Ensemble de rappel
Ensemble de rappel
Ensemble de rappel
Ensemble de rappel
Ensemble de rappel
Ensemble de rappel
Ensemble de rappel
Blockchains and Use Cases
(Refer Slide Time: 00:15)
Alright, so welcome back. So, if you recall, we talked about the Nash equilibrium andwhat actually constitutes machine learning. I will just spend maybe two minutes on. Ihave taken it from a recent book by Sinha et al. (2020). You can actually see thecomplexity and the time progression of machine intelligence and maybe what is themost dominant algorithm.
As I mentioned in the previous session, the things started, maybe 1956. You can seeneural network. There would an AI winter then we are talking about AI spring thesedays. So, you have deep learning algorithms. But the future would be we are lookingfor super intelligent machines.(Refer Slide Time: 01:17)
Machines, which can allow us to meet, their action should be such that they achieveour optimal objectives. So, they just do not achieve their objectives. They should notjust achieve our objectives, but they should achieve our optimal objectives. Theyactually sense the environment and decide what is the best action for us. So, we needcomputability of that order.
In fact, this could be highly philosophical. I do not want to spend too much time onthat looking the course objective is, the focus is more on manufacturing. But if youare keen maybe look at Church - Turing Thesis. So, I close the machine intelligencepart. Certainly, I think each of these algorithms like deep learning, you can actuallysee courses on that. So, that is not something which I really want to emphasize as partof this course.(Refer Slide Time: 02:28)
But what would be relevant for me to actually see how the blockchain as a technologycan help the manufacturing context. If you recall, we talked about the manufacturingobjectives, which was again the flexible, cheap, diverse, better, and sustainable also.So, five objectives there and for supply chain also I think these are the five importantprimitives.
What goods are moving? Who is the buyer or who is the seller? Where it has to go?Where to locate? Then two things where things move or sorry, how things move. So,you talk about logistics and where things sit. You talk about inventory. These fiveprimitives become the part of supply chain and each of these aspects offer respectiveinsights into different risk and supply chains.
So, I think I mentioned in the previous sessions that the supply chain of a companylike Intel, they could not trace the supply chain beyond two or three tiers when theJapanese earthquake came in 2011. So, what it actually brings is, you are looking fortraceability, compliance, and visibility. So, I can trace from where that material iscoming.
I can trace, so I want that all the compliance part should be taken care. So, lot ofregulatory requirements may be there. You can think of even Tesla recentlyannounced that they want to come up with a zero cobalt battery and the mainobjective is because the cobalt supply chains are still not fully traceable. So, if there isa compliance requirement that it should come from ethical sources.
Let me also add that large number of cobalt supply chain, the origin goes to Congoand you would actually see lot of child labor is involved there. We want at all thosethings should be avoided in the supply chain. So, you want traceability, you wantcompliance and you want where the goods are sitting? How much is the delay? Whatare the quality issues? All those things should be visible.
So, these three things become a part of the story of the modern supply chainmanagement and I think we extend this argument to digital supply networks also.(Refer Slide Time: 05:11)
These are some of the risk. COVID times, you would see lot of discussion on resilientsupply chains. Can blockchains allow us to make that? Can blockchains allow us toachieve that? So natural disasters, you can think of national disruption, accidents. So,I am pretty sure if you start mining, you would relate to lot of examples like this.
So where you see lot of risk, which comes in achieving these supply chain objectives.(Refer Slide Time: 05:45)
Now, I will go slightly deeper into that. When we talk about supply chains, when wetalk about blockchains and supply chains, distributed ledger is something which mostof people associate when we talk about blockchains. We have spatial, so we aretalking about now network. There could be a network where so you can think of anetwork something like this.
Some of these could be nodes and some of these nodes could be miners also. So, whatthey do is, so they actually gather blocks of transactions, and they actually compete toverify them. I think I will explain all these things and in return for their service.Miners that successfully verify a block of transaction achieve both an allocation ofnewly created currency.
So, for a distributed ledger, I will give you an example. Like if you have to transfer amoney to some other bank account, a request you send to your bank. The bank willgive it to a centralized ledger, maybe sitting in the RBI and then it is verified that youactually have that money and it will be transferred to the other bank’s account. So, itis done through a centralized ledger.
But now we are actually saying there is a distributed ledger. So, all these nodes couldbe part of the distributed ledger. In this case, the verification is not centralized, it is inthe distributed manner. Only thing you have to ensure that no single node is powerfulenough to change the ledger and that is where all this part auditability, immutabilityand disintermediation comes in.
It means that you can easily verify. It should not be immutable, means no one else,one person can be powerful enough to change that thing and it should not becentralized. You want no intermediation there. So, you just want that everythingshould be distributed and this actually, so you have to come up with some kind of amechanism where you achieve all these things at the same time and that is where thewhole notion of blockchains comes into the picture.(Refer Slide Time: 08:18)
As I mentioned that we use some ideas of game theory. In fact, the main idea of gametheory is to avoid that peer this cheap talk. Distributed ledger, in fact we sometimescalled blockchain. Blockchain and cryptocurrency could be you can saycryptocurrency could be a subset of blockchain. Was made possible by the emergenceof many innovations, which include the internet.
So, it rests on concepts from cryptography, Game Theory and peer to peernetworking. It is a free decentralized payment system. I am now emphasizing only onthe bitcoins part, in which copies of the ledger are shared between all participants anda process is established by which users agree on changes to the ledger. So, to makethat transactions valid.
Since anybody can check any proposed transaction against the ledger, this approachremoves the need for a central authority and thus for participants to have confidencein the integrity of any single entity. How it is done? In fact, this problem is coming
from the medieval times. It is called as the Byzantine Generals Problem. So, people incomputer science may be aware of this.
But the main idea here is that it should be decentralized and no single entity have thepower to change the ledger.(Refer Slide Time: 09:55)
So, as I mentioned that we use ideas from Game Theory and we call it a cryptographicproof of work, which ensures that the agents who are the miners should not beinvolved in the cheap talk. Let me see whether I have put that thing. This example wealready have seen as part of our game theory discussion that players can communicateto each other that we are going to play T, L.
But when the game is played, you again converge to Nash equilibrium. So, you wantto avoid this cheap talk of communication, which is T, L. You want to come up withsome kind of a mechanism or incentive mechanism, which ensures that you do not dothis cheap talk and that is where this cryptographic proof of work comes into thepicture. So maybe I will explain it in the subsequent slides, how that thing works.(Refer Slide Time: 11:03)
Let us take this as a network. Now, there are all this A B C, D E F, these are the nodesin that network and all these nodes are connected. All these nodes are well connected.Now there is a transaction which is happening between E and F. This red line you canactually see as part of the transaction and D, G and I are the miners in this case. So,the whole ledger actually sits on a tree.
In fact, it should be available on all other nodes also. But these are the miners. So,what they do; they actually do the verification of this transaction, whether A actuallyhas ample money, which can be transferred to F. So, there would be history of how Aachieved that money and that is also part of the ledger. Now all this information comeas part of blocks.
You cannot change a block unless or until all the miners agree that it can be changed.It means that they actually have to come up with some kind of a consensus to changethe ledger. Now the point here is whether they can actually do the cheap talk. Toavoid that thing, they actually have to spend some time in. So normally, what is thevalue of the hash?
It is more like a puzzle for them, which can actually be done by trial and error. Thepuzzle is developed as part of the network. It follows some kind of a Poissondistribution and it takes about 10 minutes to solve that puzzle. On a average, that isthe meantime and once you solve it, you communicate to all the other nodes in thenetwork.
So there is a and then so if it turns out to be correct, your puzzle turns out to becorrectly solved. Then you will actually be paid for it. Now this creates some kind ofan incentive problem, because you actually have to spend and you may not be theonly one and the complexity of the puzzle actually keeps on increasing. That isactually part of the architecture.
Now the point here is that it is not you can say cheap for you to just communicatesomething that this has been verified. You have to spend something; some effort hasto go and you need computing power. All the other nodes are competing because youare actually paid for it. You want to do it as fast as possible and let others also do theverification.
Let me also add that all these nodes are anonymous. In fact, the example which we aregiving is of the public blockchain. There would be permissioned blockchains also. Infact, I think Unilever actually implemented the permission blockchain as part of thedigital supply network. In this case, we are talking about a public blockchain and inthis case, all the nodes are anonymous.
So, you do not know where that replica sits. This actually solves this problem of theincentives to avoid the cheap talk.(Refer Slide Time: 14:24)
You can actually see, you can think of an example where so I will give you so we aregoing to talk about public, private cryptography and what happens is, if you assumethat you have to do a transaction from one person to other person, like in our case forA to F. F may generate, let us start with F. So, F will generate a pair of public and theprivate key.
Private key will stay with F and this public key will go to A and A will attach that thedigital signature and it would actually come back to F. Now even if someone knowsthe public key, they cannot actually generate the private key out of it. But if you knowthe private key you can generate a public key. So, one public and one private key.
So, even if you know the private key, you can generate the public key, but theconverse is not true. So, even if the public key is known to all the other agents, theycannot verify, what is or they cannot generate what is the private key. This issomething, so I give you an example how this whole thing work.(Refer Slide Time: 16:08)
I will take maybe a simple example, which can illustrate this idea of what I am sayingof the public and the private key. Let us take two agents. One is Alice and Bob andassume that they have to do some transaction and I am giving this analogy of paintmixing to make a sense of what we are saying. Now, Alice decides a color which isthe public color.
When I say public color, so it means that that color is known to everyone includingBob. But then there is a third party or third agent which is Carol and they want thatthe transaction should be hidden from Carol. So, she cannot make over what actuallyis the transaction. But this color, which is the public color, which is green is known toall the agents.
Alice and Bob announced that this is the public color and Carol also can observe it,but at the same time uh Alice and Bob, they also have a private color. That is theirchoice, which is not revealed to even to Alice or to Carol. So, Bob decides red as acolor, which is the private color. Alice decides blue as the private color and that is notrevealed to anyone.
Now what she does is Alice, she mixes this private and public color and get a private,public mixture. This is let us for say Alice. Bob also does the same thing. He mixeshis private color with the public color and come up with a private public key pairwhich is for Bob and let assumes that he puts so let us say that there is a center placewhere they put this mixture.
So, Bob comes and put the mixture in the center. Alice comes and put the mixture inthe center. But that private color is still not revealed. Now Bob comes and take theAlice mixture and add with his private color and Alice comes and mix the Bobmixture with her private color. Now what we will see. So it contains red, blue andgreen. It contains red, blue and green.
So, both will know they need not to know what is the private key, but they actuallymake out what was the final mixture. Both will get the same thing. This is the mainidea of this public, private and Carol even if she knows, even if she picks any of thosecolors, even if she picks the mixture also, she does not know what to mix with that.So, she may not get the same color what actually has been achieved by both Bob andAlice and that is the main idea of the public, private cryptography and that is part ofthe blockchain.
Now only thing is so this transaction is between C and F. Someone has to do theverification and that verification so as soon as this is done, this becomes as part of the
block and as I mentioned that you can add or you can edit a block only when you dothis, prove this cryptographic proof of work and which comes at a cost.(Refer Slide Time: 20:14)
So, this you can actually see in this slide, so there is a latest block, this is the minersblock and then this nonce value you have to estimate. Hash value less than thethreshold, yes. If no, you keep on trying. Blockchain is becoming a general-purposetechnology. It lowers the transaction costs. So, you relate to that idea of RonaldCoase.
Through costless verification and without the need for costly intermediation, whichimproves the efficiency and scope for markets moving them closer to a direct peer topeer ideal. Lot of things about the transparent, visibility. All these things complianceall these things come part of the blockchain. So, in this case, we are in fact calling it adistributed ledger technology or a trustless consensus engine.
Everyone converge to the consensus, and remember that we are not calling it trustful,we are actually calling it trustless. I need not to trust other person, but I trust thetechnology because I know that the technology is robust enough that you cannotactually change anything unless or until everyone agrees for it. Miners candidateblock will be changed only when you solve this puzzle and that is not cheap, it isexpensive.
You have to spend some energy some resources to achieve that. That forms the basisof this. I hope, you got some idea about how these things work and as I mentionedthat what they are achieving is they are reducing the transaction cost, and which maylead to a different kind of organization altogether.(Refer Slide Time: 22:02)
I will give you some examples for this. I think this will give you some idea. There isin fact a company called Provenance. They are actually tracing the food supply chain.You can see that you can actually trace the whole origin of a food product in theconsumption and the technology of blockchain is becoming the benchmark for it.
Large logistic company like Maersk and IBM, they actually have come up with aTradeLens Blockchain Shipping solution. They want that industry wide collaboration;it was announced in Jan. Advances as more than 90 organizations participate in theglobal trade solution. They really want to improve the traceability. They want toimprove the compliance.
They want to improve the visibility. So, all these things are becoming part of thedigital supply network.(Refer Slide Time: 23:22)
That certainly would have implications for manufacturing also. I will take anotherexample, which is the Walmart Canada, and they have come up with a blockchainbased supply network, which is similar to the for free tracking and payment. I giveyou the numbers. They operate about, they own in fact 2100 trailers and trucks. 70carrier partners and they transport over 500,000 loads of inventory to about 400 retailstores across Canada.
It is estimated that about $150 billion per day is tied up in the supply chains in US forinvoice disputes and that is where I think the Walmart Canada initiative ofimplementing a permissioned blockchain with 27 nodes. So, this is different from thepublic blockchain. In permission, you allow who should be the miner but theincentives may be very similar.
So, for a organization like Walmart, I think the permission blockchain seems to be abetter option than going for a public blockchain because you may actually sometimessee lot of spurious activity in a public blockchain. So, blockchain ledger connects tolegacy systems such as ERP accounting systems and transportation managementsystem. So that has been already done by Walmart. So, this is something interesting asan example that how they are actually integrating.(Refer Slide Time: 24:55)
Another example, which comes is from Unilever. Unilever is a company which is coheadquartered in London and Rotterdam in the Netherlands. They own about 400brands, sales in about 190 countries and revenues about 50 billion globally and youcan actually see that, if you read this, uh this plant makes Dove soap and ice creamand this is a joint offered by Unilever and Microsoft to actually establish a virtualfactory to tune up its supply chain.
What they are doing is they are actually using AI or machine intelligence connecteddevices, harness a stream of data to keep production moving and reduce the cost. Infact, they come up with the digital twin to replicate the physical system. We alreadyhave talked about that.
They want to actually make their production more efficient. They are looking forconvergence of technologies like IIoT, Cloud, ML and just to give the context thatthey have launched eight more digital twins in factories across North America, SouthAmerica, Europe and Asia. They want to name it digital rewiring. Unilever is alreadydoing that. I have taken this as part of a Wall Street report.(Refer Slide Time: 26:26)
Other company which is doing as part of manufacturing is Caterpillar. They come upwith something called as a Live Factory. They are converging. You can actually seethis. So, you in this case, we are talking maybe about a machine. You put a sensor onthe machine. You track the movement within the factory. You remove the sensor andwhat it actually does it maps the movement, location, time spent at different processesand the quality summary also. So, the quality defects.(Refer Slide Time: 27:15)
In fact, just give you the numbers. This is a kind of a dashboard, which comes as partof this Live Factory. It is called as manufacturing a common operating environment.You can actually see the defect count defect per unit. The new Live Factory was ableto identify $12 million to 18 million in savings as reduced latency, automated
activities and they could actually reduce drastically the inventory cost by $1.2 million,labor cost by about $500,000.
So, this has come as part of implementing this Live Factory, where you are actuallyconverging on all these solutions. You can actually map not just outside you can do itwithin the factory. You can optimize your resources.
The main idea of sensing, collaboration and optimizing can be seen as part ofCaterpillar. They have integrating made with a typical SAP system.(Refer Slide Time: 28:24)
So, if you look at mass, you can say personalization. Two companies are at theforefront. One is Nike and other is Adidas. Adidas in fact, have come up with 3Dprinting solutions and you can actually see there is a big change in that. We startedthis whole course when we talked about manufacturing in China. But now what ishappening is this high-tech factory is bringing the production back to Europe.
So, you can actually print it closer to the customer and this customization allowsbetter matching of the demand and supply also.(Refer Slide Time: 29:16)
So, as I mentioned about Nike, you can actually see the quotation, One day consumerswill be able to buy a shoe design file from Nike. You are not actually buying the shoe.You can actually print the shoe themselves. So that is the kind of innovation youwould be looking in the context of manufacturing. We want more customized, morepersonalized, at least for the apparels and the shoe, that thing is true.
The company has established partnerships to build a 3D digital design system totransform its product creation process enabling capacities such as digital printapplications, and rapid prototyping. They have actually brought 3D printing as part ofit, and they have established the house of innovation at different cities. This is inShanghai, and they are actually calling this approach is always on approach.
So, you are actually closer to the customer, you are customizing the products and thepoint here is it is the convergence of all these technologies. You are not just about thedata analytics; you are integrating it with the manufacturing process also. So, in thefuture what the Nike former CEO is claiming, you would actually see that is true fornot just shoes, it may be true for many other products.
This brings me to the closure of this session and this also brings me closer to theclosure of the business part of the future of manufacturing. The future sessions, wouldbe taken by Dr. Chandrashekar and would be on additive manufacturing. I hope youhave enjoyed this discussion. Thank you.
Nous enverrons les instructions pour reinitialiser votre mot de passe à votre adresse e-mail associée. Veuillez marquer votre adresse e-mail actuelle.