Hello, everyone. Welcome to Con 42 Internet of Things IoT 2024. My name is Dilshan Rabbi Kasani and welcome to my session on digital twins. So the session I'll be covering today is about the transformative role of digital twins for network resiliency and fault tolerance in IoT systems. As IoT systems grow and evolve, the increasing complexity of managing the networks for those systems increases as well. digital twins can offer solutions. for these problems. So let's dive right in. today I plan to cover, give a brief introduction about what digital twins are, what is network resiliency and what is fault tolerance and why these are important and how digital twins can enhance them. I'll touch up briefly on what digital twins are, and the role of digital twins in this, network resiliency and fault tolerance. I'll also provide some use cases, in IOT systems and advantages of using digital twins and, implementation challenges of digital twins as well. and I'll conclude with future directions and what to use within digital twins. So let's get started. so we are currently living in an age of smart, meaning that every system around us, we prefer to have it as smart system. Either it's smartwatch or smart car or smart city or whatever, right? as we go towards the smart approach and smart mindset, there are many complex IOT systems that are involved. which need to be always interconnected with each other and transfer data to the appropriate systems. So to support these systems, what we need to do is to simulate and troubleshoot any issues with the IOT systems, right? So coming to the IOT architecture. So you'll have your usual sensors, actuators, and smart devices, which are transmitting always the accurate amount of data that it's required to digitally simulate, systems. So you have, you'll have your gateways and you'll collect all this data and store it in a data lake somewhere. And once you have all this is a traditional approach in machine learning too, right? Where you'll get your data, you'll clean it and you'll put it in a data warehouse. And then perform some kind of machine learning or, build some models around that, predicting failures or, predicting any issues with the systems. So, you'll have your control applications, your models, your machine learning, your data analytics. And, at the end of the day, What you really need is everything working together. so coming to network resiliency. So what basically network resiliency means is, the ability of the network to withstand and recover, quickly from disruptions and failures. So in case of a network disruption or a failure, the network needs to come back up, right? Okay. When you have smartwatch a smartphone, you know how it feels when, it goes out of network, right? So you always need that quick, turnaround of connecting back to the network. And, in case of any failures you need, for it to connect back, as well. and the capacity of the network, you need the network to come back up fastly and adapt and continue. So not just. Basically, the same thing happening twice, right? So you'll want the system to learn basically what happened by simulating it in the backend, either through some data lake or some machine learning model. So you'll need this information in the backend. So network resiliency, meaning that, the network needs to withstand and recover quickly from the disruptions. So importance of network resilience. you need to minimize the downtime. So we all face downtimes. But, what, the companies need to do or what the systems need to be in a shape where it reduces the downtime. So you want as much downtime, as minimal downtime as possible. So you need to reduce the service, interruptions. and you need to ensure the operations are continuous. you need to protect the data as well, right? So once your network is down or some system is done, there'll be a multitude of, vulnerabilities that you will be exposed to, which you need to quickly, recover, or you'll be subject to information loss or corruption, during the outages, data corruption. So you need to maintain the productivity as well. So even though you. users lose some kind of network, access, you still want them to perform some kind of things, right? And all of this, the things that I talk about in terms of, a digital connectivity. So this is a basic digital connectivity that everyone needs to have. And I'll discuss later about how digital twins can enhance this. So enhance customer satisfaction. So customers need to have the uninterrupted service, and increased customer satisfaction or you lose customers, right? And fault tolerance. So this is similar to, network resilience, but what fault tolerance is, even if a part of the system fails, rest of the system can function, as smoothly as possible. So some parts of the system fail and everything goes on. That's not a good experience, right? So you want to minimize the downtime of, applications as well. you want to improve the reliability. So the systems need to be more dependable. and it's not very easy for IOT systems to be, simulated because the complexity of, so let's say if you have, a hundred thousand or some big amount of a number of IOT system connected together. It's hard to simulate all the systems where failure, happens. So that's where digital trends can help. protect protecting, data. during failures, you want to have a backup mechanism and, Some kind of mechanism where data corruption doesn't happen and, failure doesn't happen. So role of fault tolerance, in minimizing disruption. continued operation. So you have to enable systems to function when, some components fail and, prevent complete outages. So where you can't even access the system. you want to reduce the downtime. So all of the service interruptions that happen, you want to reduce that. So this is similar to network resiliency, where, in fault tolerance, what we talk about is it's, it's a system's capability of coming up, even when a component fails. Right and Reliability, so Overall reliability and dependability of the systems are also important and protecting the Critical information of the customers when our failures happen and enhancing the customer experience This is what the fault tolerance Role is in minimizing the disruptions. So coming to the actual core of the topic Digital twins. So what are digital twins? So digital twins are nothing but a virtual replica of the actual physical system. So you have your physical system, let's say that a manufacturing system or a component or a service or anything, right? So you have all of the data of the physical system where you can build a virtual replica of it in the cloud or on prem where it digitally. So it's a digital representation of a physical object. so the, thing with the digital twin is you need real time data. So continuously, you need to update the physical, digital system with the physical systems data. So all of the data that you have in the physical system, you'll store it, in near real time. And you'll build a digital twin. digital twin out of it. So this can help you in a multitude of reasons. So you, for any simulation, right? you can simulate, a failure happening within a particular system, digitally without having any impact on the physical system. So this way you can set up a feedback loop where in digital systems, you can, experiment with multitude of, parameters. and have the changes implemented in the physical system. So this can help the decision making process where you perform multiple number of tests in the digital twin and implement in the real system. examples of the digital twins are manufacturing equipment, where you have multiple number of. pieces, IOT systems. And once you get all of the data in a place somewhere, you can build a digital twin out of it. And you can have an exact replica of your physical system, digitally. So buildings, infrastructure, supply chain, human bodies, even in healthcare can be represented, digitally. So coming to the types of digital twins, a manufacturing component or, within an IoT system are, a simple piece of equipment. You can have all this together. in the. digital realm and have it as a digital twin of your physical system. you can have either assets so you can, it can be a car, an aeroplane or, some kind of smart device where you can have it, digitally, somewhere. So process. So even the process can be, digital made as a digital twin. So for example, you have a manufacturing process that runs somewhere, from point A to point B, you can take all of the steps and all of the IOT systems involved in that process and get the data out of the systems. and make a digital twin out of it. So you can do it for the systems and network of systems as well. So let's say you have a multiple number of systems that are connected together. anywhere you can get the data out of the IOT system, you can make a digital twin out of it. So here is a digital twin life cycle. So as I mentioned, data is at the heart of the digital twin. first step is to collect the data. So any system you have, you'll basically collect the data. You'll save the data somewhere, you'll transmit the data and you store the data somewhere. So once you have the data stored for all of your systems that you require, you'll evaluate and analyze the data with multitude of parameters. So you can simulate the parameters as well. So for example, if you have, a Tesla car or, any car, you can. crash it a hundred times, right? So in the digital world, what you can do is you can run a simulation that you can, based on the parameters. So what happens if I crash it in this way and the parameter of X is something. So you can have multiple parameters where you can adjust these parameters and test it out digitally. You can evaluate. The simulations and based on the simulations, what you can do is connect back to the physical system. So now, once you have the results and you have some parameters, now let's say you want to make a change. So you'll link back to the physical system. So this is a continuous loop. Where you're from collection of data to updating the system. So this feedback loop, you can automate the feedback loop as well, where you have a data linkage in between. So once the data link is established and you have all the process set up. Based on some parameters, you can simulate something, and you can basically, update the setting back to the physical system. coming to the challenges in IoT networks, as, IoT networks grow and evolve, so the complexity of the system evolves as well. so number of interconnected devices. Now, everything you have, you wanted to be connected to each other, either it's your phone, watch, earphones. and this is, this increases with the IOT systems as well. So where you have multiple number of devices that needs to stay connected together to perform accurately. So as it increases, as the number of systems increase, the, complexity increases as well. And the protocols with which they communicate with each other differs as well. And the standards with which they communicate change as well. as these interdependencies increase, the complexity increases. and all of this can be performed within a digital world without having any impact on the physical world. that's where digital twins really help in maintaining IoT systems. real time data dependency. So some critical applications within IoT systems require real time data for optimal performance. So that is a challenge as well. that I'll cover in the digital twin section as well. So network latency. So all of the systems to stay connected together, the systems need to be connected very fast. So network has to be. really fast and the connection speeds has to be really fast. network latency is another issue as well. So some failures, so let's say if you have a network outage or a packet loss or high latency, this can disrupt the operations. for your systems, where you can, where digital twins help in, simulating this. So security vulnerabilities. So all of your security devices, you can simulate it. let's say your firewall goes down, right? So physically, it's really hard for you to test that out when you have actual users in the system, where virtually digital twins make this easy, where you can simulate a firewall going down and make choices based on these parameters. Scalability. So as IOT systems grow in, scale, scale, managing and maintaining these systems increase as well. and limitations to the traditional methods. So rather than having proactive approach, for fault management, you can have a simulated experience in the digital involved and have the systems upgraded automatically in a feedback loop. and it's also difficult in predicting and preventing complex failures. And if you have a large IoT ecosystem, it's hard to visualize the entire system way physically. So if you have it in digital, it's easy to, visualize right and make decisions based on those. So digital twins in network resiliency and fault tolerance. How can this help? How can digital twins help in network resiliency and fault tolerance for the limitations that we saw earlier? You can do a proactive failure detection so you can identify any potential issues, either simulating a variant error of IoT systems or network congestion or before they occur. So you can predict what's going to happen when let's say 100, 000 users attack your network or use your network or use your system. You can simulate that and have systems ready in place. and you can analyze the data patterns. So this is on par with the traditional machine learning and data science model as well, where you can use traditional data science and machine learning within the digital pins and simulate those, results and update the systems back, physical systems. So predictive maintenance, so forecasting, equipment failures, and, you can schedule maintenance proactively without, having any impact on the physical systems. So you can, traditionally how it's done is, you collect all the data and you have it somewhere, but that's just a store of data. and not a physical system within itself. So a digital, a replica of the physical system, as such. But you'll have a big table of data and you'll perform some machine learning, you'll predict your analytics and whatnot and update the physical system. With the digital twins, what you can do is to set up a feedback loop. and have that, help, with the network resilience and fault tolerance. So disaster recovery, so you can, simulate the disaster scenarios where, for example, a natural, calamity occurs, like earthquakes or floods, and you can assess, you can physically, It's physically, it's impossible to, simulate these rights. You can, in the digital world, it's easy to simulate these. So you can simulate, what happens when an earthquake or flood happens. You have various systems, all are interconnected and assess the impact of the critical infrastructure. and use cases, coming to the use cases, there are multiple use cases for the IoT systems where you can have, smart cities, industrial IoT and healthcare IoT. So within the smart cities, you can have energy grids where you can optimize the energy distribution of the grid. you can predict the, blackouts or any service interruptions. and you can integrate, renewable energy sources as well. So for transportation systems, you can improve the flow of traffic. You can simulate based on the physical data, physical traffic data. You can capture all the physical data and have it, simulate in a digital environment, a digital twin of traffic. So remember I mentioned you can do a process as well. So all of the traffic that flows through a certain point, you can have that process. As a digital twin and simulate what happens and simulate the number of, accidents or, anything that can happen. and you can optimize the traffic based on that. You can optimize the public transport as well, where in holiday time or in a high flowing area, you can predict, based on the digital twins. You can set up simulation scenarios and, Prevent those, public transport issues and enhance the roads have safety as well. So coming to industrial IOT, coming to predictive maintenance. So you can, predict the machinery health or equipment failures based on all of the IOT systems that are connected to, a modern machinery system, right? equipment site so you can predict all these equipment failures and schedule the maintenance a accordingly. so optimizing, process optimization, you can, improve the efficiency of a process, reduce waste, or enhanced product quality. You count this, you call this, lean manufacturing, right? Where you have green manufacturing, lean manufacturing. where you, improve efficiency, reduce the waste that you produce, and you can enhance product quality as well. So coming to healthcare, healthcare has some, really interesting, digital twin scenarios as well, where you can monitor a patient health remotely by, creating a physical, a digital replica of the patient's health. So this helps you in early intervention. And, if any critical life support systems are needed or any medical devices are needed, this can help as well. So coming to the implementation challenges, so there are, some challenges in implementing digital twins as well. so data integration and interoperability. So connecting data from, collecting and integrating data from different sources is a challenge as well. So you have your thousands of hundreds of thousands of millions of sensors. you'll have to collect those data and you'll have to make sure that the network that the data is transported to is seamless and store the data somewhere. you will have your databases, your APIs and whatnot. So you have to ensure the data consistency, accuracy and interoperability, across the different systems as well. So scalability, so scaling digital twin models, is a challenge as well. handling the complexity and computational demands of large scale IoT deployments, with, Increasing IOT system, it's hard to manage it in the digital twin as well. forget about managing it physically, but, digitally it's really hard to manage as well. deploying scalable for complex systems. So you have a million devices, and all of those are, in the form of a digital twin. it's. It's okay to perform some kind of metrics or, some kind of simulations based on that, but it's really harder to have, that feedback loop, for millions of systems. So the scalability within the digital twins is hard as well. So real time synchronization. So this is challenged with every system, not just digital twins, but, you'll at least have to have a near real time. data ingestion loop, where you synchronize the data between your physical system and the digital system. So that is a challenge, for the digital twins as well. And you have to ensure your virtual model actually, accurately reflects the current state of the real world system. Now you have all of your million devices or, however many systems you have in the digital world. Setting up those systems initially will be a challenge as well, where you need to set up the way that it's exactly in the physical world. So that can be a challenge as well. But I think to conclude this session. let's dive into the future directions so you can use machine learning and AI and integrate it into the digital twins, and, power your decision making. you can have digital twins that can make autonomous decision based on real time data and predictions and set up a loop for that. That's a loop that I talked about where you can, simulate the systems and you can make it. take some autonomous stations and implement in the physical system. leverage of the 6G networks. So now, with the ITU, guidelines, providing for 6G networks, the bandwidth and, network connectivity. for these systems will be, very much faster and, higher. So to enhance real time, data exchange, 6G can be used for digital twins as well. exploring, UR LLC, ultra available, low latency communication for mission critical systems, mission IoT systems can be used as well. So these are some, things that can enhance the digital twin experience. And cross industry applications. So you can have it, in multiple, cross industries where one of the system that you have physically is in one industry and the other system that you have is in other industry. So cross industry, can be, another, future direction for digital twins as well. So in this talk to conclude in this talk, we have covered a digital twins, how they help IOT systems, how they help reduce network resiliency and fault tolerance in IOT systems. And how can you enhance your IOT systems based on digital twins? Thank you for coming to my talk and you can connect with me on LinkedIn. You can type in Dileesh Chandra Bikasani and thanks for attending this conference with me.