WEBVTT 00:00:00.000 --> 00:00:02.280 STEVE MILES: And so we started out with a little 00:00:02.280 --> 00:00:03.450 on the network. 00:00:03.450 --> 00:00:07.980 And it was a great deal of fun to think back and hear 00:00:07.980 --> 00:00:10.740 some of the origins of where some of the thinking was 00:00:10.740 --> 00:00:14.130 and to have a chance to discuss exactly where we 00:00:14.130 --> 00:00:17.383 are in the evolution of this-- 00:00:17.383 --> 00:00:19.800 either it's the internet of things, or according to Steve, 00:00:19.800 --> 00:00:23.520 it's things on the internet, whichever we decide it is. 00:00:23.520 --> 00:00:26.460 But we're going to move right into tags. 00:00:26.460 --> 00:00:31.680 And we have the director of the Auto-ID Labs 00:00:31.680 --> 00:00:35.940 at Fudan University in Shanghai, Professor Hao Min, who 00:00:35.940 --> 00:00:38.374 will start off the discussion. 00:00:41.800 --> 00:00:42.300 HAO MIN: OK. 00:00:42.300 --> 00:00:46.920 And the whole RFID technology, I think, 00:00:46.920 --> 00:00:50.300 the first thing starting is from the RFID tags. 00:00:50.300 --> 00:00:53.210 So people are thinking about the tag, 00:00:53.210 --> 00:00:56.720 and what the performance of the costs of the RFID tag 00:00:56.720 --> 00:01:00.200 may just influence the adoption of this technology, 00:01:00.200 --> 00:01:02.540 because this is kind of the basis 00:01:02.540 --> 00:01:04.519 for this whole technology. 00:01:04.519 --> 00:01:10.010 So today I will talk about what's the user 00:01:10.010 --> 00:01:11.840 requirement for tags right now. 00:01:11.840 --> 00:01:15.830 And so this will promote us to research what 00:01:15.830 --> 00:01:20.330 the next generation we can do for the tag performance 00:01:20.330 --> 00:01:21.620 enhancement. 00:01:21.620 --> 00:01:24.590 And also, I'll introduce some new technologies 00:01:24.590 --> 00:01:27.810 being developed right now. 00:01:27.810 --> 00:01:31.380 And there are many, already, some adoptions 00:01:31.380 --> 00:01:32.970 of this RFID technology. 00:01:32.970 --> 00:01:35.360 Some users are [? users ?] tags. 00:01:35.360 --> 00:01:39.420 So it's estimated that millions of tags be used right now. 00:01:39.420 --> 00:01:41.400 But there are still some problems. 00:01:41.400 --> 00:01:45.030 People want a longer read range. 00:01:45.030 --> 00:01:47.360 Right now, a [INAUDIBLE] tag can reach the range 00:01:47.360 --> 00:01:50.142 of approximately 5 meters. 00:01:50.142 --> 00:01:52.350 But people are thinking about whether you need, like, 00:01:52.350 --> 00:01:53.070 10 meters. 00:01:53.070 --> 00:01:57.150 If you reach 10 meters, more application will be wider. 00:01:57.150 --> 00:01:58.770 The other thing that's really critical 00:01:58.770 --> 00:02:01.170 is the 100% read coverage. 00:02:01.170 --> 00:02:04.320 From the report from Walmart and the [INAUDIBLE],, 00:02:04.320 --> 00:02:07.860 no one can reach 100% read right now, 00:02:07.860 --> 00:02:12.310 probably 99%, or someone even only gets, like, 90%. 00:02:12.310 --> 00:02:16.750 So what happens if 1% is still missing? 00:02:16.750 --> 00:02:20.590 So when we have any technologies, what we can 00:02:20.590 --> 00:02:24.130 improve that to get 100% read. 00:02:24.130 --> 00:02:27.720 Also there's the issue is the people hope to get similar read 00:02:27.720 --> 00:02:29.340 performance in worldwide frequency, 00:02:29.340 --> 00:02:31.715 because in worldwide, the frequencies are very different, 00:02:31.715 --> 00:02:35.430 from [INAUDIBLE] to [INAUDIBLE] But different countries 00:02:35.430 --> 00:02:37.102 use different frequencies. 00:02:37.102 --> 00:02:38.310 But the tag is already there. 00:02:38.310 --> 00:02:40.447 The tag will move around the world, 00:02:40.447 --> 00:02:42.030 will be read in different frequencies. 00:02:42.030 --> 00:02:45.190 Because of the RF properties of these tags, 00:02:45.190 --> 00:02:47.250 the read performance in different frequencies 00:02:47.250 --> 00:02:48.540 will be different. 00:02:48.540 --> 00:02:54.510 But people want that it should be almost a similar performance 00:02:54.510 --> 00:02:56.920 around the world. 00:02:56.920 --> 00:02:59.910 And so in that sense, the security and [INAUDIBLE].. 00:02:59.910 --> 00:03:03.060 Just the previous speakers from the web, 00:03:03.060 --> 00:03:04.920 they were all talking about securities. 00:03:04.920 --> 00:03:09.870 If there's no security, I think the RFID application will just 00:03:09.870 --> 00:03:11.170 get very limited. 00:03:11.170 --> 00:03:14.320 So just this [INAUDIBLE] say that the hack, 00:03:14.320 --> 00:03:16.650 EPC hack get attacked from-- 00:03:16.650 --> 00:03:19.180 I think it's from two days ago. 00:03:19.180 --> 00:03:22.410 So people pretend to be hack members, 00:03:22.410 --> 00:03:24.570 and they pull their posts on the hack. 00:03:24.570 --> 00:03:29.490 So all the hack members got all these junk mails. 00:03:29.490 --> 00:03:32.110 So this is the issue, is when internet comes out, 00:03:32.110 --> 00:03:33.540 there's no security concerns. 00:03:33.540 --> 00:03:35.070 Internet comes from academics. 00:03:35.070 --> 00:03:38.880 So at first, DOD sponsor the program, 00:03:38.880 --> 00:03:40.260 and then grow and grow. 00:03:40.260 --> 00:03:43.470 Just no one is planning all the security 00:03:43.470 --> 00:03:44.520 issues on the internet. 00:03:44.520 --> 00:03:46.500 Right now the problems comes out. 00:03:46.500 --> 00:03:48.730 So right now we're talking about internet of things. 00:03:48.730 --> 00:03:52.270 We need to really think about the security issues. 00:03:52.270 --> 00:03:55.470 So what we are doing right now? 00:03:55.470 --> 00:03:58.650 We are trying to improve the read range. 00:03:58.650 --> 00:04:00.540 Improve the read range is basically 00:04:00.540 --> 00:04:04.260 is to reduce the power consumption of the tag chip. 00:04:04.260 --> 00:04:09.070 If it consumes less power, than the read range will be longer. 00:04:09.070 --> 00:04:13.050 So what we are doing is we try to improve the conversion 00:04:13.050 --> 00:04:14.850 efficiency for rectifier, because this 00:04:14.850 --> 00:04:18.870 is the IF signals coming from the reader and to the tag chip. 00:04:18.870 --> 00:04:22.350 And the tag chip will convert that into a DC, DC power, 00:04:22.350 --> 00:04:23.700 to a [INAUDIBLE] chip. 00:04:23.700 --> 00:04:26.670 So if you can improve the conversion efficiency, 00:04:26.670 --> 00:04:31.710 then it's the read range will get also improved. 00:04:31.710 --> 00:04:34.710 And also, there are some special technologies 00:04:34.710 --> 00:04:37.050 being studied to reduce the power consumption. 00:04:37.050 --> 00:04:40.470 For example, as we called subthreshold digital circuits. 00:04:40.470 --> 00:04:43.290 If you have a background on digital circuits, 00:04:43.290 --> 00:04:47.160 we know why we need a power supply 00:04:47.160 --> 00:04:50.280 to power the digital circuitry, because the digital circuitry 00:04:50.280 --> 00:04:51.930 is a need to switch. 00:04:51.930 --> 00:04:57.720 But there's a threshold to switch the circuit on and off. 00:04:57.720 --> 00:04:58.720 So there is a threshold. 00:04:58.720 --> 00:05:01.500 But if the threshold is lower, then the voltage 00:05:01.500 --> 00:05:03.950 applied to that will be lower. 00:05:03.950 --> 00:05:09.170 Right now there's a circuitry which we do not use the regular 00:05:09.170 --> 00:05:11.570 on and off mechanism. 00:05:11.570 --> 00:05:14.450 But we use a circuit called subthreshold. 00:05:14.450 --> 00:05:16.340 Even if you apply a voltage which 00:05:16.340 --> 00:05:17.960 is lower than the threshold, if still 00:05:17.960 --> 00:05:20.070 controls the circuit on off. 00:05:20.070 --> 00:05:22.100 Then this will reduce the power consumption. 00:05:22.100 --> 00:05:23.683 And also there is a new circuit record 00:05:23.683 --> 00:05:24.892 called [INAUDIBLE] circuitry. 00:05:24.892 --> 00:05:26.930 Although this circuitry type is a long time ago, 00:05:26.930 --> 00:05:28.790 but recently we find out this probably 00:05:28.790 --> 00:05:31.500 can apply to the tag chip design. 00:05:31.500 --> 00:05:33.860 There's also some other conventional low-power circuit 00:05:33.860 --> 00:05:37.100 design technologies. 00:05:37.100 --> 00:05:39.140 This is what we've done in recently. 00:05:39.140 --> 00:05:41.660 We recently published a paper here, 00:05:41.660 --> 00:05:46.790 published paper on the improved the rectifier efficiency 00:05:46.790 --> 00:05:48.740 by about 30% to 100%. 00:05:48.740 --> 00:05:50.450 That means double the efficiency. 00:05:50.450 --> 00:05:54.560 Normally, the efficiency is around, like, 20%, 30%. 00:05:54.560 --> 00:05:58.610 And we can improve it into 40% or 60%. 00:05:58.610 --> 00:06:01.520 Basically we use called a bootstrap 00:06:01.520 --> 00:06:07.560 to rectifier to get rid of the VT drop of the circuitry. 00:06:07.560 --> 00:06:11.010 The second thing we are working on 00:06:11.010 --> 00:06:13.860 is there's a circuitry called adiabatic circuitry, which 00:06:13.860 --> 00:06:18.030 means the charge in the circuitry can be recycled. 00:06:18.030 --> 00:06:21.910 But previously, some people already published the papers. 00:06:21.910 --> 00:06:23.490 They used this adiabatic circuitry, 00:06:23.490 --> 00:06:25.020 but they convert it normally. 00:06:25.020 --> 00:06:28.840 Normally an electronic system is powered by a DC power. 00:06:28.840 --> 00:06:31.710 So they convert the DC power to a power clock, 00:06:31.710 --> 00:06:34.350 because adiabatic circuitry needs a power clock, 00:06:34.350 --> 00:06:35.580 and then convert-- 00:06:35.580 --> 00:06:38.160 and then to power the adiabatic circuitry. 00:06:38.160 --> 00:06:41.910 But in RFID, because we already got the power supply 00:06:41.910 --> 00:06:44.770 from the radio way, radio signal. 00:06:44.770 --> 00:06:48.550 So the radio signal, it's already powered clock. 00:06:48.550 --> 00:06:49.500 It's already clock. 00:06:49.500 --> 00:06:53.340 So we can directly use that as the power clock. 00:06:53.340 --> 00:06:58.260 So then we can rid of the rectifier circuitry, so which 00:06:58.260 --> 00:07:00.300 eliminates the loss of the energy. 00:07:00.300 --> 00:07:04.440 And also, it's because the adiabatic circuitry consumes 00:07:04.440 --> 00:07:06.240 less power than normal logic, then 00:07:06.240 --> 00:07:08.430 the whole power consumption of the tag 00:07:08.430 --> 00:07:10.580 is getting reduced a lot. 00:07:10.580 --> 00:07:15.800 By doing this, we estimated that probably we can get only 10% 00:07:15.800 --> 00:07:18.230 of the power consumption as normal circuitry. 00:07:18.230 --> 00:07:20.090 So this means what we can do-- we 00:07:20.090 --> 00:07:25.100 can increase the read range of the tag by a factor of 2 to 3. 00:07:28.560 --> 00:07:32.290 Since I already talked about the subthreshold circuitry, 00:07:32.290 --> 00:07:37.180 this curve shows that the power consumption 00:07:37.180 --> 00:07:40.690 by a different power supply voltage. 00:07:40.690 --> 00:07:44.020 If we use the supply voltage of, say, 00:07:44.020 --> 00:07:48.190 less than the threshold voltage of a transistor, 00:07:48.190 --> 00:07:51.580 we may get four orders of magnitude of reduction 00:07:51.580 --> 00:07:53.020 of power consumption. 00:07:53.020 --> 00:07:57.130 But this circuitry has the limit of the working frequency 00:07:57.130 --> 00:08:01.240 is very low, probably only go to 1 meg. 00:08:01.240 --> 00:08:02.410 We've got a simulation. 00:08:02.410 --> 00:08:07.270 If we let the circuitry only work at 1 meg, at 1 volt, 00:08:07.270 --> 00:08:12.760 a tag chip only consumes 10 nanoamp, which is only 1% 00:08:12.760 --> 00:08:16.400 of a normal power consumption of tag. 00:08:16.400 --> 00:08:19.750 By doing this, we may increase the read range 00:08:19.750 --> 00:08:21.040 by a factor of even 10. 00:08:24.450 --> 00:08:28.170 And also, we are trying to figure out what really prevents 00:08:28.170 --> 00:08:31.430 your read from 100% of the tag. 00:08:31.430 --> 00:08:33.500 So why these missing tag? 00:08:33.500 --> 00:08:35.070 The tag is missing. 00:08:35.070 --> 00:08:37.309 A reason we find our way that if you 00:08:37.309 --> 00:08:42.870 put two tags next to each other, one is closer to the reader. 00:08:42.870 --> 00:08:45.830 Another one is behind that. 00:08:45.830 --> 00:08:49.160 And the one behind that will probably cannot be read. 00:08:49.160 --> 00:08:52.370 But if you remove the one closer to the read, 00:08:52.370 --> 00:08:54.320 the one there can be read. 00:08:54.320 --> 00:08:57.950 So look like the one closer to the reader 00:08:57.950 --> 00:09:01.750 masked the tags behind. 00:09:01.750 --> 00:09:04.990 This is because the normal tag technology, 00:09:04.990 --> 00:09:07.000 they use a rectifier, because when 00:09:07.000 --> 00:09:10.560 the tag is closer to the reader, it gets more energy. 00:09:10.560 --> 00:09:16.010 So the energy goes into the tech needs to go somewhere. 00:09:16.010 --> 00:09:20.500 So it has a circuitry that convert it just to heat. 00:09:20.500 --> 00:09:25.830 So it consumes high power when the tag close to the reader. 00:09:25.830 --> 00:09:27.660 Because the tag consumes the power, 00:09:27.660 --> 00:09:30.120 then the magnetic wave will just stop there, 00:09:30.120 --> 00:09:34.630 will not go further into the tag after that. 00:09:34.630 --> 00:09:38.310 So what we do, we are trying to have a new circuitry, which 00:09:38.310 --> 00:09:43.530 try when-- it detects whether the tag got enough power. 00:09:43.530 --> 00:09:45.780 If get enough power, it just tries 00:09:45.780 --> 00:09:49.740 to detune the tune circuitry so that it does not 00:09:49.740 --> 00:09:50.940 consume energy. 00:09:50.940 --> 00:09:53.400 It only absorbs energy it needed. 00:09:53.400 --> 00:09:56.230 All the other energy is just let go away. 00:09:56.230 --> 00:10:00.720 So this can have this tag behind that can 00:10:00.720 --> 00:10:03.220 get enough energy to get power. 00:10:03.220 --> 00:10:04.740 So this is we are doing. 00:10:04.740 --> 00:10:07.050 So what we are doing right now is 00:10:07.050 --> 00:10:10.200 so we can detect the power supply of the tag 00:10:10.200 --> 00:10:13.320 and then compare to a reference, and then 00:10:13.320 --> 00:10:19.540 try to tune the [? resonant ?] capacitor of the antenna. 00:10:19.540 --> 00:10:22.340 And it just makes it work. 00:10:25.080 --> 00:10:29.140 Also, for the worldwide frequency of things, 00:10:29.140 --> 00:10:31.440 we also use this adaptive tag-- 00:10:31.440 --> 00:10:33.270 adaptive tuning waves. 00:10:33.270 --> 00:10:34.920 So it will detect the power. 00:10:34.920 --> 00:10:39.000 If get less power, it will try to tune the resonant cap 00:10:39.000 --> 00:10:42.450 to find a best point, which resonant 00:10:42.450 --> 00:10:45.190 at the frequency of the reader. 00:10:45.190 --> 00:10:47.730 So this make the performance get improved. 00:10:50.580 --> 00:10:53.740 Also, security and [? authentications ?] 00:10:53.740 --> 00:10:55.170 get considered. 00:10:55.170 --> 00:10:58.650 So there will be a two-way [? authentication ?] is 00:10:58.650 --> 00:11:01.470 designed to both the reader-- 00:11:01.470 --> 00:11:04.530 you [? authenticate ?] the tags, and the reader tag get 00:11:04.530 --> 00:11:06.810 [? authentication ?] from the readers. 00:11:06.810 --> 00:11:09.390 But this is really challenging because the tag 00:11:09.390 --> 00:11:12.490 is so tiny chip, and it consumes less power. 00:11:12.490 --> 00:11:14.760 So it needs a low cost, low power. 00:11:14.760 --> 00:11:17.100 But it also needs adequate security 00:11:17.100 --> 00:11:19.870 between tag and readers. 00:11:19.870 --> 00:11:22.790 We have tried a way which we integrate 00:11:22.790 --> 00:11:26.380 some [? authentication ?] into the gen 2 protocol 00:11:26.380 --> 00:11:29.560 to try to have the tag and the readers get authentication. 00:11:29.560 --> 00:11:32.440 We have some first simulations down. 00:11:32.440 --> 00:11:35.620 It kind of looks like work, but we still need a lot of work 00:11:35.620 --> 00:11:36.870 to do. 00:11:36.870 --> 00:11:38.780 OK, that's my presentation. 00:11:38.780 --> 00:11:39.990 Thank you. 00:11:39.990 --> 00:11:45.485 [APPLAUSE] 00:11:45.485 --> 00:11:46.860 STEVE MILES: And our next speaker 00:11:46.860 --> 00:11:50.400 is Sang-Gug Lee, who is an associate professor 00:11:50.400 --> 00:11:54.570 in the School of Engineering at the Auto-ID Labs at ICU Korea. 00:12:07.888 --> 00:12:08.680 SANG-GUG LEE: Yeah. 00:12:14.020 --> 00:12:19.630 In my talk, I would like to quickly go through the USN 00:12:19.630 --> 00:12:22.420 technology in Korea, and then we introduce 00:12:22.420 --> 00:12:25.840 what we are working on in Auto-ID Lab Korea. 00:12:25.840 --> 00:12:29.020 What I mean by "USN" is an Ubiquitous Sensor Network, 00:12:29.020 --> 00:12:32.970 which is known as a wireless sensing network. 00:12:32.970 --> 00:12:36.290 So the content would be some brief overview 00:12:36.290 --> 00:12:40.070 of what is going on inside Korea in this area, 00:12:40.070 --> 00:12:43.370 and then in relation to what's going on in Korea, what we 00:12:43.370 --> 00:12:47.080 are doing in Auto-ID Lab Korea. 00:12:47.080 --> 00:12:50.440 Probably some of you are familiar with this A39. 00:12:50.440 --> 00:12:53.800 It's called the Information Technology A39 Strategy 00:12:53.800 --> 00:12:55.480 that Korean government came up with. 00:12:55.480 --> 00:12:59.740 And there are three areas where we 00:12:59.740 --> 00:13:05.230 are trying to promote services, and then three infrastructures, 00:13:05.230 --> 00:13:08.410 and then nine different areas of growth engine 00:13:08.410 --> 00:13:11.140 that the Korean government is working on. 00:13:11.140 --> 00:13:18.160 And in this A39 strategy, you find that in the service area, 00:13:18.160 --> 00:13:20.860 the government is trying to provide 00:13:20.860 --> 00:13:23.860 the RFID-based services, and then also 00:13:23.860 --> 00:13:28.300 in the structure area by utilizing the Ubiquitous Sensor 00:13:28.300 --> 00:13:29.530 Network. 00:13:29.530 --> 00:13:32.410 We consider-- the Korean government considers-- 00:13:32.410 --> 00:13:36.010 Ubiquitous Sensor Network as one of the infrastructures 00:13:36.010 --> 00:13:39.370 that they would like to deploy in Korea, in South Korea. 00:13:39.370 --> 00:13:43.360 And then through that, hoping to be able to deploy these nine 00:13:43.360 --> 00:13:45.340 different area across engines. 00:13:48.430 --> 00:13:52.990 And if we're being more specific with the time schedule, 00:13:52.990 --> 00:13:58.000 each of these nine areas, they have schedules, and then 00:13:58.000 --> 00:14:00.220 when they would like to-- what kind of technology 00:14:00.220 --> 00:14:01.870 for the commercialization. 00:14:01.870 --> 00:14:05.380 And if you look at the pink-colored icon there, 00:14:05.380 --> 00:14:08.770 this is the RFID and USN area, where the government 00:14:08.770 --> 00:14:13.480 is trying to commercialize a mobile RFID by year 2006, which 00:14:13.480 --> 00:14:16.120 is before the end of this year, and then 00:14:16.120 --> 00:14:23.290 hoping to be able to deploy the sensor tag or sensor node 00:14:23.290 --> 00:14:27.710 technology for the public application. 00:14:27.710 --> 00:14:31.610 And the concept of a public ubiquitous sensor network 00:14:31.610 --> 00:14:37.580 is basically based on broadband convergence network, 00:14:37.580 --> 00:14:42.020 all these services, where the sensor network-based services 00:14:42.020 --> 00:14:44.570 would be available such that everywhere, 00:14:44.570 --> 00:14:46.880 everything with RFID tags. 00:14:46.880 --> 00:14:51.200 And we sense the sensing IDs and environmental information, 00:14:51.200 --> 00:14:53.990 and then real-time monitoring and the control 00:14:53.990 --> 00:14:56.060 through the network would be available. 00:14:56.060 --> 00:14:58.580 That's the idea here. 00:14:58.580 --> 00:15:00.650 And the applications, I'm sure most of you 00:15:00.650 --> 00:15:02.370 are already familiar with. 00:15:02.370 --> 00:15:06.080 They are smart buildings, factory automation 00:15:06.080 --> 00:15:08.870 and monitoring, asset monitoring and management, 00:15:08.870 --> 00:15:11.950 structural health monitoring, and environmental monitoring. 00:15:11.950 --> 00:15:15.620 All this, we call this as a public ubiquitous sensor 00:15:15.620 --> 00:15:17.060 network technology. 00:15:17.060 --> 00:15:21.590 These are the targets the government is thrusting. 00:15:21.590 --> 00:15:26.300 And if you look at the technology tree, 00:15:26.300 --> 00:15:30.920 these are the infrastructures, studying the sensor, 00:15:30.920 --> 00:15:33.650 and then service available there. 00:15:33.650 --> 00:15:36.080 While I was watching this morning, 00:15:36.080 --> 00:15:41.090 I was getting this philosophical question. 00:15:41.090 --> 00:15:43.470 We talk about all these technologies 00:15:43.470 --> 00:15:47.420 trying to make our life really easy and happy and all that. 00:15:47.420 --> 00:15:51.320 But I'm also, at the same time, finding out 00:15:51.320 --> 00:15:53.540 how all these technologies actually 00:15:53.540 --> 00:15:55.940 make our life miserable. 00:15:55.940 --> 00:15:58.520 You realize what I mean, right? 00:15:58.520 --> 00:16:01.340 All these mobile phone technologies and internet. 00:16:01.340 --> 00:16:04.770 It seems like it's helping us in many different areas. 00:16:04.770 --> 00:16:07.010 But at the same time, we're finding 00:16:07.010 --> 00:16:10.550 ourselves running like crazy. 00:16:10.550 --> 00:16:12.620 It makes our life so busy. 00:16:12.620 --> 00:16:15.230 It doesn't really make our life happy. 00:16:15.230 --> 00:16:19.000 But I guess the race has started. 00:16:19.000 --> 00:16:22.340 It's just an unstoppable race that everybody has 00:16:22.340 --> 00:16:24.820 to run all at the same time. 00:16:24.820 --> 00:16:29.240 But anyway, so these are what the government is trying to do, 00:16:29.240 --> 00:16:36.890 and this is their roadmap for actual milestones and roadmaps 00:16:36.890 --> 00:16:41.030 and how they would like to implement 00:16:41.030 --> 00:16:44.030 these technologies for the commercial applications. 00:16:44.030 --> 00:16:48.860 And in relation to that, what we try to do in Auto-ID Lab Korea 00:16:48.860 --> 00:16:49.790 is that-- 00:16:49.790 --> 00:16:52.970 these are what I have just talked about is basically 00:16:52.970 --> 00:16:56.000 of public ubiquitous sensor network technology 00:16:56.000 --> 00:16:58.220 that they try to deploy. 00:16:58.220 --> 00:17:01.850 And in Auto-ID Lab Korea here, what we're trying to do 00:17:01.850 --> 00:17:05.440 is we try to make this public ubiquitous sensor 00:17:05.440 --> 00:17:09.030 network [INAUDIBLE] connected to the EPC network. 00:17:09.030 --> 00:17:13.069 So the theme of technology thrust in Auto-ID Lab Korea 00:17:13.069 --> 00:17:16.700 is EPC network base a sensor network technology. 00:17:16.700 --> 00:17:18.710 That's what we're trying to go after. 00:17:18.710 --> 00:17:24.560 And we have seven professors involved in this thrust coming 00:17:24.560 --> 00:17:26.329 from different backgrounds. 00:17:26.329 --> 00:17:30.080 And our focus area is the hardware and communication 00:17:30.080 --> 00:17:32.910 technology for EPC-based next-generation ubiquitous 00:17:32.910 --> 00:17:37.730 sensor network, and some of the middleware technology for EPC 00:17:37.730 --> 00:17:42.650 sensor network, and then the privacy and security issues, 00:17:42.650 --> 00:17:49.190 as well as the business model development for the EPC sensor 00:17:49.190 --> 00:17:51.300 network applications. 00:17:51.300 --> 00:17:52.940 And these are some of the research work 00:17:52.940 --> 00:17:56.390 that has been done as part of those thrust. 00:17:56.390 --> 00:18:00.500 And this is some hardware [INAUDIBLE] 00:18:00.500 --> 00:18:02.970 we've been working on the impulse radio 00:18:02.970 --> 00:18:06.290 development, which is being developed for the ranging 00:18:06.290 --> 00:18:08.990 and locationing. 00:18:08.990 --> 00:18:10.850 There's some typo there. 00:18:10.850 --> 00:18:13.640 We've been working on some pulse generator circuits, 00:18:13.640 --> 00:18:16.070 very low-power pulse generator circuits on that, and then 00:18:16.070 --> 00:18:20.510 also very low-power correlator circuits and stuff like that. 00:18:20.510 --> 00:18:24.560 So basically, we're working on transceiver designs, modems 00:18:24.560 --> 00:18:31.680 for some algorithms, and then for arranging and location 00:18:31.680 --> 00:18:32.420 purposes. 00:18:32.420 --> 00:18:34.280 At the same time, we're also working 00:18:34.280 --> 00:18:36.600 on the reactive microradio technology, 00:18:36.600 --> 00:18:39.960 which means that the sensor responding to the signal. 00:18:39.960 --> 00:18:42.500 In other words, the sensor is under the sleep mode 00:18:42.500 --> 00:18:45.200 and responds only when it is being waked up 00:18:45.200 --> 00:18:49.520 by some wake-up signals, which is, I'm sure, 00:18:49.520 --> 00:18:52.940 a number of research institutes are working on at the moment. 00:18:55.450 --> 00:18:57.720 Also, this is some of the work that we're 00:18:57.720 --> 00:19:01.830 working on on the sensor network [INAUDIBLE] architecture, where 00:19:01.830 --> 00:19:05.910 the circle area, I think, the whole [INAUDIBLE] represents 00:19:05.910 --> 00:19:08.820 the EPC network. 00:19:08.820 --> 00:19:12.360 And the circle part represents the area 00:19:12.360 --> 00:19:15.240 where in order for this EPC network 00:19:15.240 --> 00:19:19.050 being connected to the public ubiquitous sensor network 00:19:19.050 --> 00:19:20.820 system. 00:19:20.820 --> 00:19:24.690 And we all saw some progress is being made in the secret 00:19:24.690 --> 00:19:27.870 and privacy area as well. 00:19:27.870 --> 00:19:30.070 And through all this activity-- in other words, 00:19:30.070 --> 00:19:33.010 [? RF-ran ?] chip sensor interfaces and networking 00:19:33.010 --> 00:19:35.490 and software, as well as a business application, 00:19:35.490 --> 00:19:38.490 privacy and security issue, we're hoping that the research 00:19:38.490 --> 00:19:42.780 that we're doing would be related to the future 00:19:42.780 --> 00:19:46.590 standardization that connects the public ubiquitous sensor 00:19:46.590 --> 00:19:49.510 network with the EPC network. 00:19:49.510 --> 00:19:51.441 That's it. 00:19:51.441 --> 00:19:54.363 [APPLAUSE] 00:19:59.240 --> 00:20:01.178 STEVE MILES: Our next speaker and member 00:20:01.178 --> 00:20:02.970 of the conference committee, Gisele Bennett 00:20:02.970 --> 00:20:04.200 from Georgia Tech. 00:20:24.210 --> 00:20:25.920 GISELE BENNETT: All right, I am going 00:20:25.920 --> 00:20:29.940 to somewhat switch gears and go as quickly as I possibly 00:20:29.940 --> 00:20:32.730 can with the 10 minutes allocated with, I think, 00:20:32.730 --> 00:20:34.170 30-some odd slides. 00:20:34.170 --> 00:20:37.290 And the whole point that I hope you walk away with 00:20:37.290 --> 00:20:39.880 is understanding the importance of requirements. 00:20:39.880 --> 00:20:42.360 We've talked about RFID, and we've talked 00:20:42.360 --> 00:20:44.340 about a number of applications. 00:20:44.340 --> 00:20:47.070 But understanding the requirements and what solution 00:20:47.070 --> 00:20:49.950 is going to meet your application is really critical. 00:20:49.950 --> 00:20:52.560 And we had a project that I'm going to focus 00:20:52.560 --> 00:20:56.520 on, on a container project, and looking at sensors that have 00:20:56.520 --> 00:20:57.270 built-in-- 00:20:59.880 --> 00:21:04.320 tags that have built-in sensors to monitor the environment, 00:21:04.320 --> 00:21:06.520 to monitor the condition of an asset. 00:21:06.520 --> 00:21:08.460 And so this is one of my favorite slides 00:21:08.460 --> 00:21:11.070 because I think it goes back to World War II, 00:21:11.070 --> 00:21:12.300 if I'm not mistaken. 00:21:12.300 --> 00:21:16.380 And really, you're really trying to find something 00:21:16.380 --> 00:21:18.480 in all the things that we're talking about. 00:21:18.480 --> 00:21:22.050 RFID happens to be one mechanism. 00:21:22.050 --> 00:21:26.220 And of course, the big motivator was the Walmart and the DOD. 00:21:26.220 --> 00:21:29.370 And one of my focus areas will be 00:21:29.370 --> 00:21:31.950 the DOD particular application. 00:21:31.950 --> 00:21:36.810 RFID's everywhere-- comes under different forms, 00:21:36.810 --> 00:21:40.560 different marketing, but everybody's pushing towards it. 00:21:40.560 --> 00:21:42.540 And really, as I indicated initially, 00:21:42.540 --> 00:21:44.900 asset tracking is really our focus. 00:21:44.900 --> 00:21:46.920 So if it's RFID, terrific. 00:21:46.920 --> 00:21:49.380 If it's some other mechanism, that's good too. 00:21:49.380 --> 00:21:52.680 Understanding the requirements is a really critical element 00:21:52.680 --> 00:21:54.307 to all of this. 00:21:54.307 --> 00:21:55.890 As you can see-- and this is, I think, 00:21:55.890 --> 00:21:57.937 actually, I have to apologize, an old slide, 00:21:57.937 --> 00:21:59.520 because I'm sure the number of patents 00:21:59.520 --> 00:22:03.690 have actually increased beyond that in the RFID arena. 00:22:03.690 --> 00:22:06.060 I don't need to get into what automatic identification 00:22:06.060 --> 00:22:08.430 tracking is with this audience. 00:22:08.430 --> 00:22:09.840 We've got a number of elements. 00:22:09.840 --> 00:22:11.670 We've talked about them in some of the other talks. 00:22:11.670 --> 00:22:13.003 We're going to talk about them-- 00:22:13.003 --> 00:22:16.050 they're going to come up in other discussions. 00:22:16.050 --> 00:22:20.310 But again, tracking something, storing information, 00:22:20.310 --> 00:22:22.470 and doing something with that data. 00:22:22.470 --> 00:22:25.080 All of these elements have come up in the talks. 00:22:25.080 --> 00:22:27.210 They're all critical elements, and they come 00:22:27.210 --> 00:22:29.930 in different shapes and forms. 00:22:29.930 --> 00:22:33.740 Biometrics is another technology that 00:22:33.740 --> 00:22:38.280 should be looked at for security, authenticity. 00:22:38.280 --> 00:22:40.580 And that's something that might be integrated in-- 00:22:40.580 --> 00:22:45.620 not RFID, but integrated within the RFID systems, if you will, 00:22:45.620 --> 00:22:48.380 in data gathering. 00:22:48.380 --> 00:22:50.023 Now, when I talk about requirements, 00:22:50.023 --> 00:22:51.440 this is one of my favorite slides, 00:22:51.440 --> 00:22:55.530 and I contribute this to Nick [? Toogis ?] from the DOD IAT 00:22:55.530 --> 00:22:56.030 office. 00:22:56.030 --> 00:22:58.920 And an understanding of requires is really critical. 00:22:58.920 --> 00:23:04.040 So Walmart has been the pin-up CPG that 00:23:04.040 --> 00:23:09.050 gets referenced as the start of all of the RFID flurry, 00:23:09.050 --> 00:23:11.900 and DOD followed suit soon thereafter. 00:23:11.900 --> 00:23:15.320 But DOD can't really apply a lot of the commercial applications, 00:23:15.320 --> 00:23:22.370 because if they could, then their stores would move-- 00:23:22.370 --> 00:23:25.010 Walmart stores would move every so often. 00:23:27.900 --> 00:23:31.710 Christmas would be a random event, maybe every five years, 00:23:31.710 --> 00:23:33.135 not once a year in December. 00:23:36.690 --> 00:23:40.870 The associates would be wearing different types of vests. 00:23:40.870 --> 00:23:43.120 And a stock-out means something completely 00:23:43.120 --> 00:23:45.140 different for the DOD than it does for Walmart. 00:23:45.140 --> 00:23:47.920 So understanding those requirements 00:23:47.920 --> 00:23:50.590 and the environment you have to work in kind of changes 00:23:50.590 --> 00:23:52.090 what technology you're going to use 00:23:52.090 --> 00:23:55.140 and how you're going to apply that technology. 00:23:55.140 --> 00:24:00.000 The way we got into this, we were approached by the Navy 00:24:00.000 --> 00:24:03.870 to solve a problem of managing their high-value assets 00:24:03.870 --> 00:24:05.460 and, in particular, the engines. 00:24:05.460 --> 00:24:08.310 The problem is they were improperly stored 00:24:08.310 --> 00:24:10.950 and improperly tracked. 00:24:10.950 --> 00:24:13.320 So when they were able to find an engine that they 00:24:13.320 --> 00:24:15.120 were looking for, they were supposed 00:24:15.120 --> 00:24:17.190 to be in pretty good condition. 00:24:17.190 --> 00:24:19.800 They'd find it floating in water. 00:24:19.800 --> 00:24:23.675 And thus, what was a perfectly good engine now 00:24:23.675 --> 00:24:25.050 had to be sent back to the depot. 00:24:25.050 --> 00:24:27.360 So you can imagine the cost, the readiness issues, 00:24:27.360 --> 00:24:28.410 all of the implications. 00:24:28.410 --> 00:24:30.410 And that's assuming you found the container that 00:24:30.410 --> 00:24:31.770 had that engine. 00:24:31.770 --> 00:24:36.930 So with that, what we ended up doing is looking at an active 00:24:36.930 --> 00:24:41.070 RFID tag and looking at a tag that would monitor 00:24:41.070 --> 00:24:44.912 the container, which is the housing element for the asset 00:24:44.912 --> 00:24:47.370 that we were interested in-- not necessarily the container, 00:24:47.370 --> 00:24:48.870 but it was the asset-- 00:24:48.870 --> 00:24:50.742 and telling you where it was, what 00:24:50.742 --> 00:24:52.950 are the temperature conditions, what are the pressure 00:24:52.950 --> 00:24:56.250 conditions, was this container dropped, where has it been, 00:24:56.250 --> 00:24:58.770 and ultimately looking into, is it 00:24:58.770 --> 00:25:02.130 emitting a chemical when you get into Homeland Security issues. 00:25:02.130 --> 00:25:05.820 And so we developed a tag, looked at common standards, 00:25:05.820 --> 00:25:11.110 integrated those standards onto a tag. 00:25:11.110 --> 00:25:13.620 And when you hear the various talks, 00:25:13.620 --> 00:25:17.340 you're going to hear about a number of issues of power, 00:25:17.340 --> 00:25:24.420 range for RFID, durability. 00:25:24.420 --> 00:25:27.960 And so what we looked at, and some of the interesting things 00:25:27.960 --> 00:25:32.460 that capitalize, actually, from the computer science world, 00:25:32.460 --> 00:25:34.740 is how do we network these tags? 00:25:34.740 --> 00:25:37.140 If we want to keep the power low, 00:25:37.140 --> 00:25:39.660 that's going to have an implication about how 00:25:39.660 --> 00:25:41.490 much distance and range we get. 00:25:41.490 --> 00:25:43.650 Now, these are active tags. 00:25:43.650 --> 00:25:47.640 So with that, if we can implement kind of a hopping 00:25:47.640 --> 00:25:50.490 or-- and it's not really an ad hoc network, to say, 00:25:50.490 --> 00:25:52.320 but it looks like it-- 00:25:52.320 --> 00:25:54.270 where we can hop from tag to tag to tag 00:25:54.270 --> 00:25:57.510 to find the furthest tag away, without increasing power, 00:25:57.510 --> 00:26:02.010 without changing anything else, then we can form this network 00:26:02.010 --> 00:26:04.680 and get a map of where all our assets are. 00:26:04.680 --> 00:26:06.180 And so that's what we ended up doing 00:26:06.180 --> 00:26:08.580 to get around the power issue. 00:26:08.580 --> 00:26:11.002 One of my last slides, I'll get into future technologies, 00:26:11.002 --> 00:26:12.960 and some of the things that we need to consider 00:26:12.960 --> 00:26:19.307 are power scavenging, ways to get power from other means. 00:26:19.307 --> 00:26:21.390 Especially when you're dealing with active systems 00:26:21.390 --> 00:26:23.282 where the power is self-contained, 00:26:23.282 --> 00:26:24.990 you've got to look at that because that's 00:26:24.990 --> 00:26:27.240 one of the greatest technology hurdles. 00:26:27.240 --> 00:26:29.490 And nothing to do with RFID, but it's 00:26:29.490 --> 00:26:32.670 a technology that has lots of research areas. 00:26:32.670 --> 00:26:34.170 Contact memory buttons are something 00:26:34.170 --> 00:26:36.030 I pointed out early on. 00:26:36.030 --> 00:26:39.030 If you want to store asset information, maintenance 00:26:39.030 --> 00:26:42.073 history on an item, and be able to gather that 00:26:42.073 --> 00:26:43.740 without having to open up the container, 00:26:43.740 --> 00:26:47.670 find the paperwork, all of that can be integrated within your-- 00:26:47.670 --> 00:26:49.785 earlier there was mention about an architecture. 00:26:49.785 --> 00:26:51.660 Not only do you have a hardware architecture, 00:26:51.660 --> 00:26:54.090 but an information architecture that you're dealing with. 00:26:54.090 --> 00:26:57.120 We had a pilot study, and everything 00:26:57.120 --> 00:26:59.850 you can think of that could go wrong went wrong. 00:26:59.850 --> 00:27:05.130 But finding power out in the field is a major issue. 00:27:05.130 --> 00:27:08.460 Having the ability to connect up to a computer 00:27:08.460 --> 00:27:13.380 without having to get lots of permissions was a major issue. 00:27:13.380 --> 00:27:15.510 In this particular case, luckily we 00:27:15.510 --> 00:27:18.330 didn't have any other RF interference conditions. 00:27:18.330 --> 00:27:20.460 But in a warehouse environment, what 00:27:20.460 --> 00:27:23.070 happened in an installation of an RFID system, 00:27:23.070 --> 00:27:25.678 it shut down the entire wireless network 00:27:25.678 --> 00:27:26.970 because they were incompatible. 00:27:26.970 --> 00:27:30.100 So a lot of things that you've got to look at. 00:27:30.100 --> 00:27:34.380 It's not just a slap and chip and it's going to work. 00:27:34.380 --> 00:27:40.440 In our active case, if we had a forklift or something come 00:27:40.440 --> 00:27:45.630 in between the containers and the tags, and read 00:27:45.630 --> 00:27:46.890 rate stopped. 00:27:46.890 --> 00:27:48.750 What we're seeing on commercial data 00:27:48.750 --> 00:27:51.120 is that you'll have an item that goes 00:27:51.120 --> 00:27:55.558 from the back stock to the store, and then back again, 00:27:55.558 --> 00:27:56.350 and back and forth. 00:27:56.350 --> 00:27:58.980 Well, that's not happening. 00:27:58.980 --> 00:28:01.310 The accuracy that was discussed earlier 00:28:01.310 --> 00:28:03.570 is another critical issue. 00:28:03.570 --> 00:28:05.530 What do you do with the data? 00:28:05.530 --> 00:28:07.460 Extremely important. 00:28:07.460 --> 00:28:10.210 And if you're not going to use the data, then 00:28:10.210 --> 00:28:13.420 why bother tagging your assets? 00:28:13.420 --> 00:28:16.210 And so that's another element that, I think, 00:28:16.210 --> 00:28:19.060 is now getting on the forefront. 00:28:19.060 --> 00:28:20.800 And there are a lot of other, actually, 00:28:20.800 --> 00:28:24.190 side benefits that came about for this particular project. 00:28:24.190 --> 00:28:26.090 They were using brand new containers that 00:28:26.090 --> 00:28:29.980 were unpressurized, and so just the testing of the containers 00:28:29.980 --> 00:28:33.550 and making sure they're protecting the assets in itself 00:28:33.550 --> 00:28:36.280 was a side benefit that came out of that project. 00:28:36.280 --> 00:28:40.630 So a lot of things to consider that we've talked about. 00:28:40.630 --> 00:28:42.340 Tag and label issue, these things 00:28:42.340 --> 00:28:44.920 are being discussed by various standards committees. 00:28:44.920 --> 00:28:47.650 Please follow those standards. 00:28:47.650 --> 00:28:51.820 The guidelines have been thought through very carefully. 00:28:51.820 --> 00:28:55.060 A lot of parameters-- we talked about how far, how fast, 00:28:55.060 --> 00:28:59.410 how much data, how much content, memory. 00:28:59.410 --> 00:29:05.260 Security is another interesting component. 00:29:05.260 --> 00:29:08.140 And I never thought about somebody 00:29:08.140 --> 00:29:10.117 walking along the street with a reader 00:29:10.117 --> 00:29:12.700 and being able to decide which car they're going to break into 00:29:12.700 --> 00:29:16.870 based on the contents in the trunk until one of my students 00:29:16.870 --> 00:29:18.400 brought up that as an application 00:29:18.400 --> 00:29:20.840 or as a problem to solve. 00:29:20.840 --> 00:29:23.020 And so there are a lot of things. 00:29:23.020 --> 00:29:25.510 Privacy, of course, is another big one. 00:29:25.510 --> 00:29:31.570 And other considerations, we'll get into in various talks 00:29:31.570 --> 00:29:33.010 on collisions. 00:29:33.010 --> 00:29:38.560 Lots of lessons learned, site surveys, power. 00:29:38.560 --> 00:29:41.470 The information system is very critical. 00:29:41.470 --> 00:29:44.812 And where I think some of the future areas are-- 00:29:44.812 --> 00:29:46.270 and we've discussed these earlier-- 00:29:46.270 --> 00:29:47.645 include the various applications. 00:29:47.645 --> 00:29:49.120 The applications are endless-- 00:29:49.120 --> 00:29:52.640 nanotechnology, power sources, packaging, 00:29:52.640 --> 00:29:54.730 which will be a session tomorrow. 00:29:54.730 --> 00:29:56.980 How can we embed some of these things in the packaging 00:29:56.980 --> 00:29:58.900 that you're using to ship your assets? 00:29:58.900 --> 00:30:02.320 And a term that we're using of performance based logistics, 00:30:02.320 --> 00:30:05.920 or logistical prognostics, taking 00:30:05.920 --> 00:30:08.860 algorithms that we use for predicting failures 00:30:08.860 --> 00:30:11.620 in equipment, and look at them for predicting failures 00:30:11.620 --> 00:30:12.700 in a logistics pipeline. 00:30:12.700 --> 00:30:15.610 These are all things to consider and take a look at. 00:30:15.610 --> 00:30:20.995 And lots of work, as Sanjay indicated, [INAUDIBLE] 00:30:24.488 --> 00:30:26.983 [APPLAUSE] 00:30:32.990 --> 00:30:36.800 STEVE MILES: And then Manfred Aigner, 00:30:36.800 --> 00:30:40.610 who's the group coordinator for the VLSI and security 00:30:40.610 --> 00:30:44.630 group at PROACT at the University of TU-Graz 00:30:44.630 --> 00:30:48.020 that has a joint research project with Phillips. 00:30:48.020 --> 00:30:50.000 And it's actually [? Ari Bachtel, ?] 00:30:50.000 --> 00:30:52.760 who's the head of EPC Global Europe, who suggested that we 00:30:52.760 --> 00:30:57.470 might want to consider some of the European experience 00:30:57.470 --> 00:31:00.500 with encryption in the smart card industry 00:31:00.500 --> 00:31:02.027 as it might relate to RFID. 00:31:02.027 --> 00:31:03.110 MANFRED AIGNER: Thank you. 00:31:10.380 --> 00:31:12.838 Thank you for the introduction. 00:31:12.838 --> 00:31:14.880 So it's a pleasure for me to present our results. 00:31:14.880 --> 00:31:19.170 We are involved in research for security tag, reader security 00:31:19.170 --> 00:31:22.770 on our RFID in, let's say, two or three years, 00:31:22.770 --> 00:31:27.180 involving from smart card implementations of crypto 00:31:27.180 --> 00:31:29.220 modules, [? NT ?] stuff. 00:31:29.220 --> 00:31:31.810 And I'm happy to show you our results. 00:31:31.810 --> 00:31:34.440 I will talk a little bit of our group, what we are doing, 00:31:34.440 --> 00:31:38.040 and tell you the requirements we defined for our developments, 00:31:38.040 --> 00:31:41.430 and tell you the problems you face as a developer of crypto 00:31:41.430 --> 00:31:47.640 modules for tag application, and show you some of our results 00:31:47.640 --> 00:31:49.450 we achieved so far. 00:31:49.450 --> 00:31:52.050 So we are a group of about 50 people doing 00:31:52.050 --> 00:31:56.740 research on IT security, from development 00:31:56.740 --> 00:31:58.140 of crypto algorithms-- 00:31:58.140 --> 00:32:01.470 Vincent Rijmen, the inventor of the Advanced Encryption 00:32:01.470 --> 00:32:03.210 Standard, is with us-- 00:32:03.210 --> 00:32:06.660 up to e-government applications. 00:32:06.660 --> 00:32:09.510 And I am the group leader of the group 00:32:09.510 --> 00:32:13.330 that is doing VLSI, so hardware implementations for crypto. 00:32:13.330 --> 00:32:16.170 We are doing implementations for smart cards, 00:32:16.170 --> 00:32:19.740 for embedded systems, accelerator 00:32:19.740 --> 00:32:22.440 cards for encrypted networks, [? NT ?] stuff. 00:32:22.440 --> 00:32:27.120 We are also doing a lot of system and chip design. 00:32:27.120 --> 00:32:30.670 Our major activity at the moment is side-channel analysis, 00:32:30.670 --> 00:32:33.130 which is a major topic in smart card industry. 00:32:33.130 --> 00:32:37.650 So there are attacks where you measure the current 00:32:37.650 --> 00:32:41.460 and try to get some secret about the key 00:32:41.460 --> 00:32:42.810 from the current measurements. 00:32:42.810 --> 00:32:47.640 This will also be a topic for RFID tags. 00:32:47.640 --> 00:32:53.220 And we are doing projects with quantum groups. 00:32:53.220 --> 00:32:55.740 And yeah, this is-- 00:32:55.740 --> 00:32:57.810 what makes us so special is probably 00:32:57.810 --> 00:33:00.030 the strong interaction with the other groups 00:33:00.030 --> 00:33:01.410 we have at our institute. 00:33:01.410 --> 00:33:03.840 Especially when developing AES, it 00:33:03.840 --> 00:33:07.140 was very helpful to have Vincent Rijmen sitting with us together 00:33:07.140 --> 00:33:10.650 to find out how we can serialize the algorithm 00:33:10.650 --> 00:33:14.400 to comply with the requirements we have in RFID. 00:33:14.400 --> 00:33:17.190 So we talked a lot about that already. 00:33:17.190 --> 00:33:20.340 I will skip that slide because Sanjay today in the morning 00:33:20.340 --> 00:33:21.250 all explained this. 00:33:21.250 --> 00:33:23.880 So there is actually a need for security. 00:33:23.880 --> 00:33:28.080 And we say if you put security on a tag or on RFID, 00:33:28.080 --> 00:33:31.150 you should use proper security, so lightweight security. 00:33:31.150 --> 00:33:34.770 In our sense, it's not lightweight crypto, 00:33:34.770 --> 00:33:38.160 so lightweight implementation of real crypto, 00:33:38.160 --> 00:33:42.990 because if you try to get in with security 00:33:42.990 --> 00:33:46.740 in a globalized technology, you will need standardization. 00:33:46.740 --> 00:33:51.250 And that prevents, of course, secrets in the algorithm. 00:33:51.250 --> 00:33:57.330 So the standardized algorithms like AES, RSA, ECC, 00:33:57.330 --> 00:33:58.680 they are approved by experts. 00:33:58.680 --> 00:34:02.100 So there is some work spent in investigations 00:34:02.100 --> 00:34:03.070 if they are secure. 00:34:03.070 --> 00:34:04.740 And if you do not use them, you're 00:34:04.740 --> 00:34:07.650 probable to flaws in your systems. 00:34:07.650 --> 00:34:10.469 And the thing is you should not only 00:34:10.469 --> 00:34:12.750 look at the algorithms you use. 00:34:12.750 --> 00:34:15.630 Most [INAUDIBLE] systems get broken because they do not 00:34:15.630 --> 00:34:20.520 use standardized protocols, so most flaws in systems 00:34:20.520 --> 00:34:24.260 are in protocols. 00:34:24.260 --> 00:34:27.330 And now I want to bring some arguments 00:34:27.330 --> 00:34:30.070 against and for standardized algorithms. 00:34:30.070 --> 00:34:34.136 So some people say that if you use standardized algorithms, 00:34:34.136 --> 00:34:35.969 they are easy to attack because there are so 00:34:35.969 --> 00:34:38.190 many publications on attacks. 00:34:38.190 --> 00:34:41.760 And you do not take into account that your developers 00:34:41.760 --> 00:34:42.600 of the system. 00:34:42.600 --> 00:34:44.270 And if you're talking about RFID, 00:34:44.270 --> 00:34:45.909 there are a lot of developers. 00:34:45.909 --> 00:34:48.719 They are potentially attackers. 00:34:48.719 --> 00:34:53.460 The secret is not a secret if so many people got involved. 00:34:53.460 --> 00:34:55.540 And some people say that especially 00:34:55.540 --> 00:35:00.600 side-channel analysis is more dangerous if the algorithms are 00:35:00.600 --> 00:35:08.640 known and if the RFID tag are in the hands of the attackers, 00:35:08.640 --> 00:35:10.910 and they are in the hands of the attackers, 00:35:10.910 --> 00:35:12.990 I would apply a side-channel attack. 00:35:12.990 --> 00:35:15.030 And if you understand side-channel attacks, 00:35:15.030 --> 00:35:17.370 it's easy to adopt them to other algorithms. 00:35:17.370 --> 00:35:19.410 It's also easier to use them to find out 00:35:19.410 --> 00:35:21.850 the specifics of secret algorithms. 00:35:21.850 --> 00:35:24.990 So that's not a question. 00:35:24.990 --> 00:35:28.860 And some people say that custom-built algorithms 00:35:28.860 --> 00:35:30.360 use less secure. 00:35:30.360 --> 00:35:33.960 You lose less resources. 00:35:33.960 --> 00:35:37.890 And I say that they are potentially less secure. 00:35:37.890 --> 00:35:39.900 If you use them, things might happen. 00:35:39.900 --> 00:35:41.790 Like, it was exactly last year, I 00:35:41.790 --> 00:35:46.110 think, this break of the John Hopkins [INAUDIBLE] 00:35:46.110 --> 00:35:49.650 of the [? speed pass. ?] And we say 00:35:49.650 --> 00:35:52.740 that proper implementations of standardized algorithms 00:35:52.740 --> 00:35:57.120 are possible for passive devices, for RFID tags. 00:35:57.120 --> 00:36:00.370 And I will show you later on our modules. 00:36:00.370 --> 00:36:03.160 So when we started, we defined some requirements 00:36:03.160 --> 00:36:05.290 for secure tags we wanted to develop. 00:36:05.290 --> 00:36:07.840 And we said right from the start we do not 00:36:07.840 --> 00:36:09.460 want lightweight crypto. 00:36:09.460 --> 00:36:13.510 We want real crypto, like the same standard as in smart card 00:36:13.510 --> 00:36:14.650 industry. 00:36:14.650 --> 00:36:18.820 We wanted to use standardized algorithms, 00:36:18.820 --> 00:36:21.310 because the high number of tags we will have 00:36:21.310 --> 00:36:24.790 is enough value that even if each tag just 00:36:24.790 --> 00:36:28.840 protects small value, it's a good point of attack 00:36:28.840 --> 00:36:32.770 for an attacker because the high number of tags 00:36:32.770 --> 00:36:36.070 makes a hack very interesting. 00:36:36.070 --> 00:36:38.200 But the high number of tags logically 00:36:38.200 --> 00:36:40.820 needs a very clever key management. 00:36:40.820 --> 00:36:44.320 So we do not have systems like that so far. 00:36:44.320 --> 00:36:49.090 We didn't want to accept a raise of costs, a significant raise 00:36:49.090 --> 00:36:52.940 of costs, due to our security implementations 00:36:52.940 --> 00:36:54.850 we wanted to have on the tags. 00:36:54.850 --> 00:36:59.380 And we didn't want to accept a reduction in reading distance. 00:36:59.380 --> 00:37:03.520 So we had to comply with this very small power consumption. 00:37:03.520 --> 00:37:07.680 And we wanted to be compatible with our already installed 00:37:07.680 --> 00:37:08.560 infrastructures. 00:37:08.560 --> 00:37:11.560 So we didn't want to suggest a system which is never 00:37:11.560 --> 00:37:14.770 accepted because there are so many readers already out there. 00:37:14.770 --> 00:37:17.820 And there we started, and then we 00:37:17.820 --> 00:37:20.840 were facing a lot of limitations due to the technology. 00:37:20.840 --> 00:37:24.080 So the main problem is the power consumption. 00:37:24.080 --> 00:37:28.180 So if we do not accept a reduction in the reading 00:37:28.180 --> 00:37:30.700 distance, we were facing that we have 00:37:30.700 --> 00:37:34.000 about 10 microamps available. 00:37:34.000 --> 00:37:36.850 The area consumption, which is less problem 00:37:36.850 --> 00:37:39.670 with new semiconductor technology, 00:37:39.670 --> 00:37:43.450 but it's still the technology that's now available. 00:37:43.450 --> 00:37:45.530 We have very limited execution time, 00:37:45.530 --> 00:37:49.510 in fact, because, well, we have a rather limited clock 00:37:49.510 --> 00:37:51.070 frequency. 00:37:51.070 --> 00:37:54.335 So the protocols, that was actually 00:37:54.335 --> 00:37:55.960 a problem, because in the other things, 00:37:55.960 --> 00:38:00.492 you can say just a client, and the server client just responds 00:38:00.492 --> 00:38:01.450 or something like that. 00:38:01.450 --> 00:38:05.310 But here always the protocol is always initiated by the reader, 00:38:05.310 --> 00:38:07.680 and that's different. 00:38:07.680 --> 00:38:13.390 And there is no physical protection available possible. 00:38:13.390 --> 00:38:15.702 And then there were a lot of publications 00:38:15.702 --> 00:38:17.160 in the last year where people state 00:38:17.160 --> 00:38:22.110 that hash is more inexpensive that the encryption, and that's 00:38:22.110 --> 00:38:23.350 simply not true. 00:38:23.350 --> 00:38:26.160 So I do not say that this research is useless, 00:38:26.160 --> 00:38:29.910 but they should use encryption modules instead of hash models 00:38:29.910 --> 00:38:35.400 because they are easier to implement in hardware. 00:38:35.400 --> 00:38:38.960 So what are the enhancements we want to propose? 00:38:38.960 --> 00:38:41.570 For tag identification, we will need 00:38:41.570 --> 00:38:43.970 an extended personalization. 00:38:43.970 --> 00:38:47.300 We will need a crypto, primitive, on a tag. 00:38:47.300 --> 00:38:49.610 And we will need to secure key storage 00:38:49.610 --> 00:38:54.840 on a tag, which is a problem that is not treated so far. 00:38:54.840 --> 00:38:57.080 And we need the cryptocapability of the reader 00:38:57.080 --> 00:39:02.120 or a secure access to a verification server. 00:39:02.120 --> 00:39:05.720 For read authentication, we will need one additional thing. 00:39:05.720 --> 00:39:07.370 That's nonce generation. 00:39:07.370 --> 00:39:09.590 Nonce a number used once. 00:39:09.590 --> 00:39:11.870 It's kind of pseudo-random number. 00:39:11.870 --> 00:39:13.970 It has to be fresh and unpredictable, 00:39:13.970 --> 00:39:19.100 and that's not an easy task for RFID tags. 00:39:19.100 --> 00:39:20.480 What are the results so far? 00:39:20.480 --> 00:39:25.400 In our institute, we developed an AES module, 00:39:25.400 --> 00:39:28.620 which complies with the requirements for a tag. 00:39:28.620 --> 00:39:33.230 So it uses 4.5 microwatts. 00:39:33.230 --> 00:39:33.920 That's produced. 00:39:33.920 --> 00:39:34.700 That's verified. 00:39:34.700 --> 00:39:36.035 That's available. 00:39:36.035 --> 00:39:37.910 The only thing is there is no countermeasures 00:39:37.910 --> 00:39:40.700 against side-channel attacks on this module. 00:39:40.700 --> 00:39:42.710 That's what we are working on so far. 00:39:42.710 --> 00:39:46.850 We presented security layers for ISO 18000. 00:39:46.850 --> 00:39:48.800 The reason why we've chosen this standard 00:39:48.800 --> 00:39:53.180 is because we were used to the ISO standards from smart card, 00:39:53.180 --> 00:39:56.120 and it was, when we started, clearer to us 00:39:56.120 --> 00:39:59.840 the concept of defining new comments like this, 00:39:59.840 --> 00:40:02.400 custom comments, which were foreseen in ISO 18000. 00:40:02.400 --> 00:40:04.260 In fact, it doesn't make a big difference 00:40:04.260 --> 00:40:07.670 if we use EPC or ISO. 00:40:07.670 --> 00:40:10.700 We have protocol security layers for anti-cloning, 00:40:10.700 --> 00:40:12.080 for privacy enhancement. 00:40:12.080 --> 00:40:18.350 And this is tested with a tag emulator and FPGA basis. 00:40:18.350 --> 00:40:21.080 And we have isometric crypto modules 00:40:21.080 --> 00:40:24.320 using elliptic curve cryptography, which 00:40:24.320 --> 00:40:28.510 are usable on passive devices. 00:40:28.510 --> 00:40:30.990 So what are our future task? 00:40:30.990 --> 00:40:32.460 We will work on-- 00:40:32.460 --> 00:40:37.620 we should work, actually, on key management and personalization, 00:40:37.620 --> 00:40:39.480 on testability of crypto tags. 00:40:39.480 --> 00:40:40.800 Nobody mentioned that so far. 00:40:40.800 --> 00:40:42.280 This will be a problem in future. 00:40:42.280 --> 00:40:44.520 So if there is more functionality in tags, 00:40:44.520 --> 00:40:45.630 how will you test them? 00:40:45.630 --> 00:40:47.580 Compliance testing is already a topic, 00:40:47.580 --> 00:40:51.150 but this will be a more interesting topic. 00:40:51.150 --> 00:40:53.760 We have to deal with this nonce generational tag 00:40:53.760 --> 00:40:59.100 and further research on isometric crypto. 00:40:59.100 --> 00:41:01.260 So what are the conclusions? 00:41:01.260 --> 00:41:05.640 Use standardized crypto if you state that you will. 00:41:05.640 --> 00:41:09.120 Design secure RFID systems, because you never 00:41:09.120 --> 00:41:11.580 know what your system is used for then. 00:41:15.550 --> 00:41:21.800 The protection will be also necessary in inexpensive tags 00:41:21.800 --> 00:41:25.640 in future because you never know the applications, also 00:41:25.640 --> 00:41:29.300 protection against side-channel analysis. 00:41:29.300 --> 00:41:33.230 And, well, according our results so far, 00:41:33.230 --> 00:41:35.870 I would say the implementation of standardized crypto 00:41:35.870 --> 00:41:38.810 is possible on passive devices if you 00:41:38.810 --> 00:41:41.480 go for a clever implementation. 00:41:41.480 --> 00:41:46.250 And more research is definitely necessary for integration 00:41:46.250 --> 00:41:49.640 into running applications, to future applications. 00:41:49.640 --> 00:41:51.900 This is a list of our recent papers. 00:41:51.900 --> 00:41:55.700 And I just want to mention our initiative, PROACT. 00:41:55.700 --> 00:42:01.160 We are looking for research, for professorship, 00:42:01.160 --> 00:42:04.760 and for visiting professors, and stronger interaction 00:42:04.760 --> 00:42:07.910 with the RFID community in the next years. 00:42:07.910 --> 00:42:10.196 And thank you. 00:42:10.196 --> 00:42:15.680 [APPLAUSE] 00:42:15.680 --> 00:42:18.270 STEVE MILES: Any questions for this distinguished panel 00:42:18.270 --> 00:42:22.110 around just the tags and the future of tags? 00:42:22.110 --> 00:42:24.090 If we could ask you to come down to the mics. 00:42:33.692 --> 00:42:35.150 AUDIENCE: This is Chris [INAUDIBLE] 00:42:35.150 --> 00:42:36.880 from the Auto-ID Lab in Switzerland. 00:42:36.880 --> 00:42:39.230 I have just a quick question regarding your work 00:42:39.230 --> 00:42:41.190 on power consumption reduction. 00:42:41.190 --> 00:42:46.220 If you look at the works, say, of [INAUDIBLE] on transponders, 00:42:46.220 --> 00:42:50.100 their transponders needed about 16 microwatts at the antenna 00:42:50.100 --> 00:42:52.070 and probably about 4 microwatts on the chip 00:42:52.070 --> 00:42:53.240 before the rectifier. 00:42:53.240 --> 00:42:55.770 Where do you think your work will take this? 00:42:55.770 --> 00:42:57.395 Do you think you can get significantly, 00:42:57.395 --> 00:43:01.250 be 1 microwatt, for power consumption of a tag for read 00:43:01.250 --> 00:43:03.950 access? 00:43:03.950 --> 00:43:06.485 HAO MIN: Actually, your question is-- 00:43:06.485 --> 00:43:11.090 you will see that the minimum power for the tags, there's 00:43:11.090 --> 00:43:11.930 IF power. 00:43:11.930 --> 00:43:14.090 Say it's, like, 16 microwatts. 00:43:14.090 --> 00:43:18.170 But the real digital power consumption is, like, a 4 watt. 00:43:18.170 --> 00:43:21.910 So this means that there's only 20% of the microwave power 00:43:21.910 --> 00:43:23.970 was converted into DC power. 00:43:23.970 --> 00:43:29.780 So our work is-- we can get the performance better by doing 00:43:29.780 --> 00:43:30.450 two things. 00:43:30.450 --> 00:43:33.320 One is we can improve the conversion rate, which 00:43:33.320 --> 00:43:38.000 is higher efficiency to convert the microwave 00:43:38.000 --> 00:43:39.740 power into DC power. 00:43:39.740 --> 00:43:42.180 The second thing is that by carefully designing 00:43:42.180 --> 00:43:44.970 so the digital circuitry would reduce the power consumption. 00:43:44.970 --> 00:43:49.250 So this is a two-way, where each go together to reduce the power 00:43:49.250 --> 00:43:50.090 consumption. 00:43:50.090 --> 00:43:55.940 But eventually, what decides the read range is the IF power, 00:43:55.940 --> 00:43:57.687 decides the whole read range. 00:43:57.687 --> 00:43:59.770 AUDIENCE: But what do you think this is moving to? 00:43:59.770 --> 00:44:02.780 We you see that at some point we will be down 00:44:02.780 --> 00:44:05.150 to, like, 1 microwatts? 00:44:05.150 --> 00:44:08.350 HAO MIN: I think probably even less than 1 microwatts. 00:44:08.350 --> 00:44:09.310 Yeah. 00:44:09.310 --> 00:44:12.250 Just like in [INAUDIBLE] stations, 00:44:12.250 --> 00:44:14.080 if we use some special circuitry, 00:44:14.080 --> 00:44:16.870 we even can reduce the power consumption by a factor of 10, 00:44:16.870 --> 00:44:18.310 even 100. 00:44:18.310 --> 00:44:21.370 We take advantage of that the tag really work in 00:44:21.370 --> 00:44:23.720 really low operating speed. 00:44:23.720 --> 00:44:25.030 So for the maximal-- 00:44:25.030 --> 00:44:31.150 the clock in a gen 2, it's only a 640k, which is much-- 00:44:31.150 --> 00:44:35.710 even 1,000 times slower than a PC is working. 00:44:35.710 --> 00:44:39.850 So probably, it's just can use some special circuitry. 00:44:39.850 --> 00:44:43.120 You can reduce the power of that to, like, a 100 times. 00:44:43.120 --> 00:44:45.600 Then [INAUDIBLE]. 00:44:45.600 --> 00:44:46.200 AUDIENCE: OK. 00:44:46.200 --> 00:44:46.908 Thanks very much. 00:44:46.908 --> 00:44:48.830 HAO MIN: OK. 00:44:48.830 --> 00:44:50.330 STEVE MILES: So thank you very much. 00:44:50.330 --> 00:44:53.210 If we could ask the next panel to come up, 00:44:53.210 --> 00:44:55.450 in the interest of time.