Harrison_Tang: Hi everyone uh welcome back uh to this week's w3c shii meeting so uh we uh they didn't have 1 yet uh last week because of the peak uh but I will resume our regular uh content program um starting this week so this week we're very excited to have your saifi and Martin uh present on a NGS uh signing engine s i l d and also authentication of NGS um today and uh. ✪
Harrison_Tang: And uh before before that just want to quickly go through some administrative stuff uh first of all just a quick reminder on the code of ethics and professional conduct I just want to make sure that we hold a respectful and constructive conversations here we've been doing that but that's continued to do that. ✪
Harrison_Tang: A quick note on the intellectual property anyone can participate in these calls however all substantial contributions to any ccg work items must be member of the ccg with full IPR agreement signed so if you have any questions in regards to the agreement or the w3c account uh feel free to uh reach out to any of the ccg cultures. ✪
Harrison_Tang: A quick call note these are meetings uh are automatically recorded and transcribed and we will publish the meeting minutes audio recordings and video recordings in the next day or 2. ✪
Harrison_Tang: We use a GC chat to cue the speaker so you can type in Q Plus to add yourself to the queue or cue minus and I will be moderating the Q and discussions. ✪
Harrison_Tang: All right just want to take a quick moment moment for the introductions and reintroductions so if you're new to the community or you haven't been active and want to re-engage uh feel free to just unmute and introduce yourself. ✪
Harrison_Tang: Actually see mostly familiar faces so move on to the next agenda um announcements and reminders so any announcements and or reminders. ✪
Harrison_Tang: I know that some of us were in TPAC last week so any announcements from that. ✪
Harrison_Tang: I actually joined 1 of The Sessions but uh there's no particular announcement uh relevant to ccg so. ✪
Harrison_Tang: All right what about updates to work items. ✪
Harrison_Tang: So we'll actually have different uh. ✪
Harrison_Tang: I will hold another quarter for review and work item updates on November 19th so just. ✪
Harrison_Tang: I'll give you a heads up for what's coming. ✪
Harrison_Tang: All right uh so last calls for introductions reintroductions announcements reminders and or work items. ✪
Harrison_Tang: All right that's quick um so just uh want to get to a main agenda so uh this week uh again very excited to have Jose and Martin here uh to present on the NGS ild uh signing NGS ild and authentication in NGS silv by the way uh NGS I stands for New Generation servers interface I think um but they can go a little bit deeper into uh NGS NGS I and also how it actually uh intersects with identity problems that uh ccg uh really uh cares about so. ✪
Martin_Bauer_(ETSI_ISG_CIM): I I'll give you a bit of an introduction um I think when when we are at the right place so. ✪
Harrison_Tang: Great great all right so without further Ado uh I think the floor is yours uh you say pay and Martin. ✪
Martin_Bauer_(ETSI_ISG_CIM): Okay should I share try to share first um okay. ✪
Martin_Bauer_(ETSI_ISG_CIM): Okay have to the right screen. ✪
Martin_Bauer_(ETSI_ISG_CIM): Okay it's not quite yet but almost there. ✪
Martin_Bauer_(ETSI_ISG_CIM): You see my my screen. ✪
Martin_Bauer_(ETSI_ISG_CIM): Jepa and me we're both um Vice chairs of etsy is G Sim. ✪
Martin_Bauer_(ETSI_ISG_CIM): I'm a principal standardization engineer at um NEC um NC Laboratories from Europe we are in Heidelberg in Germany and we are the European lab of NEC the worldwide active um japan-based um. ✪
Martin_Bauer_(ETSI_ISG_CIM): Um what I will do is try to give you a bit of an introduction to NG's LD so you have the general scope and then um just separate will focus on the core topics that we wanted to present and discuss on with you regarding signing and um Authentication. ✪
Martin_Bauer_(ETSI_ISG_CIM): So um first um to start with um who are we and and what do we do so we work in Etsy for that standardization so that's the European telecommunications standards Institute. ✪
Martin_Bauer_(ETSI_ISG_CIM): It's the producers globally applicable standards for information and Communications Technologies it's 1 of the officially recognized European standards organizations the others being sent and selac and we are an industry specification group the special thing about isgs is that they also allow the participation of non um Etsy members. ✪
Martin_Bauer_(ETSI_ISG_CIM): From isg SIM where Sim stands for context information management was established in 2017. ✪
Martin_Bauer_(ETSI_ISG_CIM): Isgs are always on for 2 years um so the current extension is until 2025. ✪
Martin_Bauer_(ETSI_ISG_CIM): Ask for another extension I'm quite soon have currently about 30 support organization and um our main um activity is the evolution of the NGS I contact interfaces to NGS and that's also a bit of the background and where this some strange acronym comes from um the story didn't start in Etsy but in MMA in before 2010 where omma was defining a sweet um of um different interfaces and um 2 of these um were the context um API then called ngi 9 and 10 so they were numbered it's not a version number it's just a number of that interface with respect to that um sweet then um this was taken up by Fireware time was a large project um in Europe On future internet and that um further developed on that um and um defined bindings um which ended up being um ngci version 1 and 2 and um Fireware became from. ✪
Martin_Bauer_(ETSI_ISG_CIM): An open source Community um and um then it was decided okay as this is the key interface really um this should go back um into standardization um and um that's what what we're doing now in Etsy um what's called um NG's ild. ✪
Martin_Bauer_(ETSI_ISG_CIM): So high level goals um it was the evolution of these um um Fireware in ghee context interfaces and we wanted to put the NG's LD information model on a solid conceptual grounding so wanted to apply the property graph model enable semantic concept definitions enable linking to existing information so I would call linked data and the idea of the API is to give applications relatively high level of abstraction enable them to specify what information um they require based on this um underlying information model including Geographic scoping and the temporal um API and also to support Central as well as distributed in Federated um architecture where information can be arbitrarily distributed and we'll see why that's important um also for our discussion later on. ✪
Martin_Bauer_(ETSI_ISG_CIM): So let's um start with the NGS information model um basically the Assumption of the world of NGS LDS that the world consists of entities entities um have a type and an identifier so you see here um car on the shop building person um and um a unique identifier in each cases all of these entities can have properties some like the speed um location or address. ✪
Martin_Bauer_(ETSI_ISG_CIM): Relationships basically so the person that owns the shop um for example on the camera that's attached to the pole and and and so on So based on having these um relationships some explicitly um we we in the end um have have a graph um that's um the underlying um property graph if we talk about properties in relationships together we also call that attribute. ✪
Martin_Bauer_(ETSI_ISG_CIM): And um not all things um um are suitable basically to be stored in in in this way and um things so if you think of um camera video stream basically um or maybe a complex 3D model of a building um this graph isn't um suitable but what you can do is basically link and also provide the right meta information for applications to find this maybe find this camera and then um there would be able to. ✪
Martin_Bauer_(ETSI_ISG_CIM): That um stream basically so you have with this graph also kind of an index structure can follow on relationship um and retrieve um um graph elements um from that. ✪
Martin_Bauer_(ETSI_ISG_CIM): Um the LD in NG's LD comes from Json LD because we decided um to use um not plain Json which was used before but um Json LD and um 1 thing you can do is basically um you you have um your concept um you can in your instance um use um nice um short names that are human readable and um in short but um with a predecessor you had basically the problem okay you use these short names but um then maybe somebody else used the same ones um but in a different way um so they were not unique um so what um um um Json LD um does is um it says um it defines some these URI and it defines the mapping so you define your mapping you can then still use um your short names um in your instances but um you also give the mapping basically um and then um it means in my document um um it means um this um unique um Yuri. ✪
Martin_Bauer_(ETSI_ISG_CIM): So I have a bit of The Best of Both Worlds um I can still use in my document um this um short names and it's um it's um relatively easy but um I have still unique um your eyes behind it. ✪
Martin_Bauer_(ETSI_ISG_CIM): Let's look at a typical um representation of NGS LD so here I have an entity which is a vehicle has an ID Json LD that needs to be a URI it has a property the brand name um with the value Mercedes which happened to be in an accident um so it has a relationship to an object which is that lamp post um and um that was observed at a certain time and provided so you can um put um meta information um in there and what you need to attach um is this um at context so that can either be directly attached to to the data or um you can in a header put um the the URL basically where you can download from that um context so for communication. ✪
Martin_Bauer_(ETSI_ISG_CIM): And um what's important is basically um you have always um entered we are always talking about entities. ✪
Martin_Bauer_(ETSI_ISG_CIM): And in the extreme case um you you can handle these um attributes um for these um things separately so for the same vehicle you may have in 1 place um stored the ID type in brand name and somewhere else on the ID um type and in accident and when retrieving um them you want to aggregate um this so that may be relevant um later on in the discussion. ✪
Martin_Bauer_(ETSI_ISG_CIM): I will not go very deeply um into the um API discussion but the kind of things you can do is um if you know um your entity you can ask a question like what is the location of Sam Sam being my entity location being okay go property um what do I need to know I need to know basically the base URL I need to know the entity ID and from the data model I need to know okay there's a location on property. ✪
Martin_Bauer_(ETSI_ISG_CIM): Probably um security credentials but that's orthogonal to the core API specification I don't need to know more complex some structures and so on where information may be stored. ✪
Martin_Bauer_(ETSI_ISG_CIM): May be Mr Geographic query so I can um ask on which costs are within a geographic um area so here I need to know the type that the type car exists um and um also how to specify that geographic location I have a place um to go and ask that and as an application that's um I need to know um to retrieve this um information. ✪
Martin_Bauer_(ETSI_ISG_CIM): From an architecture perspective um this um would be the kind of um Central um deployment um we often have right now um is you have a central broker that stores all the information you have context produces that great update delete the information here and you have context consumers that query subscripts query all that subscribe and get notifications when there are changes so we can also do temporal um based so get notification every um 10 minutes or something like that. ✪
Martin_Bauer_(ETSI_ISG_CIM): Um but we may also have um distributed um architectures where information is stored in context um sources so they also um implement the query or And subscribe notify interface and so the broker that the broker knows um how to find that information they can register what type of information they have um that can be um specific I have for this entity um these attributes or it can be a type based registration I have cards maybe in this kind of area so when I get a request the broker checks okay um which sources to have that information maybe it also um internally has um its own information it will go fetch these things aggregate these things and then return them um to the context um consumer. ✪
Martin_Bauer_(ETSI_ISG_CIM): So um in context sources um as they have the same interface can be for Brokers again so on that basis um you can build a hierarchy of of of brokers. ✪
Martin_Bauer_(ETSI_ISG_CIM): Or even a different kind of um things it doesn't have to be an explicit hierarchy. ✪
Martin_Bauer_(ETSI_ISG_CIM): Okay so that was basically the general introductory part I hope I didn't take too much time for it are there any immediate clarification questions. ✪
Harrison_Tang: If you're using a Mac you might need to change the permissions. ✪
Giuseppe_Tropea: Uh because this is the gcd okay but you should ask okay so probably is the silently failing on me yeah you're right you usually usually yes I need to change the uh Martin would you be so kind to to continue. ✪
Giuseppe_Tropea: Ing and then it's easier probably thanks a lot. ✪
Martin_Bauer_(ETSI_ISG_CIM): I I tried to continue sharing then yes okay. ✪
Our Robot Overlords are scribing.
Giuseppe_Tropea: Uh some years I guess and uh at some point we um we created uh what's in Etsy parlance is a specialist task force for implementing data Integrity in the standard so we we came up with a with a solution um okay of course starting from uh 1 uh main crucial assumption just as Martin has a highlighted um are information creators are what we call the context providers and um and it is pass. ✪
Giuseppe_Tropea: In possibly quite a complex network of Brokers before being related to clients so basically of course the hypothesis here that we won't end to end uh Integrity uh the only uh and the only uh person the only um thing that uh is trusted by clients of course is the context provider we we cannot assume that clients trust the intermediate brokers. ✪
Giuseppe_Tropea: Uh Martin I think starting from these assumptions we came with the list of basically 4 main requirements. ✪
Giuseppe_Tropea: Um this 1 is fairly obvious um the second 1 uh verification of Integrity shall be independent of syntax syntax reordering that may occur when as I said sterilizing entities that's passed among among peers um 1 of course 1 1 attribute may be serialized by 1 Brokerage in the client um also we have several NGS we have several different output formats so um once we attach uh verification information of of course we share that this is stripped out when it passes through through the network. ✪
Giuseppe_Tropea: Also as I said of course clients would not want to trust intermediate context progress so it's fairly obvious set set of requirements. ✪
Giuseppe_Tropea: Next slide so the solution we came up with uh which is implemented in in this specification there GS Sim uh 019. ✪
Giuseppe_Tropea: Um realize on mostly on the data Integrity work uh that you guys of course are very familiar with um third reference is uh exactly the API specification that Martin was mentioning beforehand. ✪
Giuseppe_Tropea: Uh just uh a quick note so this GS uh Sim 19 is as first release but an update is is in preparation. ✪
Giuseppe_Tropea: Mainly uh to align uh with the updated data Integrity versions that in the meanwhile uh have published. ✪
Giuseppe_Tropea: Okay so let me introduce 2 key terms uh the home entity in the sealed attribute and and the derivation process that takes us from the atomic entity to the sealed attribute um as you have seen uh an entity comprises generally an ID and an array of types at top level and then a bunch of attributes and or relationships. ✪
Giuseppe_Tropea: An atomic entity um an entity that comprises only 1 attribute and I'll come to this in a second um then from the atomic entity we derive a sealed attribute which is data attribute together with with a complaining cryptographic proof and the process to transform 1 into the other uh next slide so concerning the. ✪
Giuseppe_Tropea: Atomic entity um as I said it must be uh containing just 1 attribute um access slide we we we try to clarify. ✪
Giuseppe_Tropea: You can jump to the next 1 Martin. ✪
Giuseppe_Tropea: Yes so as I said um multiple entity aggregation steps can uh can happen because if you are the client and you ask for I don't know car 1 2 3 then maybe broker a uh contained information about its color and broke a big containing information about it I don't know injection date so uh basically when all this info come to the client they are or or better say to the aggregation broker or distribution broker I'll call it um this is aggregated into 1 entity uh of course also entities attributes can be filtered out because because of querying or. ✪
Giuseppe_Tropea: Because uh the entities uh travel throughout the chain of Brokers so the point is that we cannot take the whole uh NGS document representing the entity with all of its attributes and and sign all of them to get into 1 uh big sealed package this is why we have to um let's say uh sign them 1 by 1 in. ✪
Giuseppe_Tropea: In in what we call the atomic entities um okay let's go to the other 1. ✪
Giuseppe_Tropea: So the important Point here to understand is though that um when we sign for instance the color attribute being it red and we want this to be tamper evident um we also need to repackage within the color attribute also the entity ID and and the types uh that this color was referred to so basically what we do in in when we move from the atomic entity to the seal attribute is that we move this information within the proof we bound it together with the proof so we sign the color red together with information that colors red in entity I don't know car 1 2 3 that was of type car so this is basically our. ✪
Giuseppe_Tropea: Signed package what we call the sealed the sealed attribute uh so if we jump to to the next slide it's it's pretty straightforward the derivation process it it goes like that we have we have an entity with several attributes so out of that we create um a bunch of atomic entities 1 for each attribute that we want to sign uh then we created the NGS ild proof uh struct and nested inside each sealed attribute um. ✪
Giuseppe_Tropea: Taking into account as I just said that the ID and the types need to go together with the attribute. ✪
Giuseppe_Tropea: And then of course the value is exactly and this is important part and this is why I said we uh point to the uh Integrity w3c Integrity standard is it is exactly uh the proof um. ✪
Giuseppe_Tropea: The proof field when uh when it when the when the proof is is created. ✪
Giuseppe_Tropea: We derive uh the sealed attribute at this point Martin if we skip to the next slide it's the the of course the Reconstruction process is when when things get to the client then it will be more clear in in the next slide. ✪
Giuseppe_Tropea: Just straight jump the other 1 that's your typical workflow so uh context provider um generating signs uh 1 Atomic entity for each attribute each Atomic entity is transformed into a sealed attribute through the reaction process we've seen um then uh of course the provider. ✪
Giuseppe_Tropea: Uh inject the seal attributes at this point this entity can circulate a network and can undergo any number of cycles of of you know um. ✪
Giuseppe_Tropea: Signing and rearranging uh attributes um throughout uh the path until it reaches the clients when it reaches the client it's straight forward what the client needs to do is delete the reverse. ✪
Giuseppe_Tropea: And then uh validating each Atomic entity the additional step is of course that at that point when I when I validate the proof what I have to also check is that basically the entity ID that was a campaigning the proof is matches the 1 uh of the entity that I that I received. ✪
Giuseppe_Tropea: So this do we have another slide Martin what what was next let me yes okay so when we started thinking about so I think it it's pretty it's it's pretty straightforward what what we try to do but at the beginning we were even thinking but um why in the Integrity specification uh of w3c is integrity specification there's no uh such a thing as the possibility to map uh just a subset of course to sign just a subset of of of the properties of the jcd document but then and then to have a map a campaigning the document and telling you okay dear client you received this Json LD watch out this this and that attributes are signed all the others are not. ✪
Giuseppe_Tropea: Um we we started enacting with the your mailing is actually this male is from 2020 so um and and and and we promptly received the replies this is from Mano spawning he he suggested no that's the wrong way to do we have that map to uh because then uh it's it's it's a weakness in the security approach because the client expects the the whole document that he receives to be signed so this is why we took our approach of um you know making each of those sealed attributes uh stand on its own uh you can it's it's basically a uh Standalone um. ✪
Giuseppe_Tropea: Fully conformant document that you you can verify by yourself and then. ✪
Giuseppe_Tropea: You have the ID and the type inside inside the sealed attributes that refer to the external entity. ✪
Martin_Bauer_(ETSI_ISG_CIM): The idea you can check in for the type it's important we support multi typing so we maybe um during this um aggregation process more types were aggregated um but um it was signed with maybe just 1 of these types so that's why it's important to to keep the original information and then in that case okay that type should still be included in that but um for the atomic entity I mean of course have to um take only the type with the with which it was originally signed. ✪
Giuseppe_Tropea: Yep uh yeah I think yeah probably we didn't show any examples beforehand that yeah that the types can be multiple so we can have an array of times in their. ✪
Giuseppe_Tropea: Okay so this this first part was about signing um so we implemented this solution now we have to uh update it because okay the standards have evolved but mainly the idea stays basically stays the same. ✪
Giuseppe_Tropea: Uh then uh I I don't know maybe maybe I can go through the the idea of authentication part this is this is uh and then if we want to we can have a discussion this is um not yet a solution it's much less mature we we haven't published anything yet but we are now investigating. ✪
Giuseppe_Tropea: Uh uh some important some important ideas for for NGS I that that's coming from the ID authentication distributed uh uh identifiers um if if we skip to the next slide. ✪
Giuseppe_Tropea: Uh this is this is just basically my how I try to do some to summarize in in my main the core points about the the idea um. ✪
Giuseppe_Tropea: So the I think the important uh uh the important thing is that um okay they are designed to to function without the need for a centralized registry or authority but of course they can even we can in centralized case so good um we know that the the idea resolves to the IDE document uh where we have public key authentication method the service the service end point but the point in in my mind is that basically the uh the security is uh. ✪
Giuseppe_Tropea: Strictly related to to the public key infrastructure that underpins the the did uh so next slide please um. ✪
Giuseppe_Tropea: Basically the IDE authentication it boils down to to proving the the subject. ✪
Giuseppe_Tropea: Of va everything changed but um uh in the D document but of course the the the has to stay has to stay the same uh and as the the idea resolution step that um you need to perform in order to authenticate the the dat. ✪
Giuseppe_Tropea: So of course we can have multiple application workflows and. ✪
Giuseppe_Tropea: Involving for instance usual Challenger response mechanisms so that that this is not the key point so in in my mind it's what that what characterizes the the trust architecture of of the idea is basically the the did method that backs backs it up that specific the idea and in the next slide I tried to group and I also started this discussion credentials can be group list um grouping the different the ID methods um based on. ✪
Giuseppe_Tropea: Their trust architectures from from the most simple which is the ID key and people suggested also the ID jwa which is basically offline and to my understanding the subject there will be almost anything I mean as long as you able to create a key pair you can be a bot you can be a human you can be a legal entity. ✪
Giuseppe_Tropea: Uh then there's the ID method uh based on basically on. ✪
Giuseppe_Tropea: Key infrastructure which is the 1 of the cryptocurrencies the real Bitcoin what's permissions um it's Global Ledger and in this case I'm sure the guarantee is that basically the subject is at least some kind of entity that holds uh a bus in in the network. ✪
Giuseppe_Tropea: Uh did Sovereign is interesting because it's a permissioned global Ledger and I I understand it has um uh well specified trust governance and also the possibility it will have through this Steward nodes to delegate the trust to. ✪
Giuseppe_Tropea: Question is uh what what what is the subject uh typically in the ID Sovereign network is it a legal entity is it a human is there any guarantee about that. ✪
Giuseppe_Tropea: Then I think EU people are trying also to. ✪
Giuseppe_Tropea: Register another method which is the ID appz and in this case I think it's guaranteed that the identifiers are bound to Legal EU. ✪
Giuseppe_Tropea: Entities and the network is um is basically the nodes of the network of of the key infrastructure are basically um owned by EU Public public bodies so. ✪
Giuseppe_Tropea: They say they can support different trust Frameworks but bottom line they are owned by. ✪
Giuseppe_Tropea: Governments for big parties um then also interesting is the the ID web or DNS in that case I think I can be sure that the subject of the did in that case controls some kind of resource on the internet and they have the domain or. ✪
Giuseppe_Tropea: Have all of our kind of server on an IP address some somewhere on the internet. ✪
Giuseppe_Tropea: The I think we have a 1 last slide what is interesting to NGS because uh that's so now so much work about the data spaces where basically I think the goal would be to. ✪
Giuseppe_Tropea: A trust model where you have a private space is the owner of the database and you you want to um route the the Trust on on the owner of of the data space and also basically you probably want this um owner to intend be able to delegate uh trust to other other trusted members of of the data space. ✪
Giuseppe_Tropea: Members can either be humans or or legal legal entities uh so question is there a the IDE method where I identifies a guaranteed to be humans I mean it's a little bit of a rhetorical question and um. ✪
Giuseppe_Tropea: There was also a research paper about it so things are ongoing but that I think is probably the the most interesting the most interesting discussion there because well humans and and now and nowadays also AIS right which can pass uh so uh it's a Hot Topic and it's interesting I think bottom line for NGS ild um what what would be our goal is to implement this data space the the data space model where we we at least would like to be able to guarantee that what's behind the da is either a legal entity or more difficult I understand or a human. ✪
Giuseppe_Tropea: Um I think that's that's all on my side thank you for your attention and if you have questions or we can discuss. ✪
Harrison_Tang: Oh thank you thanks for the presentation and by the way like we do have uh. ✪
Harrison_Tang: I did invited the author of the person who credentials to actually talk about on November 12th so if you're interested uh make sure to mark that date on your calendar um I have also been trying to get the the the guys at the world coin to talk about their project which is also trying to uh tie credentials to uh basically uh authenticated uh humans right and uh using Iris scanners uh plus some cryptocurrencies which I know some people probably don't like as much but uh so far still working haven't got them to agree to present it here but uh but yeah those those conversations are definitely interesting and we'll continue those conversations here. ✪
Giuseppe_Tropea: Uh sorry Harrison when you you did say uh November 12. ✪
Harrison_Tang: Yeah November 12th uh we will have the Stephen uh Zoe and Shay uh the main authors to talk about a person who credentials yeah. ✪
Giuseppe_Tropea: Uh okay great but do you think we or I could join here uh freely or because this I quite didn't understand whether it's open or the calls are open for everybody or how does it work. ✪
Harrison_Tang: Yes yes so it's all free you can just join via that public link uh we also send out the the transcriptions and video recordings so you can also watch it afterwards if you're on the mailing list yeah. ✪
Giuseppe_Tropea: Right right right right right right right right right right okay great great that's that's that's interesting okay thanks for letting us know. ✪
Martin_Bauer_(ETSI_ISG_CIM): So I think what 1 key so I think the fiware context is where it's used most um but not only um and I think smart cities overall are the main users but it can also be used in other domains or is used um to a lesser degree but um I think in in smart cities it's already um quite commonly used in Europe but also in in Japan and in Korea they did their own project um in India there are using it um partially together with some of their own um standards so they did their own implementation and partly um used them and GLD so that's about the usage um. ✪
Harrison_Tang: So it's mostly governments basically. ✪
Martin_Bauer_(ETSI_ISG_CIM): Yeah but um not just yeah. ✪
Harrison_Tang: Got it and then so the next clarification question is like why do you need to Federate from different context providers I mean for example if you're a government you probably have a system of record right like that's Central so why do you need to have multiple contacts providers and then you have to uh autonomous the entities and then aggregated like what do you need to do that. ✪
Martin_Bauer_(ETSI_ISG_CIM): So why why to why why do you want the Federate system because you have different units um even within the city um you have different departments that may only want to partially share and then um if you go further um you may have um. ✪
Martin_Bauer_(ETSI_ISG_CIM): Counties um um States um um whatever and you may also have private um um operators so maybe if you're looking at parking spaces and you want that for a different areas um that comprise different cities and communities um then these may want to partially share um their data their parking spaces but not not necessarily everything and um in in such a case um it makes sense to to Federate that you don't want necessarily to have 1 single or owner that controls all the data. ✪
Harrison_Tang: Dollar okay yeah because I was assuming that if you're a government entity how we have have all in your centralized database so got it thank you. ✪
Harrison_Tang: Then the uh the point is the point of the signing mainly for data Integrity purposes like uh because we're talking about uh signing the the the data and things like that just to make sure that the data Integrity is good and then it's not being tampered with that kind of thing. ✪
Giuseppe_Tropea: Yep yep yep yep that's that's absolutely the goal yes yes because maybe you have a sensor that I don't know measures uh pollution I mean streets uh or CD and you'll be sure that I I don't know either measurements come from a certified uh uh Hardware maker or uh afterwards that they are certified by by the city I don't know you as a citizen may may want to be able to verify that and that they have not been tampered with by some malicious intermediate broker or relay node in in between. ✪
Harrison_Tang: Uh any questions from the audience. ✪
Harrison_Tang: And then 1 more question sorry uh when we're talking about uh the signature so the signature is far from the context providers right it's not does the broker actually sign the document once they aggregate all the contexts together or the signature is only from the context providers. ✪
Giuseppe_Tropea: Yeah that's the main point the signature is only from the provider because basically you don't trust the broker you don't want to trust the broker so um of course you may add yes you could add a second layer which because in in that case if if the broker for instance if the distribution broker before relaying the package to you uh Aggregates everything and then signs it on its own then it basically vouches that well yes I am the guy who is aggregating this I'm you can help me you can hold me responsible for this aggregation step but I'm not responsible for the value so if the pollution value is 25 or 27 it was not me uh who did that it's it's the original provider of of the data so. ✪
Giuseppe_Tropea: Yeah you know it's a an additional layer of of it's it's a different meaning. ✪
Giuseppe_Tropea: If if you if you sign the aggregated part. ✪
Martin_Bauer_(ETSI_ISG_CIM): So we we came to the conclusion in the end that um for us um on the attribute level it made most sense. ✪
Martin_Bauer_(ETSI_ISG_CIM): So if we send the whole entities um in Fireware sometimes we have a posteriori um Access Control check so you might filter on the entity later on so um um the broker May return an entity with all sorts of attributes and then later on it's um there's a PP that decides okay that attribute shall not go out and filters it out so then um it would break um um the signed entity. ✪
Martin_Bauer_(ETSI_ISG_CIM): So such use cases may also be there and then um in in in these cases it doesn't make sense to um sign whole entities. ✪
Harrison_Tang: About the registry like the registry of uh different uh identifiers and then also the trust framework is it the same as Brokers or is it going is it separate like basically the broker the aggregator is separate from the registry where are they do they uh basically the same they play the same role or the same entity played both 2 roles. ✪
Martin_Bauer_(ETSI_ISG_CIM): The sums of identity management and so on that can be completely separate. ✪
Harrison_Tang: Got it but in the real world is it the same entity that plays both roles the registry and aggregation or they are separate. ✪
Martin_Bauer_(ETSI_ISG_CIM): Could could well be but um what what we try some with NGS LD to to be able to work in in in also in different environments so basically so we don't want to make um just 1 single assumption so what we are also looking at now in our security is okay um we look at different scenarios and um basically want to give people an idea if this is your scenario maybe um you choose from this set of um Technologies and if you have a different 1 you're choosing another we're currently discussing the use of NGS LD in data spaces so which is a quite Hot Topic at least in Europe um and they are the the IDS um play an important role in in in this um underlying idea that you don't have um Central um entities um controlling this on so that's why 1 of the reasons why we said okay it's also for us makes sense to to dig a bit deeper into the did case. ✪
Harrison_Tang: Right so if no further questions I just want to thank you again josephe and Martin uh who actually taking the time to present here today at w3c ccg thanks a lot. ✪
Harrison_Tang: So this concludes that today so ccg meeting will publish the video audio and transcriptions and next day or 2 and then uh please uh come back um next Tuesday and next Tuesday we will have. ✪
Harrison_Tang: Jeremy and Andreas to talk about BBS signature scheme benchmarks. ✪