WEBVTT

1
00:00:00.040 --> 00:00:03.040
<v Speaker 1>Again, let's unpack this. You send us some fascinating material

2
00:00:03.160 --> 00:00:09.720
<v Speaker 1>on RFID technology, specifically focusing on its security and privacy aspects.

3
00:00:09.759 --> 00:00:12.679
<v Speaker 1>It's one of those truly invisible technologies that's kind of everywhere,

4
00:00:12.960 --> 00:00:15.640
<v Speaker 1>yet most of us don't even realize, you know, how

5
00:00:15.679 --> 00:00:18.440
<v Speaker 1>deeply it's woven into our daily lives. So for this

6
00:00:18.480 --> 00:00:20.640
<v Speaker 1>deep dive, our mission is to really cut through the

7
00:00:20.679 --> 00:00:24.879
<v Speaker 1>complexity and extract the most important insights from your sources.

8
00:00:24.480 --> 00:00:28.120
<v Speaker 2>Exactly, and RFID it's well, it's truly foundational element for

9
00:00:28.239 --> 00:00:31.120
<v Speaker 2>what people call ubiquitous computing, you know, where technology just

10
00:00:31.120 --> 00:00:34.320
<v Speaker 2>seamlessly integrates into our environment, often without us even noticing,

11
00:00:34.840 --> 00:00:38.719
<v Speaker 2>but it's widespread use. It also introduces this complex tension

12
00:00:38.759 --> 00:00:42.719
<v Speaker 2>between like incredible convenience and some pretty significant challenges for

13
00:00:42.719 --> 00:00:45.280
<v Speaker 2>security and privacy. So yeah, our deep type today will

14
00:00:45.280 --> 00:00:49.920
<v Speaker 2>focus on understanding those core elements and the critical vulnerabilities

15
00:00:49.920 --> 00:00:52.240
<v Speaker 2>detailed in the materials you've shared, right, So let's.

16
00:00:52.039 --> 00:00:54.520
<v Speaker 1>Start at the very beginning, then, What exactly is our

17
00:00:54.679 --> 00:00:57.640
<v Speaker 1>FID in its simplest form, It's like a shortcut, right

18
00:00:57.679 --> 00:01:03.560
<v Speaker 1>for electronically identifying, capturing data, controlling things, tracking, even inventoring items,

19
00:01:04.000 --> 00:01:07.200
<v Speaker 1>all using radio frequency communication. Think of it as a

20
00:01:07.239 --> 00:01:12.760
<v Speaker 1>wireless ID system basically, and its core components seem pretty straightforward.

21
00:01:12.799 --> 00:01:16.439
<v Speaker 1>First you have the RFID tag or transponder. This is

22
00:01:17.280 --> 00:01:20.879
<v Speaker 1>essentially a tiny microchip with a bit of data storage,

23
00:01:20.920 --> 00:01:23.079
<v Speaker 1>maybe some limited logic, and an antenna.

24
00:01:23.480 --> 00:01:26.200
<v Speaker 2>That's right. And then you've got the RFID reader sometimes

25
00:01:26.280 --> 00:01:29.040
<v Speaker 2>called a transceiver. This is the master unit, you could say.

26
00:01:29.159 --> 00:01:31.920
<v Speaker 2>It supplies energy to the tag. It triggers the communication.

27
00:01:32.359 --> 00:01:34.840
<v Speaker 2>Often it uses what's called the reader. Talk's first concept,

28
00:01:34.879 --> 00:01:36.799
<v Speaker 2>So the reader initiates everything, okay.

29
00:01:36.879 --> 00:01:39.719
<v Speaker 1>And finally there's the application system, which collects all that

30
00:01:39.799 --> 00:01:42.680
<v Speaker 1>data via the reader and then uses it through a

31
00:01:42.760 --> 00:01:44.239
<v Speaker 1>database for whatever it's designed for.

32
00:01:44.359 --> 00:01:46.840
<v Speaker 2>Seems like a lot packed into a small system. You

33
00:01:46.879 --> 00:01:50.000
<v Speaker 2>mentioned passive tags earlier, drawing power from the reader's signal.

34
00:01:50.239 --> 00:01:54.319
<v Speaker 2>How exactly does that work? Is it like induced electricity? Precisely,

35
00:01:54.359 --> 00:01:59.599
<v Speaker 2>it's like induced electricity. Passive tags are well, they're incredibly ingenious.

36
00:01:59.640 --> 00:02:02.879
<v Speaker 2>They literally harvest all their power for transmission and computation

37
00:02:03.159 --> 00:02:06.159
<v Speaker 2>directly from the rf signal the reader sends out. This

38
00:02:06.280 --> 00:02:10.879
<v Speaker 2>makes them extremely inexpensive and less powerful. Their read range

39
00:02:10.960 --> 00:02:14.560
<v Speaker 2>is typically just over a foot or so. Active tags

40
00:02:14.639 --> 00:02:17.080
<v Speaker 2>On the other hand, they carry their own batteries, so

41
00:02:17.120 --> 00:02:20.479
<v Speaker 2>they're more complex, more expensive, but they can communicate over

42
00:02:20.599 --> 00:02:23.919
<v Speaker 2>much longer distances up and over ten feet, and semi

43
00:02:24.000 --> 00:02:26.039
<v Speaker 2>passive tags are kind of a hybrid. They use a

44
00:02:26.039 --> 00:02:28.479
<v Speaker 2>battery just for the chip, not for sending the signal.

45
00:02:28.759 --> 00:02:31.599
<v Speaker 2>The cost of any tag usually boils down to the IC,

46
00:02:32.240 --> 00:02:35.639
<v Speaker 2>the integrated circuit, the antenna, and then assembly and testing.

47
00:02:36.000 --> 00:02:38.039
<v Speaker 1>Okay, that makes sense. And then when you look at

48
00:02:38.080 --> 00:02:41.639
<v Speaker 1>the sheer breadth of applications, Yeah, that's truly fascinating. You said,

49
00:02:41.680 --> 00:02:45.240
<v Speaker 1>tagging shipping containers is the largest business base globally.

50
00:02:45.400 --> 00:02:48.120
<v Speaker 2>Yeah, that's the biggest volume, surprisingly, but its.

51
00:02:48.039 --> 00:02:50.719
<v Speaker 1>Uses are everywhere, even if you don't see them, like

52
00:02:50.800 --> 00:02:55.199
<v Speaker 1>electronic payments, those RFID passports. Many of us carry tracking

53
00:02:55.240 --> 00:02:59.680
<v Speaker 1>office folders, tiny micro sensors inside things, intelligent labels on

54
00:02:59.680 --> 00:03:04.840
<v Speaker 1>product streamlining, ports, food production control, and yeah, animal identification.

55
00:03:04.919 --> 00:03:07.080
<v Speaker 1>It's undeniably pervasive.

56
00:03:06.680 --> 00:03:10.039
<v Speaker 2>Absolutely, And this widespread application means there isn't really a

57
00:03:10.080 --> 00:03:14.120
<v Speaker 2>single universal RFID system. It doesn't fit all needs. Systems

58
00:03:14.159 --> 00:03:17.800
<v Speaker 2>operate across different frequency bands low frequency or LF, high

59
00:03:17.840 --> 00:03:21.639
<v Speaker 2>frequency HF, and ultra high frequency UHF. The choice depends

60
00:03:21.800 --> 00:03:24.639
<v Speaker 2>entirely on the application. Right, you're considering factors like the

61
00:03:24.719 --> 00:03:28.080
<v Speaker 2>range you need, data requirements, and of course cost like

62
00:03:28.479 --> 00:03:31.520
<v Speaker 2>LF is common for animal ID or maybe car immobilizers,

63
00:03:31.599 --> 00:03:34.599
<v Speaker 2>shorter range stuff UHF is more for longer range tracking

64
00:03:34.639 --> 00:03:35.759
<v Speaker 2>like palettes in a warehouse.

65
00:03:35.879 --> 00:03:39.039
<v Speaker 1>Okay, given how deeply embedded this technology is becoming, we

66
00:03:39.039 --> 00:03:42.000
<v Speaker 1>absolutely have to talk about the crucial and maybe often

67
00:03:42.039 --> 00:03:45.400
<v Speaker 1>overlooked side of RFID, it's security and privacy.

68
00:03:45.000 --> 00:03:48.560
<v Speaker 2>Implications, right, And what's truly fascinating here is how the

69
00:03:48.680 --> 00:03:53.319
<v Speaker 2>very convenience of RFID, its ability to easily grab information

70
00:03:53.400 --> 00:03:56.719
<v Speaker 2>without needing line of sight, that simultaneously opens doors to

71
00:03:56.800 --> 00:04:01.759
<v Speaker 2>potential misuse and serious security and privacy risks. Information stored

72
00:04:01.759 --> 00:04:04.599
<v Speaker 2>in these tags, even hidden ones, can potentially be retrieved

73
00:04:04.639 --> 00:04:09.199
<v Speaker 2>by unauthorized maybe even hidden readers. Let's uh look at

74
00:04:09.240 --> 00:04:12.439
<v Speaker 2>some real world examples that really highlight this tension. Okay,

75
00:04:12.560 --> 00:04:16.079
<v Speaker 2>take financial credentials first. There was talk about embedding RFID

76
00:04:16.240 --> 00:04:19.920
<v Speaker 2>into euro banknotes to fight forgery. Sounds good, right, helps

77
00:04:19.959 --> 00:04:23.399
<v Speaker 2>fight crime, but it immediately raised concerns about potentially tracking

78
00:04:23.439 --> 00:04:25.399
<v Speaker 2>people's spending habits on a massive scale.

79
00:04:25.480 --> 00:04:26.319
<v Speaker 1>Yeah, I can see that.

80
00:04:26.560 --> 00:04:31.800
<v Speaker 2>And more concretely, contactless credit cards Visa master card Amex

81
00:04:31.920 --> 00:04:34.879
<v Speaker 2>designed for faster checkout, Well, they were found to be vulnerable.

82
00:04:35.199 --> 00:04:38.199
<v Speaker 2>Researchers showed they could skim these cards with unauthorized readers,

83
00:04:38.279 --> 00:04:41.720
<v Speaker 2>They could eavesdrop on legitimate sessions, even clone.

84
00:04:41.319 --> 00:04:42.920
<v Speaker 1>Cards clone them. Wow.

85
00:04:43.160 --> 00:04:45.319
<v Speaker 2>Yeah, And there was even this thing called a relay

86
00:04:45.360 --> 00:04:48.720
<v Speaker 2>attack that could basically place a fraudulent transaction onto a

87
00:04:48.800 --> 00:04:53.839
<v Speaker 2>nearby innocent credit card by redirecting the communication. The core

88
00:04:53.920 --> 00:04:57.480
<v Speaker 2>insight here really is that the touchless convenience, the thing

89
00:04:57.560 --> 00:05:01.439
<v Speaker 2>designed for speed, it inherently expands the attack surface. You

90
00:05:01.519 --> 00:05:04.519
<v Speaker 2>have to rethink how you secure sensitive data that can

91
00:05:04.560 --> 00:05:05.720
<v Speaker 2>be accessed remotely.

92
00:05:06.120 --> 00:05:09.399
<v Speaker 1>That's pretty startling for something meant to make life easier.

93
00:05:09.600 --> 00:05:11.759
<v Speaker 1>What about other areas like medicine You mentioned.

94
00:05:11.480 --> 00:05:15.079
<v Speaker 2>That absolutely, Then you have pharmaceutical products. The FDA considered

95
00:05:15.160 --> 00:05:18.720
<v Speaker 2>using RFID to combat the huge problem of counterfeit and

96
00:05:18.759 --> 00:05:23.120
<v Speaker 2>adulterated drugs billions of dollars lost. Now this has clear

97
00:05:23.199 --> 00:05:26.720
<v Speaker 2>benefits for public safety, supply chain integrity, great stuff, but

98
00:05:26.720 --> 00:05:30.360
<v Speaker 2>the privacy concern popped up immediately. Imagine scanning someone carrying

99
00:05:30.439 --> 00:05:34.199
<v Speaker 2>medicine and that scan inadvertently reveals their medical conditions. Maybe

100
00:05:34.199 --> 00:05:37.160
<v Speaker 2>to an insurance company or an employer without their consent.

101
00:05:37.279 --> 00:05:38.920
<v Speaker 1>Hmmm, yeah, that's a big concern.

102
00:05:39.240 --> 00:05:43.240
<v Speaker 2>And for personal identification and access control. Like RFID key cards,

103
00:05:43.600 --> 00:05:47.000
<v Speaker 2>they offer advantages over old keys sure, harder to copy,

104
00:05:47.079 --> 00:05:50.199
<v Speaker 2>easier to disable if lost, but security is still a

105
00:05:50.240 --> 00:05:54.199
<v Speaker 2>significant concern, especially when valuable assets are protected. It creates

106
00:05:54.199 --> 00:05:57.920
<v Speaker 2>a strong incentive for counterfeiting. Even our pathports aren't immune.

107
00:05:58.279 --> 00:06:02.639
<v Speaker 2>The ICAO, that's the International Civil Aviation Organization, they mandated

108
00:06:02.639 --> 00:06:06.160
<v Speaker 2>electronic passports for stronger authentication. Yet it's been shown that

109
00:06:06.199 --> 00:06:10.000
<v Speaker 2>these e passports can be well clandestinely stand and tracked.

110
00:06:10.399 --> 00:06:13.439
<v Speaker 2>There was a Dutch prototype RFID passport. Even though it

111
00:06:13.480 --> 00:06:16.079
<v Speaker 2>was weakly encrypted, it was cracked in just two hours.

112
00:06:16.120 --> 00:06:19.240
<v Speaker 2>Two hours scersly, we got two hours yielded the plaintext

113
00:06:19.240 --> 00:06:20.399
<v Speaker 2>info needed for cloning.

114
00:06:20.519 --> 00:06:23.639
<v Speaker 1>Okay, So if our FID is everywhere and it seems

115
00:06:24.399 --> 00:06:27.680
<v Speaker 1>pretty vulnerable in some cases, what are the bad actors

116
00:06:27.720 --> 00:06:30.079
<v Speaker 1>actually trying to do? What are we trying to defend

117
00:06:30.079 --> 00:06:32.199
<v Speaker 1>against when we talk about securing these systems?

118
00:06:32.360 --> 00:06:36.439
<v Speaker 2>Exactly good question. To really secure an RFID system, first

119
00:06:36.560 --> 00:06:38.759
<v Speaker 2>you need to understand the types of attacks you're facing,

120
00:06:39.079 --> 00:06:42.160
<v Speaker 2>and then what do secure even mean. In this context,

121
00:06:42.279 --> 00:06:46.800
<v Speaker 2>we can categorize attacks broadly. First, you've got hardware integrity attacks,

122
00:06:46.959 --> 00:06:51.319
<v Speaker 2>things like tag cloning, messing with the memory, reproducing tag items,

123
00:06:51.399 --> 00:06:55.279
<v Speaker 2>or even just physically damaging the tag. These are very threatening,

124
00:06:55.480 --> 00:06:58.839
<v Speaker 2>but also usually expensive. They require specialized.

125
00:06:58.279 --> 00:07:00.759
<v Speaker 1>Hardware, so that's like physically getting your ends on the

126
00:07:00.839 --> 00:07:02.279
<v Speaker 1>chip and messing with it precisely.

127
00:07:02.959 --> 00:07:06.399
<v Speaker 2>More common and often easier to pull off, are software attacks.

128
00:07:06.879 --> 00:07:09.519
<v Speaker 2>These are typically carried out by subverted readers. Think of

129
00:07:09.560 --> 00:07:12.519
<v Speaker 2>a legit reader that's been compromise, maybe hacked. And then

130
00:07:12.600 --> 00:07:14.800
<v Speaker 2>a lot of the common computer network attacks we already

131
00:07:14.839 --> 00:07:18.879
<v Speaker 2>know about, eavesdropping, impersonation, denial of service, or DOSS, they

132
00:07:18.920 --> 00:07:22.279
<v Speaker 2>translate pretty directly to RFID systems. A critical point here

133
00:07:22.319 --> 00:07:26.160
<v Speaker 2>is that an unsecured internal RFID system might simplify tag design,

134
00:07:26.240 --> 00:07:28.959
<v Speaker 2>lower costs. Sure, but if you then deploy that system

135
00:07:28.959 --> 00:07:32.319
<v Speaker 2>in an open environment, well you get serious privacy problems

136
00:07:32.360 --> 00:07:36.079
<v Speaker 2>like broadcasting unique tag IDs and plaintext. Anyone nearby could

137
00:07:36.079 --> 00:07:36.439
<v Speaker 2>track you.

138
00:07:36.720 --> 00:07:40.360
<v Speaker 1>That's a terrifying thought, just everyday items broadcasting your ID.

139
00:07:40.839 --> 00:07:43.079
<v Speaker 1>So how do we counteract that? What are the critical

140
00:07:43.120 --> 00:07:45.920
<v Speaker 1>security requirements we need for these RFID systems?

141
00:07:46.079 --> 00:07:48.759
<v Speaker 2>Right? To counter these threats, we need a robust set

142
00:07:48.800 --> 00:07:53.120
<v Speaker 2>of security requirements. Integrity is absolutely key, making sure tag

143
00:07:53.240 --> 00:07:56.759
<v Speaker 2>data is modified only by authorized readers, or at least

144
00:07:56.879 --> 00:07:58.920
<v Speaker 2>that unauthorized changes get noticed.

145
00:07:59.079 --> 00:08:01.519
<v Speaker 1>So it's about keeping the data true or at least

146
00:08:01.519 --> 00:08:03.639
<v Speaker 1>knowing if someone's tampered with it exactly.

147
00:08:04.160 --> 00:08:09.600
<v Speaker 2>Then there's confidentiality preventing unauthorized disclosure of information. Pretty straightforward.

148
00:08:09.959 --> 00:08:14.879
<v Speaker 2>Indistinguishability is crucial for privacy, preventing an adversary from distinguishing

149
00:08:14.959 --> 00:08:18.839
<v Speaker 2>or tracking tag identities or their owners remotely, even just

150
00:08:18.920 --> 00:08:21.199
<v Speaker 2>by their radio frequency signature.

151
00:08:20.759 --> 00:08:22.319
<v Speaker 1>So they can't even tell if it's the same tag

152
00:08:22.360 --> 00:08:23.079
<v Speaker 1>they saw before.

153
00:08:23.199 --> 00:08:26.800
<v Speaker 2>Ideally, yes, then forward security ensures that even if someone

154
00:08:26.839 --> 00:08:30.160
<v Speaker 2>compromises the current data on a tag, it's past history,

155
00:08:30.199 --> 00:08:33.759
<v Speaker 2>like previous reads or writes, remains untraceable. We also need

156
00:08:33.840 --> 00:08:37.159
<v Speaker 2>resistance against replay attacks, so an adversary can't just record

157
00:08:37.240 --> 00:08:40.200
<v Speaker 2>valid communication and reuse it later to impersonate a tag.

158
00:08:40.559 --> 00:08:43.200
<v Speaker 1>In authentication, making sure the right reader is talking to

159
00:08:43.240 --> 00:08:44.919
<v Speaker 1>the right tag and vice versa.

160
00:08:45.120 --> 00:08:48.440
<v Speaker 2>Yes, authentication is a big challenge, making sure tags only

161
00:08:48.480 --> 00:08:52.320
<v Speaker 2>reveal their identities to authorized readers and readers only authenticate

162
00:08:52.360 --> 00:08:55.679
<v Speaker 2>if they somehow know the tag's secret, and in complex

163
00:08:55.720 --> 00:08:59.919
<v Speaker 2>scenarios like say pharmaceutical supply chains, multi party and multi

164
00:09:00.159 --> 00:09:04.120
<v Speaker 2>level trust access is vital. Defining precisely who can access

165
00:09:04.159 --> 00:09:07.559
<v Speaker 2>which portion of the tag data read write a pend

166
00:09:07.600 --> 00:09:08.960
<v Speaker 2>that becomes really important.

167
00:09:09.240 --> 00:09:12.919
<v Speaker 1>That's a really comprehensive list of requirements. But playing devil

168
00:09:12.960 --> 00:09:15.440
<v Speaker 1>to advocate for a second, for something as cheap and

169
00:09:15.480 --> 00:09:18.720
<v Speaker 1>disposable as many RFID tags are meant to be, How

170
00:09:18.759 --> 00:09:21.159
<v Speaker 1>realistic is it to achieve all of that, especially when

171
00:09:21.200 --> 00:09:23.639
<v Speaker 1>you factor in what did you call it? Adversary modeling.

172
00:09:23.799 --> 00:09:26.159
<v Speaker 2>That's an excellent point, and it brings us to a

173
00:09:26.200 --> 00:09:31.440
<v Speaker 2>final crucial step and securitied sign adversary modeling. We absolutely

174
00:09:31.600 --> 00:09:35.320
<v Speaker 2>must understand the adversaries likely resources, their money, their time,

175
00:09:35.360 --> 00:09:38.480
<v Speaker 2>their computing power. We can't just assume they have unlimited resources.

176
00:09:38.480 --> 00:09:42.000
<v Speaker 2>That's not realistic. This helps us design.

177
00:09:41.840 --> 00:09:45.320
<v Speaker 1>Right, But isn't that a bit blunt? If you kill

178
00:09:45.360 --> 00:09:48.480
<v Speaker 1>the tag on your new gadget, you lose any potential

179
00:09:48.519 --> 00:09:51.519
<v Speaker 1>smart features that might have had down the line. Seems

180
00:09:51.559 --> 00:09:54.759
<v Speaker 1>like a limitation for that ubiquitous computing future, doesn't it?

181
00:09:54.759 --> 00:09:58.039
<v Speaker 2>It absolutely can be, yes, So an alternative is putting

182
00:09:58.120 --> 00:10:01.360
<v Speaker 2>tags to sleep, allowing them to be woken up later,

183
00:10:01.480 --> 00:10:04.360
<v Speaker 2>maybe in a safe environment, though that brings up password

184
00:10:04.360 --> 00:10:08.519
<v Speaker 2>management issues for users. Another pretty clever idea is blocker tags.

185
00:10:08.759 --> 00:10:11.840
<v Speaker 2>These are devices that basically jam the signals from selected

186
00:10:11.879 --> 00:10:14.679
<v Speaker 2>tags nearby, so you, the user, gets some control over

187
00:10:14.720 --> 00:10:17.360
<v Speaker 2>which of your tags respond if someone tries to scan them,

188
00:10:17.399 --> 00:10:20.159
<v Speaker 2>like a personal localized Faraday cage for your stuff.

189
00:10:20.360 --> 00:10:21.200
<v Speaker 1>Huh interesting.

190
00:10:21.519 --> 00:10:24.440
<v Speaker 2>Then you have things like actual physical switches on some tags,

191
00:10:24.440 --> 00:10:28.279
<v Speaker 2>maybe disconnecting the antenna or using special memory like electrically

192
00:10:28.279 --> 00:10:31.200
<v Speaker 2>erasable ROM or even magnetic bits to turn a tag

193
00:10:31.240 --> 00:10:35.000
<v Speaker 2>on or off. There are also light enabled switches tags

194
00:10:35.000 --> 00:10:37.679
<v Speaker 2>that can be deactivated by a light source, useful for

195
00:10:37.720 --> 00:10:41.279
<v Speaker 2>things like banknotes, maybe to prevent tracking, though an attacker

196
00:10:41.279 --> 00:10:43.919
<v Speaker 2>would need to point a light right at it. And finally,

197
00:10:44.000 --> 00:10:48.440
<v Speaker 2>there's time delay functionality. The tag deliberately slows down sending

198
00:10:48.480 --> 00:10:52.480
<v Speaker 2>sensitive data in say an unprotected environment, making scanning take longer,

199
00:10:52.759 --> 00:10:56.759
<v Speaker 2>but it responds instantly in protected settings. Okay. Now moving

200
00:10:56.799 --> 00:11:00.480
<v Speaker 2>over to cryptographic protocols and as young principles, A common

201
00:11:00.519 --> 00:11:04.000
<v Speaker 2>strategy is to frequently change the RF identifier, use pseudonyms,

202
00:11:04.080 --> 00:11:07.399
<v Speaker 2>so malicious parties can't easily trace a specific tag over time.

203
00:11:07.799 --> 00:11:11.240
<v Speaker 2>Password protected right access is already implemented in some standard tags,

204
00:11:11.279 --> 00:11:14.600
<v Speaker 2>like EPC tags, but it's still potentially vulnerable to eavesdropping

205
00:11:14.679 --> 00:11:17.759
<v Speaker 2>if the whole session is intercepted. Hash based protocols like

206
00:11:17.799 --> 00:11:20.480
<v Speaker 2>the hashlock scheme we're developed. The tag responds with a

207
00:11:20.480 --> 00:11:23.240
<v Speaker 2>hash of its secret key, like a digital fingerprint. A

208
00:11:23.320 --> 00:11:26.080
<v Speaker 2>randomized version sends a nonce, which is just a number

209
00:11:26.159 --> 00:11:29.600
<v Speaker 2>used once of random value, and then a hash combining

210
00:11:29.679 --> 00:11:32.559
<v Speaker 2>the ID and the noce. Now, this helps against tracking,

211
00:11:32.679 --> 00:11:35.000
<v Speaker 2>but it forces the back end database to do a

212
00:11:35.000 --> 00:11:37.159
<v Speaker 2>lot of searching to find the matching tag, and it

213
00:11:37.159 --> 00:11:40.799
<v Speaker 2>can still be vulnerable to impersonation or spoofing, and doesn't

214
00:11:40.840 --> 00:11:44.279
<v Speaker 2>really guarantee forward security if the tag itself gets compromised.

215
00:11:45.039 --> 00:11:48.039
<v Speaker 2>The ocubo scheme uses something called a hash chain to

216
00:11:48.120 --> 00:11:51.200
<v Speaker 2>constantly renew the tag's secret information. Think of it like

217
00:11:51.240 --> 00:11:54.159
<v Speaker 2>a sequence where each new secret depends on the last one,

218
00:11:54.320 --> 00:11:57.240
<v Speaker 2>constantly updating that digital fingerprint, making it harder to track

219
00:11:57.279 --> 00:11:57.759
<v Speaker 2>over time.

220
00:11:57.960 --> 00:12:01.360
<v Speaker 1>Okay, so we're layering these cryptographic complexes onto these tiny chips,

221
00:12:01.799 --> 00:12:05.320
<v Speaker 1>but the real tests often comes against real world adversaries, right,

222
00:12:05.840 --> 00:12:09.639
<v Speaker 1>which leads us to that pretty fascinating story about digital

223
00:12:09.639 --> 00:12:11.440
<v Speaker 1>signature transponders DSTs.

224
00:12:11.600 --> 00:12:15.080
<v Speaker 2>Indeed, DSTs are used in things like car and mobilizers

225
00:12:15.120 --> 00:12:19.279
<v Speaker 2>aiming for higher security using challenge response authentication. But the

226
00:12:19.320 --> 00:12:22.919
<v Speaker 2>proprietary DST forty algorithm it used a forty bit encryption

227
00:12:23.039 --> 00:12:26.240
<v Speaker 2>key was famously reverse engineered by a team at Johns

228
00:12:26.279 --> 00:12:29.240
<v Speaker 2>Hopkins University back in two thousand and four. They used

229
00:12:29.240 --> 00:12:32.159
<v Speaker 2>what they called a black box method, basically just collecting

230
00:12:32.200 --> 00:12:36.279
<v Speaker 2>pairs of keys, challenges and responses by programming their own DST.

231
00:12:35.919 --> 00:12:37.360
<v Speaker 1>Devices they just listened in.

232
00:12:37.600 --> 00:12:40.480
<v Speaker 2>Essentially, they generated lots of data, and then they even

233
00:12:40.519 --> 00:12:44.519
<v Speaker 2>built a hardware key cracker using FPGA's sixteen boards that

234
00:12:44.600 --> 00:12:46.519
<v Speaker 2>could crack a key in less than an hour.

235
00:12:46.480 --> 00:12:47.720
<v Speaker 1>Less than an hour. Wow.

236
00:12:48.039 --> 00:12:51.360
<v Speaker 2>Yeah. This showed that tag cloning and stealing data was

237
00:12:51.440 --> 00:12:55.879
<v Speaker 2>possible even from supposedly secure car systems. The huge lesson

238
00:12:55.960 --> 00:13:00.000
<v Speaker 2>learned there is critical You have to review security requirements constantly,

239
00:13:00.200 --> 00:13:04.320
<v Speaker 2>maybe annually, because computing power increases Moore's law and it

240
00:13:04.440 --> 00:13:07.039
<v Speaker 2>just erodes the strength of fixed length keys over time.

241
00:13:07.639 --> 00:13:11.600
<v Speaker 2>What's secure today might be breakable tomorrow for even stronger security.

242
00:13:11.639 --> 00:13:16.000
<v Speaker 2>People talk about public ecryptography PKC, like elliptic curve cryptography

243
00:13:16.080 --> 00:13:19.960
<v Speaker 2>or ECC, it offers really robust security, but it's typically

244
00:13:20.039 --> 00:13:23.639
<v Speaker 2>been seen as too computationally heavy, too expensive for these

245
00:13:23.679 --> 00:13:26.559
<v Speaker 2>low cost tags. However, research is exploring ways to make

246
00:13:26.559 --> 00:13:29.000
<v Speaker 2>it feasible. For instance, there's a one hundred and thirty

247
00:13:29.039 --> 00:13:32.399
<v Speaker 2>one bit ECC implementation that offers security comparable to the

248
00:13:32.440 --> 00:13:35.679
<v Speaker 2>old DEES algorithm from nineteen eighty two, which would still

249
00:13:35.679 --> 00:13:38.320
<v Speaker 2>cost billions of dollar days to break today. They use

250
00:13:38.399 --> 00:13:42.200
<v Speaker 2>clever math tricks like using projective coordinates to avoid field inversion,

251
00:13:42.480 --> 00:13:45.639
<v Speaker 2>making strong crypto possible even on tiny constrained tags.

252
00:13:45.879 --> 00:13:47.679
<v Speaker 1>So the math is getting smarter to fit.

253
00:13:47.559 --> 00:13:52.399
<v Speaker 2>The hardware exactly. And finally, there are lightweight symmetric algorithms

254
00:13:52.600 --> 00:13:57.240
<v Speaker 2>things like as tha xtea hash functions like SAHA one

255
00:13:57.320 --> 00:13:59.679
<v Speaker 2>or two fifty six MD five, even newer ones like

256
00:13:59.679 --> 00:14:04.120
<v Speaker 2>grain truvium. They're designed specifically for passive RFID tags. The

257
00:14:04.120 --> 00:14:07.159
<v Speaker 2>big challenges are super low power consumption and a really

258
00:14:07.159 --> 00:14:10.360
<v Speaker 2>small chip area. For example, as one twenty eight, a

259
00:14:10.399 --> 00:14:13.600
<v Speaker 2>strong standard can be implemented in a remarkably small area

260
00:14:13.679 --> 00:14:16.879
<v Speaker 2>around three threersy five hundred gait equivalents they call it.

261
00:14:16.879 --> 00:14:19.480
<v Speaker 2>It uses tiny amounts of power like three micro amps

262
00:14:19.480 --> 00:14:22.639
<v Speaker 2>at one hundred killohertz and finishes encrypting super fast in

263
00:14:22.720 --> 00:14:24.320
<v Speaker 2>about thousand clock cycles.

264
00:14:24.399 --> 00:14:25.480
<v Speaker 1>That's incredibly efficient.

265
00:14:25.679 --> 00:14:28.960
<v Speaker 2>It is. Designers use clever tricks like clock gating, putting

266
00:14:28.960 --> 00:14:30.720
<v Speaker 2>parts of the chip to sleep when not needed to

267
00:14:30.759 --> 00:14:34.320
<v Speaker 2>minimize power draw AES in particular is often favored for

268
00:14:34.360 --> 00:14:36.960
<v Speaker 2>symmetric crypto and RFID because it hits a really good

269
00:14:36.960 --> 00:14:39.679
<v Speaker 2>balance between strong security and resource efficiency.

270
00:14:40.240 --> 00:14:43.720
<v Speaker 1>Okay, now, with all these different security measures potentially in place,

271
00:14:43.759 --> 00:14:46.480
<v Speaker 1>the big question becomes how do they scale? How do

272
00:14:46.559 --> 00:14:48.759
<v Speaker 1>they hold up When we're talking about not just thousands,

273
00:14:48.759 --> 00:14:52.879
<v Speaker 1>but millions, maybe billions of tags, Because that's the ubiquitous

274
00:14:52.919 --> 00:14:54.120
<v Speaker 1>future people talk about.

275
00:14:53.919 --> 00:14:57.960
<v Speaker 2>Right, Scalability is absolutely a critical challenge. The sheer number

276
00:14:58.000 --> 00:15:01.679
<v Speaker 2>of RFID elements is growing in Daly fast. Think about

277
00:15:01.679 --> 00:15:04.519
<v Speaker 2>the border control example that was in the sources. If

278
00:15:04.559 --> 00:15:07.240
<v Speaker 2>every car has an RFID tag and all the border

279
00:15:07.240 --> 00:15:11.000
<v Speaker 2>readers connect back to one single centralized database holding every

280
00:15:11.039 --> 00:15:14.279
<v Speaker 2>possible ID, well, the delays would just grow linearly with

281
00:15:14.320 --> 00:15:16.840
<v Speaker 2>the number of cars. It just doesn't scale, wouldn't work practically,

282
00:15:17.360 --> 00:15:21.000
<v Speaker 2>But a distributed approach makes more sense. You organize the readers,

283
00:15:21.000 --> 00:15:23.679
<v Speaker 2>maybe in an adjacency graph, where each reader or local

284
00:15:23.759 --> 00:15:26.440
<v Speaker 2>database only needs to store the IDs likely to be

285
00:15:26.559 --> 00:15:30.759
<v Speaker 2>in its immediate influence area. This drastically reduces the number

286
00:15:30.799 --> 00:15:33.039
<v Speaker 2>of ID that each database needs to handle and really

287
00:15:33.080 --> 00:15:34.279
<v Speaker 2>improves scalability.

288
00:15:34.480 --> 00:15:37.200
<v Speaker 1>Right, So, instead of one massive phone book, it's more

289
00:15:37.240 --> 00:15:40.039
<v Speaker 1>like a network of smaller local address books that talk to.

290
00:15:40.039 --> 00:15:43.639
<v Speaker 2>Each other precisely. And to manage this effectively, an information

291
00:15:43.799 --> 00:15:47.320
<v Speaker 2>sharing protocol suite was proposed for the readers. It involves

292
00:15:47.320 --> 00:15:51.720
<v Speaker 2>messages like tag arrival, tag roaming, tag departure. This helps

293
00:15:51.759 --> 00:15:54.679
<v Speaker 2>manage the information flow and keeps all the local caches

294
00:15:54.759 --> 00:15:58.519
<v Speaker 2>reasonably up to date, making the whole distributed system scalable.

295
00:15:58.200 --> 00:16:01.559
<v Speaker 1>Okay, and looking for their ahead out. How does all

296
00:16:01.600 --> 00:16:05.240
<v Speaker 1>this tech and its security challenges fit into that bigger picture,

297
00:16:05.320 --> 00:16:08.519
<v Speaker 1>the vision of ubiquitous computing in smart spaces.

298
00:16:08.759 --> 00:16:12.200
<v Speaker 2>Yeah, this brings us to the emergence of mobile RFID services.

299
00:16:12.519 --> 00:16:15.919
<v Speaker 2>Think about integrating RFID reader chips directly into our mobile

300
00:16:15.919 --> 00:16:18.799
<v Speaker 2>phones or other terminals. This would let us read tags

301
00:16:18.840 --> 00:16:22.399
<v Speaker 2>on everyday objects. Then we could perform code resolution kind

302
00:16:22.399 --> 00:16:25.000
<v Speaker 2>of like DNS for websites. But for objects using something

303
00:16:25.039 --> 00:16:28.440
<v Speaker 2>called an ODS server Object Directory service and then access

304
00:16:28.480 --> 00:16:32.720
<v Speaker 2>information about that object via an OIS an Object Information service.

305
00:16:33.240 --> 00:16:37.480
<v Speaker 2>And crucially alongside this, an RFID Privacy Management System or

306
00:16:37.600 --> 00:16:40.200
<v Speaker 2>RPS is being developed. This is a system where we,

307
00:16:40.279 --> 00:16:43.600
<v Speaker 2>the tag owners, can define our own privacy policies, often

308
00:16:43.679 --> 00:16:46.879
<v Speaker 2>using a format like XML. These owner defined policies create

309
00:16:46.919 --> 00:16:49.799
<v Speaker 2>profiles that tell different service providers exactly who gets to

310
00:16:49.840 --> 00:16:52.960
<v Speaker 2>access what information about our tagged items. It aims to

311
00:16:52.960 --> 00:16:56.000
<v Speaker 2>put control back in the user's hands, which is vital.

312
00:16:56.120 --> 00:16:58.279
<v Speaker 1>So I get to decide who sees the data from

313
00:16:58.320 --> 00:17:00.759
<v Speaker 1>the tag on my says coffee mug.

314
00:17:00.919 --> 00:17:04.240
<v Speaker 2>That's the goal. Yes. And finally there's this really interesting

315
00:17:04.279 --> 00:17:08.880
<v Speaker 2>concept of RFID based touch for intuitive user interaction, especially

316
00:17:08.920 --> 00:17:13.640
<v Speaker 2>with smart space security. This often uses near field communication NFC,

317
00:17:13.759 --> 00:17:17.319
<v Speaker 2>which is basically a very short range form of RFID.

318
00:17:18.000 --> 00:17:19.799
<v Speaker 2>The goal here is to make things like smart home

319
00:17:19.839 --> 00:17:24.839
<v Speaker 2>security super easy, even for non experts, using simple touch gestures.

320
00:17:25.160 --> 00:17:27.160
<v Speaker 2>For example, you could set up a new smart device,

321
00:17:27.279 --> 00:17:29.839
<v Speaker 2>bootstrap it into your home network just by tapping your

322
00:17:29.880 --> 00:17:33.920
<v Speaker 2>mobile phone to the network access point, establishes connectivity instantly.

323
00:17:34.440 --> 00:17:37.000
<v Speaker 2>Or say you have visitors over, you could grant them

324
00:17:37.039 --> 00:17:40.319
<v Speaker 2>temporary access to specific services like the guest Wi Fi

325
00:17:40.400 --> 00:17:43.640
<v Speaker 2>or smart lights using passlets. These are like little digitally

326
00:17:43.720 --> 00:17:47.880
<v Speaker 2>signed permission slips containing connectivity and access details transferred simply

327
00:17:47.920 --> 00:17:49.519
<v Speaker 2>by tapping their device to yours.

328
00:17:49.839 --> 00:17:53.200
<v Speaker 1>And with NFC that really short range just a few

329
00:17:53.240 --> 00:17:57.279
<v Speaker 1>centimeters usually that actually makes the communication inherently authentic, doesn't it.

330
00:17:57.279 --> 00:17:58.759
<v Speaker 1>It's hard to ease drop if you have to be

331
00:17:58.799 --> 00:18:01.480
<v Speaker 1>that close. It almost some security problems just by its

332
00:18:01.480 --> 00:18:02.240
<v Speaker 1>physical nature.

333
00:18:02.359 --> 00:18:05.400
<v Speaker 2>It absolutely does that limited range is a powerful built

334
00:18:05.440 --> 00:18:07.160
<v Speaker 2>in security feature in many ways.

335
00:18:07.440 --> 00:18:09.839
<v Speaker 1>Wow. Okay, so we really covered a lot. We've seen

336
00:18:09.839 --> 00:18:14.039
<v Speaker 1>how this seemingly simple technology are FID underpins so much,

337
00:18:14.119 --> 00:18:17.279
<v Speaker 1>from massive supply chains down to our passports and potentially

338
00:18:17.319 --> 00:18:20.720
<v Speaker 1>everyday objects, but also how incredibly complex the layers of

339
00:18:20.720 --> 00:18:23.640
<v Speaker 1>its security and privacy can be. From those physical kill

340
00:18:23.680 --> 00:18:27.799
<v Speaker 1>commands all the way to sophisticated crypto hash chains ECC,

341
00:18:28.200 --> 00:18:30.319
<v Speaker 1>and then the huge challenges of scaling it all up

342
00:18:30.359 --> 00:18:34.680
<v Speaker 1>to a truly ubiquitous world. It's clear rfid's invisible world

343
00:18:34.759 --> 00:18:35.519
<v Speaker 1>is anything.

344
00:18:35.240 --> 00:18:38.279
<v Speaker 2>But simple, and it really raises an important final thought

345
00:18:38.400 --> 00:18:42.119
<v Speaker 2>or maybe a question as RFID becomes truly ubiquitous, As

346
00:18:42.119 --> 00:18:45.079
<v Speaker 2>it gets integrated into countless devices and environments around us.

347
00:18:45.200 --> 00:18:48.799
<v Speaker 2>What are the potential implications if we don't prioritize privacy

348
00:18:48.839 --> 00:18:52.240
<v Speaker 2>and security, if they aren't deeply embedded features in every

349
00:18:52.279 --> 00:18:54.920
<v Speaker 2>new application right from the start, rather than being tacked

350
00:18:54.920 --> 00:18:57.480
<v Speaker 2>on as an afterthought. It just underscores, I think, the

351
00:18:57.599 --> 00:19:00.720
<v Speaker 2>critical need for continued technical innovation yet, but also for

352
00:19:00.799 --> 00:19:03.880
<v Speaker 2>public education and for strong policy efforts. We need to

353
00:19:03.960 --> 00:19:06.279
<v Speaker 2>ensure that all this convenience doesn't ultimately come at the

354
00:19:06.319 --> 00:19:08.640
<v Speaker 2>cost of our digital safety in our personal autonomy.