera/amcrest-camera-processed.pcap";
207 //        final String triggerTimesFile = path + "/2018-08/amcrest-camera/amcrest-camera-aug-9-2018.timestamps";
208 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .235 == camera
209
210         // 11) TODO: ARLO CAMERA August 10 experiment
211 //        final String inputPcapFile = path + "/2018-08/arlo-camera/arlo-camera.wlan1.local.pcap";
212 //        final String outputPcapFile = path + "/2018-08/arlo-camera/arlo-camera-processed.pcap";
213 //        final String triggerTimesFile = path + "/2018-08/arlo-camera/arlo-camera-aug-10-2018.timestamps";
214 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .140 == camera
215         // TODO: EXPERIMENT - November 13, 2018
216 //        final String inputPcapFile = path + "/experimental_result/standalone/arlo-camera/wlan1/arlo-camera.wlan1.local.pcap";
217 //        final String outputPcapFile = path + "/experimental_result/standalone/arlo-camera/wlan1/arlo-camera-processed.pcap";
218 ////        final String inputPcapFile = path + "/experimental_result/standalone/arlo-camera/eth0/arlo-camera.eth1.local.pcap";
219 ////        final String outputPcapFile = path + "/experimental_result/standalone/arlo-camera/eth0/arlo-camera-processed.pcap";
220 //        final String triggerTimesFile = path + "/experimental_result/standalone/arlo-camera/timestamps/arlo-camera-nov-13-2018.timestamps";
221 ////        final String deviceIp = "192.168.1.140"; // .246 == phone; .140 == camera
222 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .140 == camera
223
224         // 12) TODO: BLOSSOM SPRINKLER August 13 experiment
225 //        final String inputPcapFile = path + "/2018-08/blossom/blossom.wlan1.local.pcap";
226 //        final String outputPcapFile = path + "/2018-08/blossom/blossom-processed.pcap";
227 //        final String triggerTimesFile = path + "/2018-08/blossom/blossom-aug-13-2018.timestamps";
228 //        final String deviceIp = "192.168.1.229"; // .246 == phone; .229 == sprinkler
229 //        // 2 November
230 //        final String inputPcapFile = path + "/2018-10/blossom-sprinkler/blossom-sprinkler.wlan1.local.pcap";
231 //        final String outputPcapFile = path + "/2018-10/blossom-sprinkler/blossom-sprinkler-processed.pcap";
232 //        final String triggerTimesFile = path + "/2018-10/blossom-sprinkler/blossom-sprinkler-nov-2-2018.timestamps";
233 //        final String deviceIp = "192.168.1.229"; // .246 == phone; .229 == sprinkler
234         // TODO: EXPERIMENT - January 9, 11, 13, 14
235 //        final String inputPcapFile = path + "/experimental_result/standalone/blossom-sprinkler/wlan1/blossom-sprinkler.wlan1.local.pcap";
236 //        final String outputPcapFile = path + "/experimental_result/standalone/blossom-sprinkler/wlan1/blossom-sprinkler-processed.pcap";
237 //        final String triggerTimesFile = path + "/experimental_result/standalone/blossom-sprinkler/timestamps/blossom-sprinkler-standalone-jan-14-2019.timestamps";
238 ////        final String triggerTimesFile = path + "/experimental_result/standalone/blossom-sprinkler/timestamps/blossom-sprinkler-standalone-jan-11-2019.timestamps";
239 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .229 == sprinkler
240 ////        final String deviceIp = "192.168.1.229"; // .246 == phone; .229 == sprinkler
241
242 //        // 13) TODO: D-LINK SIREN August 14 experiment
243 //        final String inputPcapFile = path + "/2018-08/dlink-siren/dlink-siren.wlan1.local.pcap";
244 //        //final String inputPcapFile = path + "/evaluation/dlink-siren/dlink-siren.data.wlan1.pcap";
245 //        final String outputPcapFile = path + "/2018-08/dlink-siren/dlink-siren-processed.pcap";
246 //        final String triggerTimesFile = path + "/2018-08/dlink-siren/dlink-siren-oct-12-2018.timestamps";
247 //        //final String triggerTimesFile = path + "/2018-08/dlink-siren/dlink-siren-aug-14-2018.timestamps";
248 //        //final String triggerTimesFile = path + "/actual/timestamps/dlink-siren-8hr-data-oct-10-2018.timestamps";
249 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .183 == siren
250         // TODO: EXPERIMENT - November 9, 2018
251 //        final String inputPcapFile = path + "/experimental_result/standalone/dlink-siren/wlan1/dlink-siren.wlan1.local.pcap";
252 //        final String outputPcapFile = path + "/experimental_result/standalone/dlink-siren/wlan1/dlink-siren-processed.pcap";
253 //        final String triggerTimesFile = path + "/experimental_result/standalone/dlink-siren/timestamps/dlink-siren-nov-9-2018.timestamps";
254 ////        final String deviceIp = "192.168.1.183"; // .246 == phone; .183 == siren
255 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .183 == siren
256
257         // 14) TODO: NEST THERMOSTAT August 15 experiment
258 //        final String inputPcapFile = path + "/2018-08/nest/nest.wlan1.local.pcap";
259 //        final String outputPcapFile = path + "/2018-08/nest/nest-processed.pcap";
260 //        final String triggerTimesFile = path + "/2018-08/nest/nest-aug-15-2018.timestamps";
261 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .127 == Nest thermostat
262 //        // TODO: EXPERIMENT - November 14, 2018
263 //        final String inputPcapFile = path + "/experimental_result/standalone/nest-thermostat/wlan1/nest-thermostat.wlan1.local.pcap";
264 //        final String outputPcapFile = path + "/experimental_result/standalone/nest-thermostat/wlan1/nest-thermostat-processed.pcap";
265 ////        final String inputPcapFile = path + "/experimental_result/standalone/nest-thermostat/eth0/nest-thermostat.eth1.local.pcap";
266 ////        final String outputPcapFile = path + "/experimental_result/standalone/nest-thermostat/eth0/nest-thermostat-processed.pcap";
267 //        final String triggerTimesFile = path + "/experimental_result/standalone/nest-thermostat/timestamps/nest-thermostat-nov-15-2018.timestamps";
268 ////        final String deviceIp = "192.168.1.127"; // .246 == phone; .127 == Nest thermostat
269 //        final String deviceIp = "192.168.1.246"; // .246 == phone; .127 == Nest thermostat
270
271         // 15) Alexa August 16 experiment
272 //        final String inputPcapFile = path + "/2018-08/alexa/alexa.wlan1.local.pcap";
273 //        final String outputPcapFile = path + "/2018-08/alexa/alexa-processed.pcap";
274 //        final String triggerTimesFile = path + "/2018-08/alexa/alexa-aug-16-2018.timestamps";
275 //        final String deviceIp = "192.168.1.225"; // .246 == phone; .225 == Alexa
276         // August 17
277 //        final String inputPcapFile = path + "/2018-08/alexa/alexa2.wlan1.local.pcap";
278 //        final String outputPcapFile = path + "/2018-08/alexa/alexa2-processed.pcap";
279 //        final String triggerTimesFile = path + "/2018-08/alexa/alexa-aug-17-2018.timestamps";
280 //        final String deviceIp = "192.168.1.225"; // .246 == phone; .225 == Alexa
281
282         // September 17
283 //        final String inputPcapFile = path + "/2018-08/noise/noise.eth1.pcap";
284 //        final String outputPcapFile = path + "/2018-08/noise/noise-processed.pcap";
285 //        final String triggerTimesFile = path + "/2018-08/noise/noise-sept-17-2018.timestamps";
286 //        final String deviceIp = "192.168.1.142"; //  .142 == SmartThings Hub; .199 == dlink plug; .183 == siren
287         // September 26 - D-Link noise
288 //        final String inputPcapFile = path + "/2018-08/noise/noise.dlink.wlan1.pcap";
289 //        final String outputPcapFile = path + "/2018-08/noise/noise-processed.pcap";
290 //        final String triggerTimesFile = path + "/2018-08/noise/dlink-noise-sept-26-2018.timestamps";
291 //        final String deviceIp = "192.168.1.183"; //  .199 == dlink plug; .183 == siren
292         // September 27 - Kwikset noise
293 //        final String inputPcapFile = path + "/2018-08/noise/noise.kwikset.eth1.pcap";
294 //        final String outputPcapFile = path + "/2018-08/noise/noise-processed.pcap";
295 //        final String triggerTimesFile = path + "/2018-08/noise/kwikset-doorlock-noise-sept-27-2018.timestamps";
296 //        final String deviceIp = "192.168.1.142"; //  .142 == SmartThings Hub;
297
298         // TODO: The below part is just for 15-second time sensitivity experiment
299         // TODO: The below part is just for 15-second time sensitivity experiment
300         // TODO: The below part is just for 15-second time sensitivity experiment
301         // D-Link plug
302 //        final String triggerTimesFile = path + "/experimental_result/standalone/dlink-plug/timestamps/dlink-plug-nov-7-2018.timestamps";
303 ////        final String onSignatureFile = path + "/experimental_result/standalone/dlink-plug/signatures/dlink-plug-onSignature-phone-side.sig";
304 ////        final String offSignatureFile = path + "/experimental_result/standalone/dlink-plug/signatures/dlink-plug-offSignature-phone-side.sig";
305 //        final String onSignatureFile = path + "/experimental_result/standalone/dlink-plug/signatures/dlink-plug-onSignature-device-side.sig";
306 //        final String offSignatureFile = path + "/experimental_result/standalone/dlink-plug/signatures/dlink-plug-offSignature-device-side.sig";
307         // TP-Link plug
308 //        final String triggerTimesFile = path + "/experimental_result/standalone/tplink-plug/timestamps/tplink-plug-nov-8-2018.timestamps";
309 //////        final String onSignatureFile = path + "/experimental_result/standalone/tplink-plug/signatures/tplink-plug-onSignature-phone-side.sig";
310 //////        final String offSignatureFile = path + "/experimental_result/standalone/tplink-plug/signatures/tplink-plug-offSignature-phone-side.sig";
311 ////        final String onSignatureFile = path + "/experimental_result/standalone/tplink-plug/signatures/tplink-plug-onSignature-device-side-outbound.sig";
312 ////        final String offSignatureFile = path + "/experimental_result/standalone/tplink-plug/signatures/tplink-plug-offSignature-device-side-outbound.sig";
313 //        final String onSignatureFile = path + "/experimental_result/standalone/tplink-plug/signatures/tplink-plug-onSignature-device-side.sig";
314 //        final String offSignatureFile = path + "/experimental_result/standalone/tplink-plug/signatures/tplink-plug-offSignature-device-side.sig";
315
316         // D-Link siren
317 //        final String triggerTimesFile = path + "/experimental_result/standalone/dlink-siren/timestamps/dlink-siren-nov-9-2018.timestamps";
318 //        final String onSignatureFile = path + "/experimental_result/standalone/dlink-siren/signatures/dlink-siren-onSignature-phone-side.sig";
319 //        final String offSignatureFile = path + "/experimental_result/standalone/dlink-siren/signatures/dlink-siren-offSignature-phone-side.sig";
320         // Kwikset door lock
321 //        final String triggerTimesFile = path + "/experimental_result/standalone/kwikset-doorlock/timestamps/kwikset-doorlock-nov-10-2018.timestamps";
322 //        final String onSignatureFile = path + "/experimental_result/standalone/kwikset-doorlock/signatures/kwikset-doorlock-onSignature-phone-side.sig";
323 //        final String offSignatureFile = path + "/experimental_result/standalone/kwikset-doorlock/signatures/kwikset-doorlock-offSignature-phone-side.sig";
324         // SmartThings plug
325 //        final String triggerTimesFile = path + "/experimental_result/standalone/st-plug/timestamps/st-plug-nov-12-2018.timestamps";
326 //        final String onSignatureFile = path + "/experimental_result/standalone/st-plug/signatures/st-plug-onSignature-phone-side.sig";
327 //        final String offSignatureFile = path + "/experimental_result/standalone/st-plug/signatures/st-plug-offSignature-phone-side.sig";
328         // Arlo Q
329 //        final String triggerTimesFile = path + "/experimental_result/standalone/arlo-camera/timestamps/arlo-camera-nov-13-2018.timestamps";
330 //        final String onSignatureFile = path + "/experimental_result/standalone/arlo-camera/signatures/arlo-camera-onSignature-phone-side.sig";
331 //        final String offSignatureFile = path + "/experimental_result/standalone/arlo-camera/signatures/arlo-camera-offSignature-phone-side.sig";
332         // Nest thermostat
333 //        final String triggerTimesFile = path + "/experimental_result/standalone/nest-thermostat/timestamps/nest-thermostat-nov-15-2018.timestamps";
334 //        final String onSignatureFile = path + "/experimental_result/standalone/nest-thermostat/signatures/nest-thermostat-onSignature-phone-side.sig";
335 //        final String offSignatureFile = path + "/experimental_result/standalone/nest-thermostat/signatures/nest-thermostat-offSignature-phone-side.sig";
336         // Blossom sprinkler
337 //        final String triggerTimesFile = path + "/experimental_result/standalone/blossom-sprinkler/timestamps/blossom-sprinkler-standalone-jan-14-2019.timestamps";
338 //        final String onSignatureFile = path + "/experimental_result/standalone/blossom-sprinkler/signatures/blossom-sprinkler-onSignature-device-side.sig";
339 //        final String offSignatureFile = path + "/experimental_result/standalone/blossom-sprinkler/signatures/blossom-sprinkler-offSignature-device-side.sig";
340 //        final String onSignatureFile = path + "/experimental_result/standalone/blossom-sprinkler/signatures/blossom-sprinkler-onSignature-phone-side.sig";
341 //        final String offSignatureFile = path + "/experimental_result/standalone/blossom-sprinkler/signatures/blossom-sprinkler-offSignature-phone-side.sig";
342         // TP-Link bulb
343 //        final String triggerTimesFile = path + "/experimental_result/standalone/tplink-bulb/timestamps/tplink-bulb-nov-16-2018.timestamps";
344 //        final String onSignatureFile = path + "/experimental_result/standalone/tplink-bulb/signatures/tplink-bulb-onSignature-phone-side.sig";
345 //        final String offSignatureFile = path + "/experimental_result/standalone/tplink-bulb/signatures/tplink-bulb-offSignature-phone-side.sig";
346         // Philips hue
347 //        final String triggerTimesFile = path + "/2018-08/hue-bulb/hue-bulb-aug-7-2018.timestamps";
348 //        final String onSignatureFile = path + "/training/hue-bulb/signatures/hue-bulb-onSignature-phone-side.sig";
349 //        final String offSignatureFile = path + "/training/hue-bulb/signatures/hue-bulb-offSignature-phone-side.sig";
350         // WeMo plug
351 //        final String triggerTimesFile = path + "/experimental_result/standalone/wemo-plug/timestamps/wemo-plug-nov-20-2018.timestamps";
352 //        final String onSignatureFile = path + "/experimental_result/standalone/wemo-plug/signatures/wemo-plug-onSignature-phone-side.sig";
353 //        final String offSignatureFile = path + "/experimental_result/standalone/wemo-plug/signatures/wemo-plug-offSignature-phone-side.sig";
354         // WeMo Insight plug
355 //        final String triggerTimesFile = path + "/experimental_result/standalone/wemo-insight-plug/timestamps/wemo-insight-plug-nov-21-2018.timestamps";
356 //        final String onSignatureFile = path + "/experimental_result/standalone/wemo-insight-plug/signatures/wemo-insight-plug-onSignature-phone-side.sig";
357 //        final String offSignatureFile = path + "/experimental_result/standalone/wemo-insight-plug/signatures/wemo-insight-plug-offSignature-phone-side.sig";
358
359
360 //        TriggerTimesFileReader ttfr = new TriggerTimesFileReader();
361 //        List<Instant> triggerTimes = ttfr.readTriggerTimes(triggerTimesFile, false);
362 //
363 //        System.out.println("ON signature file in use is " + onSignatureFile);
364 //        System.out.println("OFF signature file in use is " + offSignatureFile);
365 //
366 //        List<List<List<PcapPacket>>> onSignature = PrintUtils.deserializeSignatureFromFile(onSignatureFile);
367 //        List<List<List<PcapPacket>>> offSignature = PrintUtils.deserializeSignatureFromFile(offSignatureFile);
368 //
369 //        List<Instant> signatureTimestamps = new ArrayList<>();
370 //        // Load ON signature last packet's timestamp
371 //        // Get the last only
372 //        List<List<PcapPacket>> lastListOn = onSignature.get(onSignature.size()-1);
373 //        for (List<PcapPacket> list : lastListOn) {
374 //            // Get timestamp Instant from the last packet
375 //            int lastPacketIndex = list.size()-1;
376 //            signatureTimestamps.add(list.get(lastPacketIndex).getTimestamp());
377 //        }
378 //        // Load OFF signature last packet's timestamp
379 //        // Get the last only
380 //        List<List<PcapPacket>> lastListOff = offSignature.get(offSignature.size()-1);
381 //        for (List<PcapPacket> list : lastListOff) {
382 //            // Get timestamp Instant from the last packet
383 //            int lastPacketIndex = list.size()-1;
384 //            signatureTimestamps.add(list.get(lastPacketIndex).getTimestamp());
385 //        }
386 //        // Sort the timestamps
387 //        signatureTimestamps.sort((p1, p2) -> {
388 //            return p1.compareTo(p2);
389 //        });
390 //
391 //        Iterator<Instant> iterTrig = triggerTimes.iterator();
392 //        Iterator<Instant> iterSign = signatureTimestamps.iterator();
393 //        System.out.println("Trigger to Last Packet:");
394 //        while (iterTrig.hasNext() && iterSign.hasNext()) {
395 //            Instant trigInst = (Instant) iterTrig.next();
396 //            Instant signInst = (Instant) iterSign.next();
397 //            Duration dur = Duration.between(trigInst, signInst);
398 //            long duration = dur.toMillis();
399 //            // Check duration --- should be below 15 seconds
400 //            if (duration >= 0 && duration <= 15000) {
401 //                System.out.println(dur.toMillis());
402 //            } else if (duration > 15000) {
403 //                while (duration > 15000) { // that means we have to move to the next trigger
404 //                    trigInst = (Instant) iterTrig.next();
405 //                    dur = Duration.between(trigInst, signInst);
406 //                    duration = dur.toMillis();
407 //                }
408 //                System.out.println(dur.toMillis());
409 //            } else { // below 0 / negative --- that means we have to move to the next signature
410 //                while (duration < 0) { // that means we have to move to the next trigger
411 //                    signInst = (Instant) iterSign.next();
412 //                    dur = Duration.between(trigInst, signInst);
413 //                    duration = dur.toMillis();
414 //                }
415 //                System.out.println(dur.toMillis());
416 //            }
417 //        }
418 //
419 //
420 //        // ==========================================================================
421 //        List<Instant> firstSignatureTimestamps = new ArrayList<>();
422 //        List<Instant> lastSignatureTimestamps = new ArrayList<>();
423 //        List<List<PcapPacket>> firstListOnSign = onSignature.get(0);
424 //        List<List<PcapPacket>> lastListOnSign = onSignature.get(onSignature.size()-1);
425 //        // Load ON signature first and last packet's timestamps
426 //        for (List<PcapPacket> list : firstListOnSign) {
427 //            // Get timestamp Instant from the last packet
428 //            firstSignatureTimestamps.add(list.get(0).getTimestamp());
429 //        }
430 //        for (List<PcapPacket> list : lastListOnSign) {
431 //            // Get timestamp Instant from the last packet
432 //            int lastPacketIndex = list.size()-1;
433 //            lastSignatureTimestamps.add(list.get(lastPacketIndex).getTimestamp());
434 //        }
435 //
436 //        List<List<PcapPacket>> firstListOffSign = offSignature.get(0);
437 //        List<List<PcapPacket>> lastListOffSign = offSignature.get(offSignature.size()-1);
438 //        // Load OFF signature first and last packet's timestamps
439 //        for (List<PcapPacket> list : firstListOffSign) {
440 //            // Get timestamp Instant from the last packet
441 //            firstSignatureTimestamps.add(list.get(0).getTimestamp());
442 //        }
443 //        for (List<PcapPacket> list : lastListOffSign) {
444 //            // Get timestamp Instant from the last packet
445 //            int lastPacketIndex = list.size()-1;
446 //            lastSignatureTimestamps.add(list.get(lastPacketIndex).getTimestamp());
447 //        }
448 //        // Sort the timestamps
449 //        firstSignatureTimestamps.sort((p1, p2) -> {
450 //            return p1.compareTo(p2);
451 //        });
452 //        // Sort the timestamps
453 //        lastSignatureTimestamps.sort((p1, p2) -> {
454 //            return p1.compareTo(p2);
455 //        });
456 //
457 //        Iterator<Instant> iterFirst = firstSignatureTimestamps.iterator();
458 //        Iterator<Instant> iterLast = lastSignatureTimestamps.iterator();
459 //        System.out.println("First to Last Packet:");
460 //        while (iterFirst.hasNext() && iterLast.hasNext()) {
461 //            Instant firstInst = (Instant) iterFirst.next();
462 //            Instant lastInst = (Instant) iterLast.next();
463 //            Duration dur = Duration.between(firstInst, lastInst);
464 //            long duration = dur.toMillis();
465 //            // Check duration --- should be below 15 seconds
466 //            if (duration >= 0 && duration <= 15000) {
467 //                System.out.println(dur.toMillis());
468 //            } else if (duration > 15000) {
469 //                while (duration > 15000) { // that means we have to move to the next trigger
470 //                    firstInst = (Instant) iterFirst.next();
471 //                    dur = Duration.between(firstInst, lastInst);
472 //                    duration = dur.toMillis();
473 //                }
474 //                System.out.println(dur.toMillis());
475 //            } else { // below 0 / negative --- that means we have to move to the next signature
476 //                while (duration < 0) { // that means we have to move to the next trigger
477 //                    lastInst = (Instant) iterLast.next();
478 //                    dur = Duration.between(firstInst, lastInst);
479 //                    duration = dur.toMillis();
480 //                }
481 //                System.out.println(dur.toMillis());
482 //            }
483 //            if (duration > 8000) {
484 //                break;
485 //            }
486 //        }
487
488         // TODO: The above part is just for 15-second time sensitivity experiment
489         // TODO: The above part is just for 15-second time sensitivity experiment
490         // TODO: The above part is just for 15-second time sensitivity experiment
491
492
493
494
495         TriggerTimesFileReader ttfr = new TriggerTimesFileReader();
496         List<Instant> triggerTimes = ttfr.readTriggerTimes(triggerTimesFile, false);
497         // Tag each trigger with "ON" or "OFF", assuming that the first trigger is an "ON" and that they alternate.
498         List<UserAction> userActions = new ArrayList<>();
499         for (int i = 0; i < triggerTimes.size(); i++) {
500             userActions.add(new UserAction(i % 2 == 0 ? Type.TOGGLE_ON : Type.TOGGLE_OFF, triggerTimes.get(i)));
501         }
502         TriggerTrafficExtractor tte = new TriggerTrafficExtractor(inputPcapFile, triggerTimes, deviceIp);
503         final PcapDumper outputter = Pcaps.openDead(DataLinkType.EN10MB, 65536).dumpOpen(outputPcapFile);
504         DnsMap dnsMap = new DnsMap();
505         TcpReassembler tcpReassembler = new TcpReassembler();
506         TrafficLabeler trafficLabeler = new TrafficLabeler(userActions);
507         tte.performExtraction(pkt -> {
508             try {
509                 outputter.dump(pkt);
510             } catch (NotOpenException e) {
511                 e.printStackTrace();
512             }
513         }, dnsMap, tcpReassembler, trafficLabeler);
514         outputter.flush();
515         outputter.close();
516
517         if (tte.getPacketsIncludedCount() != trafficLabeler.getTotalPacketCount()) {
518             // Sanity/debug check
519             throw new AssertionError(String.format("mismatch between packet count in %s and %s",
520                     TriggerTrafficExtractor.class.getSimpleName(), TrafficLabeler.class.getSimpleName()));
521         }
522
523         // Extract all conversations present in the filtered trace.
524         List<Conversation> allConversations = tcpReassembler.getTcpConversations();
525         // Group conversations by hostname.
526         Map<String, List<Conversation>> convsByHostname = TcpConversationUtils.groupConversationsByHostname(allConversations, dnsMap);
527         System.out.println("Grouped conversations by hostname.");
528         // For each hostname, count the frequencies of packet lengths exchanged with that hostname.
529         final Map<String, Map<Integer, Integer>> pktLenFreqsByHostname = new HashMap<>();
530         convsByHostname.forEach((host, convs) -> pktLenFreqsByHostname.put(host, TcpConversationUtils.countPacketLengthFrequencies(convs)));
531         System.out.println("Counted frequencies of packet lengths exchanged with each hostname.");
532         // For each hostname, count the frequencies of packet sequences (i.e., count how many conversations exchange a
533         // sequence of packets of some specific lengths).
534         final Map<String, Map<String, Integer>> pktSeqFreqsByHostname = new HashMap<>();
535         convsByHostname.forEach((host, convs) -> pktSeqFreqsByHostname.put(host, TcpConversationUtils.countPacketSequenceFrequencies(convs)));
536         System.out.println("Counted frequencies of packet sequences exchanged with each hostname.");
537         // For each hostname, count frequencies of packet pairs exchanged with that hostname across all conversations
538         final Map<String, Map<String, Integer>> pktPairFreqsByHostname =
539                 TcpConversationUtils.countPacketPairFrequenciesByHostname(allConversations, dnsMap);
540         System.out.println("Counted frequencies of packet pairs per hostname");
541         // For each user action, reassemble the set of TCP connections occurring shortly after
542         final Map<UserAction, List<Conversation>> userActionToConversations = trafficLabeler.getLabeledReassembledTcpTraffic();
543         final Map<UserAction, Map<String, List<Conversation>>> userActionsToConvsByHostname = trafficLabeler.getLabeledReassembledTcpTraffic(dnsMap);
544         System.out.println("Reassembled TCP conversations occurring shortly after each user event");
545
546
547
548         /*
549          * NOTE: no need to generate these more complex on/off maps that also contain mappings from hostname and
550          * sequence identifiers as we do not care about hostnames and sequences during clustering.
551          * We can simply use the UserAction->List<Conversation> map to generate ON/OFF groupings of conversations.
552          */
553
554         // Contains all ON events: hostname -> sequence identifier -> list of conversations with that sequence
555         Map<String, Map<String, List<Conversation>>> ons = new HashMap<>();
556         // Contains all OFF events: hostname -> sequence identifier -> list of conversations with that sequence
557         Map<String, Map<String, List<Conversation>>> offs = new HashMap<>();
558         userActionsToConvsByHostname.forEach((ua, hostnameToConvs) -> {
559             Map<String, Map<String, List<Conversation>>> outer = ua.getType() == Type.TOGGLE_ON ? ons : offs;
560             hostnameToConvs.forEach((host, convs) -> {
561                 Map<String, List<Conversation>> seqsToConvs = TcpConversationUtils.
562                         groupConversationsByPacketSequence(convs, verbose);
563                 outer.merge(host, seqsToConvs, (oldMap, newMap) -> {
564                     newMap.forEach((sequence, cs) -> oldMap.merge(sequence, cs, (list1, list2) -> {
565                         list1.addAll(list2);
566                         return list1;
567                     }));
568                     return oldMap;
569                 });
570             });
571         });
572
573 //        System.out.println("==== ON ====");
574 //        // Print out all the pairs into a file for ON events
575 //        File fileOnEvents = new File(onPairsPath);
576 //        PrintWriter pwOn = null;
577 //        try {
578 //            pwOn = new PrintWriter(fileOnEvents);
579 //        } catch(Exception ex) {
580 //            ex.printStackTrace();
581 //        }
582 //        for(Map.Entry<String, Map<String, List<Conversation>>> entry : ons.entrySet()) {
583 //            Map<String, List<Conversation>> seqsToConvs = entry.getValue();
584 //            for(Map.Entry<String, List<Conversation>> entryConv : seqsToConvs.entrySet()) {
585 //                List<Conversation> listConv = entryConv.getValue();
586 //                // Just get the first Conversation because all Conversations in this group
587 //                // should have the same pairs of Application Data.
588 //                for(Conversation conv : listConv) {
589 //                    // Process only if it is a TLS packet
590 //                    if (conv.isTls()) {
591 //                        List<PcapPacketPair> tlsAppDataList = TcpConversationUtils.extractTlsAppDataPacketPairs(conv);
592 //                        for(PcapPacketPair pair: tlsAppDataList) {
593 //                            System.out.println(PrintUtils.toCsv(pair, dnsMap));
594 //                            pwOn.println(PrintUtils.toCsv(pair, dnsMap));
595 //                        }
596 //                    } else { // Non-TLS conversations
597 //                        List<PcapPacketPair> packetList = TcpConversationUtils.extractPacketPairs(conv);
598 //                        for(PcapPacketPair pair: packetList) {
599 //                            System.out.println(PrintUtils.toCsv(pair, dnsMap));
600 //                            pwOn.println(PrintUtils.toCsv(pair, dnsMap));
601 //                        }
602 //                    }
603 //                }
604 //            }
605 //        }
606 //        pwOn.close();
607 //
608 //        System.out.println("==== OFF ====");
609 //        // Print out all the pairs into a file for ON events
610 //        File fileOffEvents = new File(offPairsPath);
611 //        PrintWriter pwOff = null;
612 //        try {
613 //            pwOff = new PrintWriter(fileOffEvents);
614 //        } catch(Exception ex) {
615 //            ex.printStackTrace();
616 //        }
617 //        for(Map.Entry<String, Map<String, List<Conversation>>> entry : offs.entrySet()) {
618 //            Map<String, List<Conversation>> seqsToConvs = entry.getValue();
619 //            for(Map.Entry<String, List<Conversation>> entryConv : seqsToConvs.entrySet()) {
620 //                List<Conversation> listConv = entryConv.getValue();
621 //                // Just get the first Conversation because all Conversations in this group
622 //                // should have the same pairs of Application Data.
623 //                for(Conversation conv : listConv) {
624 //                    // Process only if it is a TLS packet
625 //                    if (conv.isTls()) {
626 //                        List<PcapPacketPair> tlsAppDataList = TcpConversationUtils.extractTlsAppDataPacketPairs(conv);
627 //                        for(PcapPacketPair pair: tlsAppDataList) {
628 //                            System.out.println(PrintUtils.toCsv(pair, dnsMap));
629 //                            pwOff.println(PrintUtils.toCsv(pair, dnsMap));
630 //                        }
631 //                    } else { // Non-TLS conversations
632 //                        List<PcapPacketPair> packetList = TcpConversationUtils.extractPacketPairs(conv);
633 //                        for (PcapPacketPair pair : packetList) {
634 //                            System.out.println(PrintUtils.toCsv(pair, dnsMap));
635 //                            pwOff.println(PrintUtils.toCsv(pair, dnsMap));
636 //                        }
637 //                    }
638 //                }
639 //            }
640 //        }
641 //        pwOff.close();
642
643
644         // ================================================ CLUSTERING ================================================
645         // Note: no need to use the more convoluted on/off maps; can simply use the UserAction->List<Conversation> map
646         // when don't care about hostnames and sequences (see comment earlier).
647         List<Conversation> onConversations = userActionToConversations.entrySet().stream().
648                 filter(e -> e.getKey().getType() == Type.TOGGLE_ON). // drop all OFF events from stream
649                 map(e -> e.getValue()). // no longer interested in the UserActions
650                 flatMap(List::stream). // flatten List<List<T>> to a List<T>
651                 collect(Collectors.toList());
652         List<Conversation> offConversations = userActionToConversations.entrySet().stream().
653                 filter(e -> e.getKey().getType() == Type.TOGGLE_OFF).
654                 map(e -> e.getValue()).
655                 flatMap(List::stream).
656                 collect(Collectors.toList());
657         //Collections.sort(onConversations, (c1, c2) -> c1.getPackets().)
658
659         List<PcapPacketPair> onPairs = onConversations.stream().
660                 map(c -> c.isTls() ? TcpConversationUtils.extractTlsAppDataPacketPairs(c) :
661                         TcpConversationUtils.extractPacketPairs(c)).
662                 flatMap(List::stream). // flatten List<List<>> to List<>
663                 collect(Collectors.toList());
664         List<PcapPacketPair> offPairs = offConversations.stream().
665                 map(c -> c.isTls() ? TcpConversationUtils.extractTlsAppDataPacketPairs(c) :
666                         TcpConversationUtils.extractPacketPairs(c)).
667                 flatMap(List::stream). // flatten List<List<>> to List<>
668                 collect(Collectors.toList());
669         // Note: need to update the DnsMap of all PcapPacketPairs if we want to use the IP/hostname-sensitive distance.
670         Stream.concat(Stream.of(onPairs), Stream.of(offPairs)).flatMap(List::stream).forEach(p -> p.setDnsMap(dnsMap));
671         // Perform clustering on conversation logged as part of all ON events.
672         double eps = 10.0;
673         int minPts = 45;
674         DBSCANClusterer<PcapPacketPair> onClusterer = new DBSCANClusterer<>(eps, minPts);
675         List<Cluster<PcapPacketPair>> onClusters = onClusterer.cluster(onPairs);
676         // Perform clustering on conversation logged as part of all OFF events.
677         DBSCANClusterer<PcapPacketPair> offClusterer = new DBSCANClusterer<>(eps, minPts);
678         List<Cluster<PcapPacketPair>> offClusters = offClusterer.cluster(offPairs);
679         // Sort the conversations as reference
680         List<Conversation> sortedAllConversation = TcpConversationUtils.sortConversationList(allConversations);
681         // Output clusters
682         System.out.println("========================================");
683         System.out.println("       Clustering results for ON        ");
684         System.out.println("       Number of clusters: " + onClusters.size());
685         int count = 0;
686         List<List<List<PcapPacket>>> ppListOfListReadOn = new ArrayList<>();
687         List<List<List<PcapPacket>>> ppListOfListListOn = new ArrayList<>();
688         List<List<List<PcapPacket>>> corePointRangeSignatureOn = new ArrayList<>();
689         for (Cluster<PcapPacketPair> c : onClusters) {
690             System.out.println(String.format("<<< Cluster #%02d (%03d points) >>>", ++count, c.getPoints().size()));
691             System.out.print(PrintUtils.toSummaryString(c));
692             if(c.getPoints().size() > 45 && c.getPoints().size() < 55) {
693                 // Print to file
694                 List<List<PcapPacket>> ppListOfList = PcapPacketUtils.clusterToListOfPcapPackets(c);
695                 // Check for overlaps and decide whether to do range-based or conservative checking
696                 corePointRangeSignatureOn.add(PcapPacketUtils.extractRangeCorePoints(ppListOfList, eps, minPts));
697                 ppListOfListListOn.add(ppListOfList);
698             }
699         }
700         System.out.println("========================================");
701         System.out.println("       Clustering results for OFF       ");
702         System.out.println("       Number of clusters: " + offClusters.size());
703         count = 0;
704         List<List<List<PcapPacket>>> ppListOfListReadOff = new ArrayList<>();
705         List<List<List<PcapPacket>>> ppListOfListListOff = new ArrayList<>();
706         List<List<List<PcapPacket>>> corePointRangeSignatureOff = new ArrayList<>();
707         for (Cluster<PcapPacketPair> c : offClusters) {
708             System.out.println(String.format("<<< Cluster #%03d (%06d points) >>>", ++count, c.getPoints().size()));
709             System.out.print(PrintUtils.toSummaryString(c));
710             if(c.getPoints().size() > 45 && c.getPoints().size() < 55) {
711                 // Print to file
712                 List<List<PcapPacket>> ppListOfList = PcapPacketUtils.clusterToListOfPcapPackets(c);
713                 // Check for overlaps and decide whether to do range-based or conservative checking
714                 corePointRangeSignatureOff.add(PcapPacketUtils.extractRangeCorePoints(ppListOfList, eps, minPts));
715                 ppListOfListListOff.add(ppListOfList);
716             }
717         }
718         // TODO: Merging test
719         ppListOfListListOn = PcapPacketUtils.mergeSignatures(ppListOfListListOn, sortedAllConversation);
720         // TODO: Need to remove sequence 550 567 for Blossom phone side since it is not a good signature (overlap)!
721 //        PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 1);
722         // TODO: Need to remove sequence 69 296 for Blossom device side since it is not a good signature (overlap)!
723 //        PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 2);
724         // TODO: Need to remove sequence number 2 for ST plug since it is not a good signature!
725         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 2);
726         // TODO: Need to remove sequence number 0 for Arlo Camera since it is not a good signature!
727 //        PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 0);
728         // TODO: Need to remove sequence number 0 for TP-Link plug since it is not a good signature!
729         // TODO: This sequence actually belongs to the local communication between the plug and the phone
730 //        PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOn, 0);
731         ppListOfListListOn = PcapPacketUtils.sortSignatures(ppListOfListListOn);
732
733         // TODO: Merging test
734         ppListOfListListOff = PcapPacketUtils.mergeSignatures(ppListOfListListOff, sortedAllConversation);
735         // TODO: Need to remove sequence 69 296 for Blossom device side since it is not a good signature (overlap)!
736 //        PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 3);
737         // TODO: Need to remove sequence number 1 for Nest Thermostat since it is not a good signature!
738         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 1);
739         // TODO: Need to remove sequence number 0 for Arlo Camera since it is not a good signature!
740 //        PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 1);
741         // TODO: Need to remove sequence number 2 for ST plug since it is not a good signature!
742         //PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 2);
743         // TODO: Need to remove sequence number 0 for TP-Link plug since it is not a good signature!
744         // TODO: This sequence actually belongs to the local communication between the plug and the phone
745 //        PcapPacketUtils.removeSequenceFromSignature(ppListOfListListOff, 0);
746         ppListOfListListOff = PcapPacketUtils.sortSignatures(ppListOfListListOff);
747
748         // Write the signatures into the screen
749         PcapPacketUtils.printSignatures(ppListOfListListOn);
750         PcapPacketUtils.printSignatures(ppListOfListListOff);
751
752         // Printing signatures into files
753         PrintUtils.serializeIntoFile("./onSignature.sig", ppListOfListListOn);
754         PrintUtils.serializeIntoFile("./offSignature.sig", ppListOfListListOff);
755         //ppListOfListReadOn = PrintUtils.deserializeFromFile("./onSignature.sig");
756         //ppListOfListReadOff = PrintUtils.deserializeFromFile("./offSignature.sig");
757
758         // Printing cluster analyses into files
759         PrintUtils.serializeIntoFile("./onClusters.cls", corePointRangeSignatureOn);
760         PrintUtils.serializeIntoFile("./offClusters.cls", corePointRangeSignatureOff);
761         System.out.println("========================================");
762
763         // ============================================================================================================
764
765 //        // TODO: This part is just for DBSCAN sensitivity experiment
766 //        // TODO: This part is just for DBSCAN sensitivity experiment
767 //        // TODO: This part is just for DBSCAN sensitivity experiment
768 //        // TODO: This part is just for DBSCAN sensitivity experiment
769 //        // TODO: This part is just for DBSCAN sensitivity experiment
770 //        List<Conversation> onConversations = userActionToConversations.entrySet().stream().
771 //                filter(e -> e.getKey().getType() == Type.TOGGLE_ON). // drop all OFF events from stream
772 //                map(e -> e.getValue()). // no longer interested in the UserActions
773 //                flatMap(List::stream). // flatten List<List<T>> to a List<T>
774 //                collect(Collectors.toList());
775 //        List<Conversation> offConversations = userActionToConversations.entrySet().stream().
776 //                filter(e -> e.getKey().getType() == Type.TOGGLE_OFF).
777 //                map(e -> e.getValue()).
778 //                flatMap(List::stream).
779 //                collect(Collectors.toList());
780 //        //Collections.sort(onConversations, (c1, c2) -> c1.getPackets().)
781 //
782 //        List<PcapPacketPair> onPairs = onConversations.stream().
783 //                map(c -> c.isTls() ? TcpConversationUtils.extractTlsAppDataPacketPairs(c) :
784 //                        TcpConversationUtils.extractPacketPairs(c)).
785 //                flatMap(List::stream). // flatten List<List<>> to List<>
786 //                collect(Collectors.toList());
787 //        List<PcapPacketPair> offPairs = offConversations.stream().
788 //                map(c -> c.isTls() ? TcpConversationUtils.extractTlsAppDataPacketPairs(c) :
789 //                        TcpConversationUtils.extractPacketPairs(c)).
790 //                flatMap(List::stream). // flatten List<List<>> to List<>
791 //                collect(Collectors.toList());
792 //        // Note: need to update the DnsMap of all PcapPacketPairs if we want to use the IP/hostname-sensitive distance.
793 //        Stream.concat(Stream.of(onPairs), Stream.of(offPairs)).flatMap(List::stream).forEach(p -> p.setDnsMap(dnsMap));
794 //
795 //        double eps = 10; // loop from eps 1-10
796 //        int minPts = 50; // loop from minPts 30-50
797 //        for(int epsCount = 7; epsCount <= eps; epsCount++) {
798 //            for(int minPtsCount = 30; minPtsCount <= minPts; minPtsCount++) {
799 //                System.out.println("Eps: " + epsCount + " --- minPts: " + minPtsCount);
800 //                DBSCANClusterer<PcapPacketPair> onClusterer = new DBSCANClusterer<>(epsCount, minPtsCount);
801 //                DBSCANClusterer<PcapPacketPair> offClusterer = new DBSCANClusterer<>(epsCount, minPtsCount);
802 //                List<Cluster<PcapPacketPair>> onClusters = onClusterer.cluster(onPairs);
803 //                List<Cluster<PcapPacketPair>> offClusters = offClusterer.cluster(offPairs);
804 //                // Sort the conversations as reference
805 //                List<Conversation> sortedAllConversation = TcpConversationUtils.sortConversationList(allConversations);
806 //                // Output clusters
807 //                System.out.println("========================================");
808 //                System.out.println("       Clustering results for ON        ");
809 //                System.out.println("       Number of clusters: " + onClusters.size());
810 //                int count = 0;
811 //                List<List<List<PcapPacket>>> ppListOfListListOn = new ArrayList<>();
812 //                for (Cluster<PcapPacketPair> c : onClusters) {
813 //                    System.out.println(String.format("<<< Cluster #%02d (%03d points) >>>", ++count, c.getPoints().size()));
814 ////                    System.out.print(PrintUtils.toSummaryString(c));
815 //                    if (c.getPoints().size() > 45 && c.getPoints().size() < 55) {
816 ////                        if(c.getPoints().size() > 25) {
817 //                        // Print to file
818 //                        List<List<PcapPacket>> ppListOfList = PcapPacketUtils.clusterToListOfPcapPackets(c);
819 //                        ppListOfListListOn.add(ppListOfList);
820 //                    }
821 //                }
822 //                PcapPacketUtils.printSignatures(ppListOfListListOn);
823 //
824 //                System.out.println("========================================");
825 //                System.out.println("       Clustering results for OFF       ");
826 //                System.out.println("       Number of clusters: " + offClusters.size());
827 //                count = 0;
828 //                List<List<List<PcapPacket>>> ppListOfListListOff = new ArrayList<>();
829 //                for (Cluster<PcapPacketPair> c : offClusters) {
830 //                    System.out.println(String.format("<<< Cluster #%03d (%06d points) >>>", ++count, c.getPoints().size()));
831 ////                    System.out.print(PrintUtils.toSummaryString(c));
832 //                    if (c.getPoints().size() > 45 && c.getPoints().size() < 55) {
833 //                        //if(c.getPoints().size() > 25) {
834 //                        // Print to file
835 //                        List<List<PcapPacket>> ppListOfList = PcapPacketUtils.clusterToListOfPcapPackets(c);
836 //                        ppListOfListListOff.add(ppListOfList);
837 //                    }
838 //                }
839 //                PcapPacketUtils.printSignatures(ppListOfListListOff);
840 //                System.out.println();
841 //                System.out.println();
842 //                System.out.println();
843 //                // ============================================================================================================
844 //            }
845 //        }
846
847
848 //        // ================================================================================================
849 //        // <<< Some work-in-progress/explorative code that extracts a "representative" sequence >>>
850 //        //
851 //        // Currently need to know relevant hostname in advance :(
852 //        String hostname = "events.tplinkra.com";
853 ////        String hostname = "rfe-us-west-1.dch.dlink.com";
854 //        // Conversations with 'hostname' for ON events.
855 //        List<Conversation> onsForHostname = new ArrayList<>();
856 //        // Conversations with 'hostname' for OFF events.
857 //        List<Conversation> offsForHostname = new ArrayList<>();
858 //        // "Unwrap" sequence groupings in ons/offs maps.
859 //        ons.get(hostname).forEach((k,v) -> onsForHostname.addAll(v));
860 //        offs.get(hostname).forEach((k,v) -> offsForHostname.addAll(v));
861 //
862 //
863 //        Map<String, List<Conversation>> onsForHostnameGroupedByTlsAppDataSequence = TcpConversationUtils.groupConversationsByTlsApplicationDataPacketSequence(onsForHostname);
864 //
865 //
866 //        // Extract representative sequence for ON and OFF by providing the list of conversations with
867 //        // 'hostname' observed for each event type (the training data).
868 //        SequenceExtraction seqExtraction = new SequenceExtraction();
869 ////        ExtractedSequence extractedSequenceForOn = seqExtraction.extract(onsForHostname);
870 ////        ExtractedSequence extractedSequenceForOff = seqExtraction.extract(offsForHostname);
871 //
872 //        ExtractedSequence extractedSequenceForOn = seqExtraction.extractByTlsAppData(onsForHostname);
873 //        ExtractedSequence extractedSequenceForOff = seqExtraction.extractByTlsAppData(offsForHostname);
874 //
875 //        // Let's check how many ONs align with OFFs and vice versa (that is, how many times an event is incorrectly
876 //        // labeled).
877 //        int onsLabeledAsOff = 0;
878 //        Integer[] representativeOnSeq = TcpConversationUtils.getPacketLengthSequence(extractedSequenceForOn.getRepresentativeSequence());
879 //        Integer[] representativeOffSeq = TcpConversationUtils.getPacketLengthSequence(extractedSequenceForOff.getRepresentativeSequence());
880 //        SequenceAlignment<Integer> seqAlg = seqExtraction.getAlignmentAlgorithm();
881 //        for (Conversation c : onsForHostname) {
882 //            Integer[] onSeq = TcpConversationUtils.getPacketLengthSequence(c);
883 //            if (seqAlg.calculateAlignment(representativeOffSeq, onSeq) <= extractedSequenceForOff.getMaxAlignmentCost()) {
884 //                onsLabeledAsOff++;
885 //            }
886 //        }
887 //        int offsLabeledAsOn = 0;
888 //        for (Conversation c : offsForHostname) {
889 //            Integer[] offSeq = TcpConversationUtils.getPacketLengthSequence(c);
890 //            if (seqAlg.calculateAlignment(representativeOnSeq, offSeq) <= extractedSequenceForOn.getMaxAlignmentCost()) {
891 //                offsLabeledAsOn++;
892 //            }
893 //        }
894 //        System.out.println("");
895 //        // ================================================================================================
896 //
897 //
898 //        // -------------------------------------------------------------------------------------------------------------
899 //        // -------------------------------------------------------------------------------------------------------------
900     }
901
902     /**
903      * Check if there is any overlap between the signature stored in this class and another signature.
904      * Conditions:
905      * 1) If the signatures do not have any range, then we need to do conservative checking (return true).
906      * 2) If the signatures have the same number of packets/packet lengths, then we check the range; if the
907      *    numbers of packets/packet lengths are different then we assume that there is no overlap.
908      * 3) If there is any range in the signatures, then we need to check for overlap.
909      * 4) If there is overlap for every packet/packet length, then we return false (range-based checking); otherwise,
910      *    true (conservative checking).
911      *
912      * @param signatures Multiple {@code List} of {@code List} of {@code List} of {@code PcapPacket} objects to be checked
913      *                       for overlaps.
914      * @return A boolean that is true if there is an overlap; false otherwise.
915      */
916     public boolean isConservativeChecking(List<List<List<PcapPacket>>>...signatures) {
917
918         // If none of the pairs/sequences is range-based then we go conservative
919         boolean isRange = false;
920         for(List<List<List<PcapPacket>>> signature : signatures) {
921             if (isRangeBased(signature)) {
922                 isRange = true;
923             }
924         }
925
926         return isRange;
927 //        if (!isRangeBased(signatureRanges) && !isRangeBased(otherSignatureRanges)) {
928 //            // Conservative checking when there is no range
929 //            return true;
930 //        } else if(signatureRanges.size() != otherSignatureRanges.size()) {
931 //            // The two signatures have different numbers of packets/packet lengths
932 //            return false;
933 //        } else {
934 //            // There is range; check if there is overlap
935 //            return checkOverlap(signatureRanges, otherSignatureRanges);
936 //        }
937     }
938
939     /*
940      * Check for overlap since we have range in at least one of the signatures.
941      * Overlap is only true when all ranges overlap. We need to check in order.
942      */
943 //    private boolean checkOverlap(List<List<List<PcapPacket>>> signatureRanges,
944 //                                 List<List<List<PcapPacket>>> otherSignatureRanges) {
945 //
946 //        for(List<List<PcapPacket>> listListPcapPacket : signatureRanges) {
947 //            // Lower bound of the range is in index 0
948 //            // Upper bound of the range is in index 1
949 //            int sequenceSetIndex = signatureRanges.indexOf(listListPcapPacket);
950 //            List<PcapPacket> minSequenceSignature = listListPcapPacket.get(0);
951 //            List<PcapPacket> maxSequenceSignature = listListPcapPacket.get(1);
952 //            for(PcapPacket pcapPacket : minSequenceSignature) {
953 //                // Get the lower and upper bounds of the current signature
954 //                int packetIndex = minSequenceSignature.indexOf(pcapPacket);
955 //                int lowerBound = pcapPacket.length();
956 //                int upperBound = maxSequenceSignature.get(packetIndex).length();
957 //                // Check for range overlap in the other signature!
958 //                // Check the packet/packet length at the same position
959 //                List<PcapPacket> minSequenceSignatureOther = otherSignatureRanges.get(sequenceSetIndex).get(0);
960 //                List<PcapPacket> maxSequenceSignatureOther = otherSignatureRanges.get(sequenceSetIndex).get(1);
961 //                int lowerBoundOther = minSequenceSignatureOther.get(packetIndex).length();
962 //                int upperBoundOther = maxSequenceSignatureOther.get(packetIndex).length();
963 //                if (!(lowerBoundOther <= lowerBound && lowerBound <= upperBoundOther) &&
964 //                    !(lowerBoundOther <= upperBound && upperBound <= upperBoundOther)) {
965 //                    return false;
966 //                }
967 //            }
968 //        }
969 //
970 //        return true;
971 //    }
972
973     /*
974      * Check and see if there is any range in the signatures
975      */
976     private boolean isRangeBased(List<List<List<PcapPacket>>> signatureRanges) {
977
978         for(List<List<PcapPacket>> listListPcapPacket : signatureRanges) {
979             // Lower bound of the range is in index 0
980             // Upper bound of the range is in index 1
981             List<PcapPacket> minSequence = listListPcapPacket.get(0);
982             List<PcapPacket> maxSequence = listListPcapPacket.get(1);
983             for(PcapPacket pcapPacket : minSequence) {
984                 int index = minSequence.indexOf(pcapPacket);
985                 if (pcapPacket.length() != maxSequence.get(index).length()) {
986                     // If there is any packet length that differs in the minSequence
987                     // and maxSequence, then it is range-based
988                     return true;
989                 }
990             }
991         }
992
993         return false;
994     }
995
996 //    /* Find the sequence with the minimum packet lengths.
997 //     * The second-layer list should contain the minimum sequence for element 0 and maximum sequence for element 1.
998 //     */
999 //    private List<List<List<PcapPacket>>> getSequenceRanges(List<List<List<PcapPacket>>> signature) {
1000 //
1001 //        // Start from the first index
1002 //        List<List<List<PcapPacket>>> rangeBasedSequence = new ArrayList<>();
1003 //        for (List<List<PcapPacket>> listListPcapPacket : signature) {
1004 //            List<List<PcapPacket>> minMaxSequence = new ArrayList<>();
1005 //            // Both searches start from index 0
1006 //            List<PcapPacket> minSequence = new ArrayList<>(listListPcapPacket.get(0));
1007 //            List<PcapPacket> maxSequence = new ArrayList<>(listListPcapPacket.get(0));
1008 //            for (List<PcapPacket> listPcapPacket : listListPcapPacket) {
1009 //                for (PcapPacket pcapPacket : listPcapPacket) {
1010 //                    int index = listPcapPacket.indexOf(pcapPacket);
1011 //                    // Set the new minimum if length at the index is minimum
1012 //                    if (pcapPacket.length() < minSequence.get(index).length()) {
1013 //                        minSequence.set(index, pcapPacket);
1014 //                    }
1015 //                    // Set the new maximum if length at the index is maximum
1016 //                    if (pcapPacket.length() > maxSequence.get(index).length()) {
1017 //                        maxSequence.set(index, pcapPacket);
1018 //                    }
1019 //                }
1020 //            }
1021 //            // minSequence as element 0 and maxSequence as element 1
1022 //            minMaxSequence.add(minSequence);
1023 //            minMaxSequence.add(maxSequence);
1024 //            rangeBasedSequence.add(minMaxSequence);
1025 //        }
1026 //
1027 //        return rangeBasedSequence;
1028 //    }
1029 }
1030
1031
1032 // TP-Link MAC 50:c7:bf:33:1f:09 and usually IP 192.168.1.159 (remember to verify per file)
1033 // frame.len >= 556 && frame.len <= 558 && ip.addr == 192.168.1.159