So we're going to take an inside look at the markets, the sizes and the dollar estimates. This is huge. The university purchased a market research report that's couple 100 pages long and they interviewed CEOs of Cisco and Intel and HP and all these big companies that you are familiar with and produced estimates of how much revenue is available. So the bigger these numbers are, the more opportunity there is for you from an employment perspective. Because wherever that money is, they want to hire savvy engineers and bring them in and help them continue growing their businesses. I wish I had access to information like that when I was in the university. There wasn't the internet, it didn't even exist then. So, it's pretty cool. Platforms, we're going take a look at IBM Bluemix and Watson is one of the extra credit projects. So you're familiar with the game Jeopardy? Did you happen to watch when Watson played against Ken Jennings? It's pretty amazing, wasn't it? I was really astounded. I heard Jennings is really good and I don't remember the name of the other gentleman, but they're the top two winners, and of course, Watson was programmed to play Jeopardy and we'll see that for machine learning and data analytics there's a tremendous amount of human involvement in crafting a solution for a particular segment. So a lot of work went into programming the Watson, setting up and conditioning Watson to be able to play Jeopardy, but it was pretty impressive. We'll study networks. There's a number of topics, but two big ones are network functions virtualization and software defined networks. I have mentioned security already. One of the areas when Andy and I were brainstorming subjects for this course we were just thrown out ideas, writing them down on a whiteboard, and I said well, what about project planning and staffing and execution, what if one of these students want to go off and start their own business? Would it be useful to them to understand what it takes to successfully staff and execute a project? You got this great idea as an engineering student, you graduate and your uncles in the venture capital business, so he arranges for a million dollars of startup money. Okay, now what? What do I do? Who do I hire? How do I organize my project? We're going to take a look at that. Sensors and file systems, talked about that, machine learning big, data analytics, system C. I think system C is so cool. One, it's free and it's way faster than Verilog, we've ever done. Has anyone used Verilog or SystemVerilog to do modeling? Not RTL for synthesis, but the Model a system? No. It can, you can do it, but you pay a fairly significant overhead. There's a couple of free Verilog simulators out there, but I don't think they compare in terms of execution speed against the big three, synopsis, canes and mentors simulators. Going to system C, everything is C and it runs way, way faster, and we're going to take a look at when a physical plant, when a physical system is highly instrumented with sensors can generate a ton of data and collect all that data and operate on it and learn from it, but also we can create a virtual version of a physical system like a virtual manufacturing system, for instance, and perform what-if scenarios on that system using the data that was extracted from the real physical plant and you would develop that model of the physical system using systems C. We'll see how that works. Debugging deeply embedded, referenced that already, and I've got three guest speakers lined up. So one of them is Don Matthews, he was my security mentor at Seagate, he is now doing security for MD. He's here locally and he's agreed to come in and talk to you. So once I'm done with my lecture on security, Dan will come in and add on to what I have to say about security and I think you will enjoy it. I have an engineering manager from Trimble Corporation, anyone familiar with Trimble? Do you know what they do? Yeah, yes. I think it's like geographic either mapping or spatial modeling or might be, just guessing here, but I recall, it's something like navigation related. So they do use global navigation systems, but their primary gig is controlling graders and bulldozers and diggers. Trimble a few years ago got acquired by Caterpillar and they have highly instrumented their machines to the point where there doesn't have to be a human operator in them anymore. They can be driven remotely and they, you know what a lay this? A machining lay that turns a piece of material and a cutting blade comes down. Their motto was leaving the world and because they could create roads and spaces for apartment buildings and complexes and all of that, all that grading work in crafting the ground. So, he's going to come in and give you a talk and it's pretty cool. One of the things he'll talk about is he'll show you a machine, a digging machine and then he has another drawing that hasn't exploded view, if you know what that is. It just strips all the electronics off of it so you can see the computer, you can see the GPS receivers, you can see all the control units and it's really cool and he might bring some other stuff, but I'll leave that as a teaser for you. Then on the week where we're studying deeply embedded debugging, deeply embedded systems. I've got the application engineer for Lauterbach coming in and will give you a demonstration of their tools and their capabilities. Has anyone heard of Lauterbach? Yes. Have you used their tools before? You have? trace 32, Yeah. So he's going to come in and talk about trace 32 and some related material there. So he's coming in from, I think he is in either in Portland or Seattle, I can't remember, but he's going to come in and give a talk and that will be really cool though. When I was at Seagate, we used all Lauterbach equipment just because it just state-of-the-art in terms of debugging. It's very, very cool stuff. The last time I taught this course the introduction ended there and I just wanted to take a minute and talk, I grab some slides from later material, so this course isn't all just fluff and me talking about market sizes and dollars, blah, blah, blah. We're going to get into some things that are really important that you need to know. I'm going to go through this really quick, and we're going to revisit it in the appropriate week, just to give you an idea of the depth that we're going to go into, and security is huge. So, what does it mean to me secure? So, if you've ever done any reading about security or encryption, Bob and Alice are the good people, the nice people, and they want to communicate with each other, and E here as the adversary whose eavesdropping on this conversation between Bob and Alice and is trying to listen in on what's happening. So, Bob says, "Hi Bob", and Alice says, "Hi Bob", and Bob says, "Hi Alice and Eve is listening in on that and we want to fought that, you don't want that to happen. So, what we want to do is we want to scramble and mix this up and make Eve really frustrated, so frustrated that Eve gives up and decides not to eavesdrop anymore. We're going to look at a number of encryption techniques, just going to buzz through these really quickly. Long time ago, Julius Caesar created this Cipher, it's named after him, this Caesar Cipher and it's just a simple substitution where one letter is substituted for another character. It's pretty easy to break by today's standard. Another one is called the one-time pad and this is the so-called "perfect" encryption because it is, well, we'll get into it. It doesn't have a key as we understand it in terms of AES encryption or types of encryption that we use, that our web browsers use when it establishes a secure connection to a server someplace, but it's not practical for real world. We're going to take a look at AES, encryption, how it works, it's key lengths series of rounds it goes through to iterate as it scrambling the data, we're going to look at asymmetric encryption. Asymmetric encryption generates a pair of keys and we'll come to understand what the relationship and use of the public and the private keys are, or learn about Diffie-Hellman algorithm which is a way to establish a shared secret over an insecure public transport. We'll learn about hashes, hashes and uses for hashes. So here's an example showing how a password can be hashed in start in table and compared, skipper over that stuff. The takeaways from the security section that we'll look at is to start to develop a security mindset and it's really about thinking orthogonally. When you work in security for a while, you start to take this attitude or this viewpoint of, how can I break that? Okay, there's a boundary there, how can I get around that boundary? How can I break this thing? We're going to study methods for data integrity, has this file I downloaded from the Internet, has it been modified or tampered with by anybody? We care about that because there might be malware, and never download software from BitTorrent and those other places, unless you want your machine to be compromised. Authentication, what does that really mean? What does it mean to authenticate a couple of number of dating encryption algorithms? This one's fun. I got a book from Keith Graham, and it has all bunch of examples of security blunders, and we're going to take a look at those. The three that we'll take a look at I thought were pretty good obviously, oh my gosh. Now machine learning, again I'm going to go through this really, really quickly, this is an example of supervised learning, it's called linear regression, I guess this is like the kindergarten example, I think every or many instructors start with, but here's a dataset that has square footage number of bedrooms and the price and they don't have to be aligned in or organized in ascending order, but when I was writing the Python code for this I just made these numbers up on the fly. Just needed some data. There's datasets out there, you can go get boston. The boston dataset has real data, I don't know what year it's from, but real numbers, I just manufactured some numbers, so it's got square footage and the number of bedrooms and the price and we'll come to understand that the square footage and the number of bedrooms are features and what we want to predict is the price of the house. So we train the network, giving it this data and then we give it new data, it's never seen before and then it makes a prediction. So, there's what the data looks like now, there's two dimensions to the inputs of the prices in the x axis and the number of bedrooms to sort of sticking out this way, but I couldn't figure out how to plot it in 3D so that data isn't shown, so that's what happens to look like. We can draw a straight line through that, just put a ruler down and drawn it and then we can manually come in and pick a price and just go up to where it intersects the straight line, but that's not very fun. So, linear regression and we create this hypothesis function which is a linear combination of these weights, theta and these features, so the x1 would be the square footage and x2 would be the number of bedrooms, and these thetas represent awaiting. Instead of just figuring out by trial and error what the thetas are, we'll learn an algorithm where this process is applied by which the thetas are learned automatically. So if we arrange right out all our thetas as a column vector and all of our features as a column vector, we can represent a hypothesis of theta given x is the summation of the thetas times the x's which is equal to theta transpose times x which is shown over here and we get a real number which is the price of the house. This process uses or defines a cost function, there's many cost functions that these algorithms can use, but this one uses the least mean square, and I'm going through this pretty quickly, so we take this derivative of the J of theta that we want to minimize, and it eventually take the partial derivative with respect to theta of J of theta for all n and we get this repeat until convergence, the summation equation here. This process is called batch gradient descent, and of course, I'll go through this much more slowly when we get into the machine learning, but I just want to give you a taste of what's coming so that it isn't all fluff, of course. The consumer internet connects about five billion devices. Think about all the new business models that came about, all the changes our lives. Now imagine, we connect 50 billion devices to each other's. Devices that are manufacturing for us, that are on energy plants, that are on healthcare systems, what kind of value can they create in the world, what kind of business models can happen through this? We need businesses to rethink their business models without start looking differently at your consumer relationship, the way you price your product, provide the services and increasing UBC, you don't pay for product per se but increasing, you pay for service. You see increasing importance of these ecosystems that together start creating these tremendous new solutions that, they're medically can improve people's lives. We can really increase the utilization rate of products. Today, a utilization rate of cars is maybe 15 percent, half of the time is used to search for parking spaces or standing in traffic jams. Using data and software, we will be enabled to automate driving, you will bring up the utilization rate and I think the same might happened to the industry. We will see devices that no longer fail, or are repaired before the failure happens, and that will change the way that we buy products, it will change the way we rent products, it will change the way we think about products. In the past the focus was on automation of muscle work, in the future, software will really help us to automate knowledge work to come up with better decisions, faster decisions. The skills sets required for jobs in manufacturing will automatically change. We would see far more white-collar workers in factories than what we have today. We will see many products being offered as services. Some people estimate that the Industrial Internet of Things is as big as the entire US economy. To capture data in the right framework, engaging the right parties. Is essential so that at the end of the day, the benefits can outweigh the risk. There are some pretty significant questions that leaders within those organizations have. By coming together and sharing perspectives and understanding how others are thinking about these issues, they can collectively come up with better approaches to be able to create the enabling environment, as well as, individually walk away with more forms, position to make individual decisions. We need to rethink products and services, re-invent business models, and re-train and re-tool the workforce. Doing this and more will be important to make sure that this trend, Industrial Internet of Things doesn't just deliver the business and economic advantage, but also the human advantage. It's the people behind it that have the power to make it truly transformative.