Embedded - 70: Make Us All into Sherlock Holmes

Episode Date: October 3, 2014

Rob Faludi (@Faludi), author of Building Wireless Sensor Networks and chief innovator at Digi International, spoke with us about Zigbee, writing, and experimenting. Rob's blog Books we talked about: Building Wireless Sensor Networks (of course!) Creativity, Inc.  Make: Wearable Electronics

Welcome to Embedded, the show for people who love gadgets. I'm Elysea White, here with co-host Christopher White, and our guest is Rob Faludi, wireless sensor author and engineer. Hi, Rob. Thanks for being on the show. Hi, thanks for having me. Could you tell us a bit about yourself? Well, sure. And I'd like to point out to start with that I am not a professional engineer. I actually have a very eclectic background. I did my undergrad in theater and film. I also have a master's in cognitive psychology. And then I got another master's degree. I started

a collection at the ITP program at New York University. Is that where you teach now? And yes, I also teach as an adjunct there from time to time. But my main day job is I am chief innovator for a company called Digi International, and Digi makes a bunch of different wireless communication devices. And Digi is not to be confused with DigiKey. That's correct. DigiKey is a distributor of electronics components. And so they actually distribute some of the modules that we make.

So we have a relationship with DigiKey, but we are not the same company. They're not really related. That's not confusing at all. No, no, not at all. Okay, Digi International, and they make wireless modules. That's correct. Among other Digi International, and they make wireless modules. That's correct. Among other wireless communication devices. So we make cellular gateways, satellite routers, stuff like that as well.

So big stuff too. And you wrote a book for O'Reilly. I did. I wrote a book called Building Wireless Sensor Networks, which has some delightful dachshunds on the cover. It does. That was how we actually got talking was that I like my bird, but he wasn't what I wanted. And apparently the dachshunds weren't what you wanted? No. Well, I never requested dachshunds.

They just showed up. And I went back and asked, because you can always make requests for other animals. They may or may not be honored. Yeah, I made several requests, too. And none of them were requested extinct animals. They may or may not be honored. Yeah, I made several requests, too. And none of them were requested extinct animals. Yeah, that seems to be typical. But yeah, I asked maybe pigeons, because I thought, you know, they're used in communications. Or I said bees, you know, Zigbee bees, that seemed to make sense.

And they said, no, well, you can't have insects, because those are for databases. Which doesn't make sense with Michael Barr's ticks or whatever he's got. I think they just found something to say. But I've grown to love the wiener dogs, so they're mine now. They live in my heart. Yes, I feel that way about my bird. But, I'm sorry, what is your book about? Other than dachshunds.

So, my book isn't about dachshunds at all, which is why it was very surprising to see them. My book is about, basically, it's instructions on how to make wireless sensor networks using tools like XBee radios, Arduino boards, processing. And it's intended for anyone with a bit of a technical bent, but it doesn't assume any specific kind of experience, to go ahead and make wireless sensor networks for themselves. Also, wireless actuation networks. So two-way communications. And you said XBee. That's an implementation of the ZigBee radio protocol. Exactly.

Yes, that's correct. So the XBee radio, which is made by Digi International. I didn't work for Digi when I wrote the book. I got picked up by them later. The book is pretty, I mean, it talks about Digi specifically quite a lot. So I'm been informed that it was a good idea to focus on one physical device rather than covering a whole lot of them, that it would be a lot easier to write the book and write the code. So since I knew the XBee radios, I focused on those. And because I was using them, the whole ecosystem around them,

which includes a device cloud system, it includes gateways to get ZigBee information onto the internet and a device cloud to actually put your data on. And so those things just sort of came along with it. From what I hear, the people at Digi were thrilled, and later they hired me. It was a very good resume. It's a bit long. Yeah.

Apparently you can go over one page. Here's my audition. It's a 300 page book yeah and you build a number of projects in the book i mean there's a little bit of theory but there's a lot of here's how you put it together uh doorbell uh lighting yes so there's a the first project is a doorbell project and then i i needed to do a project that encompassed some sensing and processing, but I wanted to make it very simple. And so I invented the romantic lighting sensor, which basically goes with, it's monitoring, it's basically monitoring a sensor and then having an output. But I wanted a story around it.

And so the story is that you decide you want to have a romantic dinner. And, you know, you don't want to have your romantic dinner be in glaring sunlight. That's not very romantic. But you also don't want to go too dark because that's creepy. So this sensor was going to help you understand the optimum light level. So that gave me an excuse to have some processing. What about your sensing? So this is just using straight up a lighting sensor, a photoresistor.

Oh, good, good, good. I had this idea you were going to measure the romanticness of the evening. I wondered how that was going to go. Oh, I would love to have sensors. I feel like we don't ever have sensors for the things we want to sense. We have sensors that are left over from the automotive industry. The romanticness would be perfect, wouldn't it? I mean, that's something humans want to know.

Lighting, meh. But, you know, it's indirect, but it's something. Well, and that allows the light to turn on softly as the sun goes down. Okay. Right. And for you to set the lights in the room so that it's dim but not too dark. Why Zigbee? So I was interested in Zigbee actually for a long time because it is an implementation of mesh networking.

And mesh networking just really is a network where any node on the network, any point on the network, can act as a router to any other point. And so it makes for local area networks a very resilient type of network, where if you can either talk directly to a radio from one radio to another, but if the radio you want to talk to is too far away, you can hop, skip, and jump through other intermediary radios along the way. And the Zigbee protocol figures that all out for you. So as a developer, you don't have to do anything. You just say where you want to go,

and the robots in the black boxes figure out how to get your message there. Mesh networking has been a goal for a long time, for a lot of different types. I mean, ZigBee is one variant. How does mesh networking work? Well, actually, I'd like to address the first part of the mesh networking goal. I think there's been an interest in mesh networking for large area networks. And the idea being that there would be no one central data point, there wouldn't be any backbones that you could attack or that were run by large companies. Everyone would be peer-to-peer. And it's a lofty dream, but unfortunately, the physics of it don't work that well.

Mesh networks are really good in local area networks, but they're also inherently very chatty. And so if you try to scale them up to very large-scale networks, you get a network that really you just can't communicate. There's so much noise. Just as if you got 500 people in a room and they're all talking at the top of their lungs, no one would be able to hear anyone else. Is that because there's radio noise or is it because the way

the network is implemented, there's a lot of rebroadcasting? It's really the way the network is implemented. It's very efficient to actually have a network that's more hierarchical, which is why you see wide area networks that have things like backbones. And if you look at networks, you know, telephone networks or the human bloodstream, there are always these kind of peripheral, very dense networks and these central, very large pipes, because that's,

it turns out, the efficient way to move information around. But mesh networks on a local level are extremely useful because the chattiness isn't a problem, and you don't have to worry about doing a very resilient design because the radios will actually take over for each other. So the idea is really that you can throw some radios into a space in a non-optimal way, which is typical for residential and a lot of industrial. And they will figure out how to work together as best they can. And then basically you'll ship that information typically through a gateway off to another network. Maybe you'll log data remotely.

Maybe you'll log it locally. Or maybe you'll just connect to the internet, go up to the cloud, and do what you want to do there. So you got into this through art, or you came from a theater background? Yep, I did. I was very interested in – and actually, so my theater background, I was very interested in things like lighting and sound and set design. I did it starting in high school. And so I've always been interested in technology, but I've been very focused on technology as it applies to humans, as it applies to people. And this whole back and forth of people and devices and objects and back to people is the reason that I decided to do the master's in cognitive psychology. I really wanted to understand how that worked. Were you planning to build a brain?

No, I've never been interested in building a brain because we don't know how to make one. That's why I minored in cognitive, because I thought it would help with AI. Yeah, no, I think by the time I got to it, AI was sort of the dream deferred. We tried it. We thought, hey, this should be easy. We have computers now. And then it turns out it didn't reveal itself to us as readily as we thought. So we have a lot more to learn. But my main interest was in learning how to build

things that people could work with easily. And that led to my second master's degree, which was at the ITP program at NYU. ITP program is a media technology program. And so, it combines both the arts and technology or the sciences, if you will, together. And the point of the program is they put together people from all different walks of life, poets, artists, scientists, electrical engineers, and give them sort of a basic common skill set. And then they have two years to build projects together, hopefully projects that celebrate the human spirit rather than create startups. But startups come out of it. Some people take it back and bring it into their modern dance practice. Some people go off and become chief innovators like me. I heard that Arduino started out partially because it was a way to get into technology for artists.

That's exactly right. It actually started at a university in Italy called Ivrea, which is a design school. And Arduino was created specifically for interaction designers so that if the interaction designer wanted to build something, they could do it quickly and they could make their own prototypes rather than trying to make a design and then hire someone who could do the electrical engineering and programming. They wanted to put all the skills into – the initial skills into one brain. So the interaction designer could make anything that they needed to, and then later on they could hire professionals to make it resilient and work the kinks out of it. So I actually ran into Arctuino very early. I was lucky at ITP.

One of the instructors and my advisor at the time, a person who I consider to continue to be my mentor, a guy named Tom Igoe, had gotten involved. He knew people from Ivrea, and they sent over some. I believe it is of very limited historical importance, but still, it's got to be worth more than what I paid, the $12 I paid for it. And did Arduino make you think more about how ZigBee networks should go together and this crossover of art and technology? Sure.

I learned them both at the same time. Because they do work well together. They do work well together. The XBee acts very much as though it's a serial port between two units. Yes, the XBee radio is really popular with makers because it's got this easy-to-use mode where you just turn it on and

one radio will, by default, talk to the other radio. Maybe not the most efficient way, but you can really plug and play. And so it's been a popular solution really from the get-go. Before the XBee radio, wireless communications with microcontrollers, especially for artists and maker technologists, was very complicated. It failed a lot. You had to write your own protocols. You had to write your own error correction. And the XB, pretty much, you open the box, you plug it in, and all of that stuff is built in. So it's extremely popular in the maker community. And it's one of the reasons that I ended up working for Digi International is a lot of people, you know, the XB Radio was designed for industry.

And it turns out that a lot of makers adopted it. And so I got kind of pulled in for that. They decided they really needed a better link to the maker community. And I was brought on board with one of that as one of my goals. And so now we're passing a few years very quickly. Now you have the XBee LilyPad. The LilyPad is the Arduino variant that is usually used in wearables. Right. So the LilyPad is actually a modular system for wearables.

There is an Arduino LilyPad, there are sensor LilyPads, there are LED output LilyPads, and there is a lily pad XP designed by myself and Kate Hartman, who is currently at OCAD in Toronto. She's a professor there. And her book actually just came out. I think it's Make Wearables is her book. It came out Friday. A terrific book. And so the LilyPad XB was something that we designed while we were both at ITP because we realized that here was a wearable, a sew-on wearable electronic system that had no direct way to communicate information off of the body, off of the piece that it was in. And so we thought that this was a glaring omission and went ahead and made, because Lilypad is open source, we made a prototype. The first prototype I actually designed in a single day as part of a seven projects in seven days marathon that we designed at ITP. And so it took a day to prototype it, and then it took quite a while to actually bring it fully into existence. Got to know the people at SparkFun Electronics.

They took what we had done, did a little redesign so that it was easier to manufacture, and brought it into being there. And now anyone can sew an XB radio to their clothing. I should point out that XB, ZigBee is one of the protocols that XB radio speak. However, XB, there is also a Wi-Fi variant to the XB radio. There's an 80215-4, which basically is a ZigBee radio without mesh networking. 80215-4 is the layer under ZigBee. At Digi, we also have 900 megahertz and 868 megahertz radios, which do a little better performance but are not globally licensed. 900 is good in North America, 868 for Europe.

And we're always trying to think of new and interesting XB radios to make. So what they all have in common is the same footprint, the small size, the low power requirements, which is extremely important if you're dealing with a battery-operated system, which typically is the reason that you wanted wireless in the first place. You have something that's going to move around. But all the implementations are slightly different. Obviously, you talk Wi-Fi systems differently than you talk to ZigBee systems. So the LilyPad is ZigBee? The LilyPad is a – so the LilyPad XBee is a LilyPad board, so a round board with sew holes on it so you can sew it into clothing with conductive thread, with a mounting for XB radio.

So it's a 20-pin mounting. You can put any of the XB radios on it. And you said they were footprint compatible. That's right. So as long as you get two of the same variety, you're good. Yep, that's correct. Okay. And working with LilyPad and having worked with Arduino, do you see a big gender difference between people using them, or is it just everybody uses whatever they're comfortable with? at all. People will just get a lily pad Arduino and maybe the lily pad XP boards,

solder them together and stick them in a project box because, hey, these things seem smaller than the other components. Well, you don't even have to solder them. You can just wire tie them, which is kind of nice. Sure, you could wire tie them together and you can sew them together with conductive thread. I don't see a giant gender difference. Obviously, wearables, I think, you know, fashion is appealing to women and to girls when you're doing high school education. Actually, the biggest gender difference I see in wearables is in sewing skills. Men decide that they can do electronics, and they put some effort into it and do it. Women, just the same.

They go ahead and they decide they want to do electronics. Sometimes there's a feeling that they're maybe incapable of it. But if they work through that, they can learn it just as quickly. I don't see any real significant gender difference there. And then if you teach a class where people are actually making their own wearables, men decide that they can sew. And how hard could it be? You've got a pointy thing, you've got some thread on it, and you've got some fabric. Electronics was easy. Lots of other things they do are easy. Why not this? And women don't think this is a skill. I'm talking generally.

They think that sewing is just something that you do. Well, it turns out that the, and I ask women typically who sew, when did you start to learn sewing? And they say, well, I don't know. I was five or something. It's not hard. And so, it turns out that they have 10 or 20 years of expertise in sewing. Even if it's just the sort of sewing that, you know, my grandmother taught me how to sew on a button and not make it all crinkly. Right.

And yet, so I've always sewn on my own buttons. That's not hard. And I love watching guys learn to sew because they think they can do it, which is good, you know, lean in. But when they lean into sewing, they end up, you know, they have a piece of fabric and all of a sudden they've like got thread wrapped around their finger and they've sewn the i've done that they've sewn the shirt to their pants and themselves in the finger it's incredibly tight and now everything touches yeah and and you know cutting patterns and they realize that there's this whole universe of skills that need to be learned so i'm i'm just i can sew a little but i am in awe of people who can sew, and they're never in awe of themselves. Yeah. I know a few people who are really, really good at sewing, and it goes well beyond my buttons, and I'm just shocked.

And they're like, oh, yeah, I made this tank top this weekend. Yeah. So with wearables, the real gender difference I see is that men have to spend about 10 years learning how to sew, and then the genders would really be equal. So getting back to the technology, why do people need to know about using XB, and ZigBee in particular? Because that is what your book's about, and I'm kind of, ZigBee is different than so many of the other ones. Well, I think Zigbee and the XBee radios that are biggest sellers are the 80215-4 and the Zigbee radios. And we sell a ton of them. We sell a bunch of them into the maker community, but we also sell a ton of them for industry.

They're used in things like street lighting and medical devices and in energy, smart energy systems. Factory monitoring? And factory industrial also. They're used on vehicle monitoring, commercial vehicle monitoring, to actually get information from one part of the vehicle to another, and then usually there'll be an uplink of some sort. So what interested me about them was first the mesh networking

and that they were easy to use. And I think they're most useful in networks that are local. A lot of times people think, well, I'll just put Wi-Fi in everything. And it turns out that's very power hungry. It's very difficult to provision. You need a username and password and you have to say which Wi-Fi network you're going to be on. And as soon as you move your project to another location, you have to do it all over again. Zigbee doesn't have those requirements to it. So it's very simple to set up a network, and the radials will sort of figure out what the network architecture should be. So both for power requirements, the ease of provisioning, and cost, I think there are a lot of things about Zigbee that make it the perfect solution to certain types of problems. Wi-Fi is obviously good for other types of problems, and the same can be said of Bluetooth. Bluetooth won't form a network for you. It won't, and it might manage your power well, but it has different behavioral characteristics.

So I guess I kind of came of age doing ZigBee networking, and so that's the thing I focus on. But I always think you have to look at your application first. What are you doing? And then the last choice should be what kind of radio you're going to put in there. Yeah, I think that's a good way to look at it because they do each have their strengths and weaknesses. But provisioning, and that is a weakness of just about all of them, does that mean if you aren't provisioning systems, does that mean when you drop a new sensor in, you can't? I mean, is it a closed system and you have to send it from the factory

with all the sensors, or how does provisioning work? Well, it works differently in all sorts of different situations. So provisioning, and what provisioning means is, okay, you're going to have radios, how are you going to join them together to a network? How are you going to tell them who they can talk to and who they shouldn't talk to? And so- That's from foes. Yes. And this is something that is true of any wireless device. I guess it's true of wired devices too, but we provision them by physically plugging them in and you're done. So your headphones are paired with your laptop because you plug your headphones into your laptop and you're done. Well, if they're wireless, they don't know what your laptop is and you're going to have to give them some sort of authentication information.

You also don't want your laptop being joined by someone else's headphones who's spying on you. So there always has to be some sort of process to join the device to the right network, and hopefully only to the right network, and prevent other devices, foreign devices, that shouldn't be on your network from coming in. And this is something that can be configured at a factory. It's something often in consumer devices the consumer has to do themselves. So you have to sign in to your own Wi-Fi network or pair your

Bluetooth watch to your phone. But industrially it is something where you wish you didn't have to do it at all. You wish the device is just new. And so having a simple process to get that done is very attractive. And also, in the case of a ZigBee network, let's say you're doing

leak monitoring. You're putting the ZigBee network into a building. You're going to look for leaks so that when water starts pouring out of a pipe, you know it right away, and you can go fix the problem before it causes a lot of damage. Well, you're going to put that all over the building. You're probably not going to get a choice of putting the radios in an optimal position for a radio network. So just being able to plug them in and turn them on and have them figure out what the optimal paths are, so have them distribute routing tables, have them recover from any problems on the network. They can essentially auto-provision themselves. Now, you still have to let them know which networks they're allowed to join and which they aren't.

You might need security on your network. Zigbee comes with a security layer. We only turn it on when it's really needed because security slows everything down. It makes it more difficult to join things. But in the case of a network carrying revenue information like smart energy, it's essential. In the case of someone's interactive kinetic sculpture, probably it isn't essential and you can do without it.

So how do you provision it? So Zigbee is provisioned. So the first thing to know, in the case of the XBee radio implementation, you can just turn the radios on and they will talk to each other. You don't have to do anything. They're very promiscuous. They'll talk to anybody. Absolutely. And that is terrific for a maker who's just starting out or for someone who's learning. Because really, I think, and the reason in my book that there are so many projects, I think you want to promote early success. A lot of people, I think –

It prevents discouragement. It prevents discouragement, and I think it also creates a framework that you can hang the rest of your knowledge on. Once you've built something yourself, you understand what it is that you're doing. You probably understand some of the problems you ran into. But if you were successful, now you can become interested in learning more. And at the same time, as you learn more, you can apply that knowledge to a project that you did.

So it's not just abstractions, but realities that you can hang on your cognitive framework, if you will. We were going somewhere with that. There was, how do you provision? Ah, and I lost you there. And I totally agree. We should make it simple to start with.

And I like the idea that they talk to everyone when they come up. But if I was using this in a professional capacity, I might agree with you that security is not required for leak sensors because who cares if somebody is sniffing my leak sensors or they don't want them to spoof them that would be annoying so I do care about provisioning and you said it was fine to get technical so we'll just get technical for a moment

Zigbee networks, and this is true actually of any 802.15.4 type of radio communication, Zigbee, basic 802.15.4 or other layers put on top of that, has the ability to support what are called PAN IDs. These are personal area network IDs. And there's about 64,000 different IDs that you can subscribe to. And each radio may hear other radios, but it'll only pay attention to radios on its own pan ID. So it's like creating a virtual city. cities, we know when you talk about Broadway here, you're not talking about the Broadway in New York. And so the Zigbee IDs work the same way, that you have an ID that kind of designates what city you're in, and then you have an address yourself as a radio. So two different numbers. But if I talk to radio address three on pan ID 257, if there is another radio address three on some other pan id it won't pay any attention to me it knows now that information isn't for me and so it allows each project or product to kind of create its own little world so that's one layer of security or provisioning um it's not it ends up being security because you'd have to guess a pan ID. That's certainly not impossible.

Well, not if I can sniff it. Then you also have a channel that you talk on. So you're only on one physical channel of the spectrum at a time. That's another layer where you provision yourself. You kind of wall off other people. That doesn't count as security directly. But again, it's a way that you may not be leaking information to other things. And then there is elliptic key encryption security, and that's real security where you're actually distributing a public-private key, and each radio has that, and you have to know that key to join the network. And once you have that, all the communications are encrypted and cannot be sniffed. But passing out that key is probably non-trivial. A lot like logging in and telling

802.11 radio your password. Right. And the right way to do that is offline. Each radio gets programmed. So for example, in a revenue network where you're using that, every radio would get programmed directly with its key so that it's not ever passed out online. It's a pretty good system. I'm sure there'll be newer and better ones in the future. I'm not aware of any cracks to it. Usually, sometimes people want to implement security too fast. It's not something you want to deal with at the prototyping stage, I think. But as you said, if you're putting a leak sensor in a building, you may well want to secure that network so you don't get spoofed. You really have to look at what the dangers are.

Yeah. And it's how much does it cost for me to roll a truck to go find the leak? And that may be, you know, maybe there's a superintendent you don't mind bugging, and maybe that's a $10,000 truck. So you got to work on what your costs are, your benefits. Make sure your benefits are good. That's exactly right. And that's the only way you can really do security is you have to figure out how much you need, how much it's going to cost you, and what the alternatives are. So we do use the security, but I certainly don't advise it in every project.

Fair enough. Are the key lengths and things comparable to current Wi--fi security or is it a shorter simpler uh i i believe the security i'm not a security expert but i believe the security is uh comparable to current wi-fi okay it's it's it's uh for this day and age it's very good security okay i imagine some point in the future we'll have hyper fast computers and you'll crack it right you'll crack all those keys in a millisecond. But we don't live in that world yet, so you're still good with what we have. And I know from your book there are three different kinds of radios, three different kinds of XBee widgets. Oh, well, there's more than three.

Connectors, routers, and endpoints? Oh, in ZigBee, yes. So in ZigBee, your radio is one of three things. It's either every network has one and only one coordinator radio. And that radio starts the network. So I usually explain it, you know, if we wanted to have a club, we couldn't just have a club, someone would have to start it and say what the name was, and then other people could join. So if I started the club, I could name it the XB Radio Club, and then each of you could decide whether or not you wanted to join my club. I really hope you do.

And that joiner chooses the channel, right? Right. That sets which little tiny chunk of the wireless spectrum you're going to. The coordinator will pick a channel. And actually, this is one of the lovely things about the protocol. It'll do a scan of all the channels. Basically, it's called an energy scan. So it looks for a lot of radio energy on each channel.

And the channel with the least radio energy on it, it says, hmm, here's a plot of land that I could build a house on. It doesn't seem to be anyone else using it. So it picks that channel. And it will also pick a PAN ID. Now, you can designate one or the radio can be told to just pick one on its own. And then your coordinator radio will broadcast that out. Now, as your routers, those are the second-level radios. They're almost like the coordinator except they join clubs.

They don't start them. We'll look around. And the router radio will try to join any network that will allow it to join so when you turn a router on it looks around and says oh what clubs are out there and it looks at the clubs and maybe it has a list of clubs it's allowed to join or maybe it's been told it can join any club and so it follows its instructions can it join multiple no only one at a time. Okay, so how does it decide whichever is first? It will typically pick the first one that it comes to that it is allowed to join. So often, if you're provisioning a network, you're going to set a PAN ID for that network and tell the coordinator, okay, I only want you to broadcast this PAN ID, and tell your routers you can only join that PAN ID.

Another possible layer of security is you can turn this joining process on and off so you can have it started basically press a button on the coordinator and say hey now you can people can join your club but uh tomorrow morning not so much not tomorrow morning you know typically in you know much shorter time frames maybe a minute maybe maybe 60 minutes, no more. And at that point, hey, you can sniff the network, you can do a bunch of things, but even if it's unencrypted, you can't join it because joining has been prohibited. That takes care of my spoof problem. It does take care of your spoof problem. And it's a way of securing the network enough in the most efficient way.

You may not need to encrypt the information on it. The information may be nothing's leaking, nothing's leaking, something's leaking. You don't care. You can look at the outside of the building and find that out, right? So depending on what type of information you're passing, you may not need encryption. But there are other ways to essentially form a network that's unlikely for someone else to be able to join it or look at the information, if you will. And then you have one more type of device, and these are end devices. In Zigbee networks, end devices are allowed to sleep. So they'll go to sleep, and when they turn on, they will pick a parent device, which

will be a router or the coordinator. The coordinator is just a special type of router. And the end device then says, hey, okay, I want to join a club, but I sleep a lot. So I really can't get messages from this club. I don't want to be an officer. Yeah. So I'm, I'm going to, I'm going to pick another radio, a router as my parent, and that router will store any messages for me.

And when I wake up briefly, I'll check and see if there's any messages, uh, waiting for me. I'll go look in my mailbox and I'll send anything out that I have to send out and then I'll go and see if there's any messages waiting for me. I'll go look in my mailbox, and I'll send anything out that I have to send out, and then I'll go back to sleep. And that router basically will act as my proxy on the network while I'm sleeping. And Zigbee has, within the protocol, is able to designate a way to handle situations where the end device wakes up, it can't find its parent, but there's another potential parent available. So if it's orphaned, it'll go find someone else to be its mommy and take its messages for it. The routers have to be on all the time.

In ZigBee networks, yes. And can the routers, by virtue of their name, I'm getting the idea that maybe they route things. They do. They must be able to route from the endpoint to the coordinator. They can route to other endpoints and other routers as well? Yeah. So routers can, and they will configure themselves. They will pass a message along that they receive to, according to their routing table, the next hop, whatever that is.

So the routing tables are figured out automatically and passed around automatically. How often do they change? They change as often as needed. Now, this is the thing where you can do configuration. You can let them just chatter away as much as needed, which on a small network is fine. As your network starts to get around 40 devices or so,

that chatter can overwhelm the amount of bandwidth available on the radio channel. small network is fine. As your network starts to get around 40 devices or so, that chowder can overwhelm the amount of bandwidth available on the radio channel. And you don't, of course, want to do that. So then you might set up a system where everybody, depending on the type of application you're running, you can set up a system where only the inbound routing tables are available. So if you're making a sensor network that's going to send everything to a base station, then every radio will just learn how to get to a base station and store that information. And you can do the same in reverse, basically, where if mostly what you're doing is sending from a base station out to radios, you can actually build all the routing tables on the base station. So the important thing is it works automatically on small networks.

You don't really have to worry about it. As you build a large network, you may want to take a look at my book or some of the stuff online that's available to look at some of the more sophisticated configurations. These things are running real routing protocols and dynamically updating tables when nodes leave. That's pretty impressive, actually. Yep, and they're doing that on very simple chips with very low power so the whole protocol is designed around what's

called low bandwidth networking now low bandwidth isn't that low bandwidth you can go up to you can talk to up to about 115k uh through this radio and so for things like bits per second. Yeah. So for sensor networks, that's great for something that where you're talking to a vending machine, the vending machines telling you hasn't been tipped over. I'm out of Skittles. I have too many seven ups. No problem. Those are very small messages that happen? Too many 7-ups? Nobody buys 7-up anymore. I don't know. Sorry. Okay. I'm out of Coke. I've been tipped over. I have no Skittles.

Right. Any of those things can be passed along, but those are small messages. You can send that message with a single number if you want to. So low bandwidth and everything about the network has been optimized for that. So here you've got these sophisticated routing tables, security, encryption, all of it running on very low power with very little bandwidth. Because it's been optimized for that, it's not a good network for sending video or streaming audio like this podcast over. You can get some of that done, but typically you're going to want to look at another protocol that's been optimized for that. It's maybe going to use more power, it's maybe going to need more expensive chips but it's going to do the job that you need it to do.

So I'm getting the idea of the coordinator and the routers and the endpoints and I understand the endpoints can sleep but it sounds like the routers have to be on all the time and they're going to chatter a lot so those have to be powered. Right, well the reason the routers have to be powered is because they don't know when they're going to chatter a lot. So those have to be powered. Right. Well, the reason the routers have to be powered is because they don't know when they're going to receive a message. If the endpoints could say, I mean, if it had 10 endpoints and it could tell all 10 endpoints,

only talk to me once every 10 minutes in this time slot, could the router go to sleep? Then it can, yes. And so the thing is that requires pretty accurate clocks because as you drift away, you won't be able to talk correctly. So you bring up an interesting point because one of the protocols that XBee radios speak is a proprietary protocol called DigiMesh, named after the company. similar to Zigbee, and it's good at some things. It's better at some things that Zigbee isn't good at, but it also is a proprietary protocol, so you only can buy XBee radios to talk Digimash. Nothing else talks it, which is fine if you are making... It's fine for you. Well, it's fine for us in a way.

The reason that we do both is that proprietary protocol is great as long as you know you're only going to be talking to your own brand of radio. If you are putting a smart energy system into a home, you'll probably want to talk Zigbee because you may be talking to other manufacturers, and DigiMesh can't do that. But what DigiMesh does, it actually has what you're describing. So all radios are routers. There's no hierarchy of three, if you will. And they all can sleep, which means they all save battery power. And they'll all wake up at the same time and go to sleep at the same time. And they have a protocol for figuring out if they've gotten out of sync and coming back into sync, which means that one of your DigiMesh radios might drop off the network for a little while because it got colder and so it's crystal is running it a little slower and eventually it falls out of sync and then it will stay awake for longer and longer periods until it sees other radios and get back on the network. If that works for your application and it works wonderfully for applications like street lighting

where you don't have to have immediate control messages are passed around the system but you you know are looking at seconds a minute, a minute, it's okay. A minute. Hours are fine in a street lighting network. It's not that big a deal. If a radio is offline for an hour in a street lighting network, probably there's no effect at all. So you're not worried about that.

Because the light goes on and off, it just doesn't coordinate with everybody else and it doesn't tell you if it's got an error, which it only has an error once a year. So who cares? It's out for an hour. So a system like that is very fault tolerant. And as a fault tolerant system, DigiMesh is a great implementation for that. We don't expect it to be offline for an hour or even 10 seconds, but knowing that it can happen and understanding what the repercussions of that might be are very important when you're

designing a system. So DigiMesh is good at a lot of things but won't talk to anyone. No one else makes a DigiMesh radio, so we can't talk to anyone else. And there's some good things, great things in the ZigBee protocol as well. So it's really picking the right protocol for the right job. Okay. You know, it's the fight between having radios on all the time and calling it low power and mesh. And it's like you can have two of those. You can have low power and mesh, but it's really hard to get them without better crystals and to get them to have cheap implementations. Right. Low power, low cost, and low bandwidth are all desirable. But the only way to implement them is to sort of implement them in a way that works properly for your application. Remember, all the

end devices in an ISIGB network can sleep. And many networks are just end devices and a coordinator. Let's say you're doing moisture monitoring in soil, and they're all in range of your coordinator radio, then everything else can sleep, in which case all of your moisture monitors can be on battery, and you have no problem with that. Well, and if we go back to the leak detection in a building, every floor probably has a router. And then you – this isn't a problem. You have one thing empowered in a wall socket, and everything else is on batteries and only needs maintenance once a year. Exactly, exactly. So you change the batteries once a year, maybe open a panel under the sink, whatever it is you have to do. But the rest of the system that needs to be on all the time is there and no radio ever falls out of sync because the radios you need to be synchronized are always on.

So it's a wonderful solution for something like an apartment building. And do the radios, I mean, two of them come up at the same time? Do they fight? Sure. They do the automatic back off with random numbers? Exactly, yeah. So they have what's called collision detection. So if two radios speak at the same time, they will understand that they've made a mess of it. They will stop. They will wait a random amount of time and then speak again. And because they wait random amounts of time, they're very unlikely to speak simultaneously again. And they'll do this for a few cycles so that basically the message is guaranteed to get through. Collisions are a normal part of networking. They're always, even on Ethernet or TCPIP, there's something that occur all the time, and they just need to be handled.

But this is one of the reasons that having a whole bunch of units in close proximity becomes difficult. Yes, it becomes difficult, especially if they are on and talking a lot. At a unit. So if you have a network where, say, let's just go back to our soil moisture monitoring. Let's say each soil moisture monitor reports once every five minutes. You could have tons of them because you're really not going to have very much traffic. You're going to have a little blip out of each one of them every five minutes. No big deal. Unless you turn them all on in the same instant. Well, yeah, but the collision detection will help you with that. Eventually you'll sort that out, yeah.

So, but the less they're talking, the more you can have. If they're talking every five milliseconds, so basically constantly, then you will only be able to handle fewer inside your network. And there's strategies to deal with that, of course. Well, that's a bandwidth problem. You just have to acknowledge what your bandwidth limitations are and use them properly. Right. How do I choose the XB antenna?

I've heard that's kind of a tricky thing. Wow. Radio antennas are a black art in the first place, but I can tell you a lot about them. Yeah, they're absolutely black magic. We have radio engineers, and I understand a little bit about it, and I understand why. It's not really black magic. We have radio engineers, and I understand a little bit about it, and I understand why. It's not really black magic. It's just that you have to understand a lot of different things

and then think very carefully about how they're going to come together within a specific application. And that's what makes it kind of an art. It's not that we don't understand radio energy or how to design antennas. It's that... Explaining what a cartoid pattern is is not trivial. No, no. We only have 2D. Yeah, I shouldn't trivialize the math and the physics involved with it. But even knowing those things, the real art, I think, to antenna design and to radio frequency

engineering is understanding all the trade-offs, what you want to get accomplished in your application, and which set of trade-offs is what you want to get accomplished in your application, and which set of trade-offs is going to optimize for that particular application. But I can talk about antennas much more simply than this. The XBee radios come with a few different types of antenna. There's a wire antenna, which is just a little piece of wire sticking up.

It's a quarter wavelength. It's extremely efficient. It's omnidirectional and only a little bit fragile. So if your radio is sitting inside a box, no problem having a wire antenna. It's not fragile in that situation. However, if you're going to get a lot of mechanical interference with it, in other words, if someone's going to bend that antenna back and forth, back and forth, back and forth, for example, on a wearable, then you'll probably be more interested in one of our printed circuit board antennas, which actually etches the antenna right into the circuit board. There's nothing external to bend back and forth, and those antennas are almost as good as the wire antenna. We also have connectors for – But they have a directionality to them. But they have a directionality to them.

They do have a directionality to them. When I wrote the book, there was a chip antenna, and that chip antenna was highly directional. That's been upgraded to this new printed circuit board antenna. The printed circuit board antenna is less directional and operates more efficiently. So it's actually, it's not such a stark choice between the two anymore as it was probably when I wrote the book. And then there are external antenna connectors too. The external antennas are approximately as good as the little wire antenna. However, because they're external, you can put them in situations that are better.

Outside your metal box. Outside your metal box, up a little higher in the window so that it's not down behind the metal window sash. You can turn it the right way. In some cases, you need that external antenna to properly make FCC requirements. If you have multiple radios inside the box, for example, you may not be able to have all your antennas inside. So all of those are useful. I always tell people, if you don't know, start with a wire antenna. That's a good way to start, and you'll get the job done.

Definitely for makers and folks who are just trying it out. That's right. Although that board antenna sounds pretty nice. It is pretty good, and I have to take my XB radios with me, so I've been skewing more and more towards that type of antenna because it lasts forever. Yeah. Even though I'm shoving it in my bag and taking it out and bending the pins and that sort of thing. What kind of range do you get from these things? So typically, there are two models of most of the XB radios, a regular and a pro, and the pro is higher power.

The regular model works about 100 meters outdoor line of sight, perfect conditions. Typical implementation, about 30 meters indoors, and they're mostly used indoors. So that's going to cover a small house, a set of rooms, that sort of thing. Well, you just have to drop a router house, a set of rooms, that sort of thing. Well, you just have to drop a router someplace. If you're not getting good connection, you drop a router. Right. You can always add more router radios to your network. The pro radios are line-of-sight best conditions, about 1.2 miles, about 2 kilometers in perfect conditions.

In a city, in a building with walls and people and other radios, that might take you down to 100 meters. So it's like a football field area. It's pretty big. Your performance will be different in a stucco house that has the chicken wire grating inside the walls because that acts to filter radio energy. But in most cases, the Pro will take care of kind of any project needs that you have in a local area network. They are a little physically bigger, they cost a little more, and they use more power

because you can't get longer range for free. You've got to put more energy in to get it to the places that it needs to go. So you only want to use the Pro radios if you actually need them. If it's a situation where you're putting together some musical instruments that you're going to play in your bedroom and you want them to talk wirelessly, the regular radio is going to be fine and it won't run your batteries down as quickly. Makes sense. And so now going back to your book, actually. Great. Who is your book, actually. Great. Who is your book directed towards?

We've talked a lot about makers, and now we're talking a little bit about professional. So the book was written so that it wasn't intended to be an intro to everything. So if you've never touched Arduino before, you've never written a line of code, you've never soldered anything, the book is still going to be useful, but you're going to need to get a couple of other books to go along with it. So you would certainly want to learn – maybe getting started with Arduino would be a good way to get going. We have the code, and I walk through all the steps required, but if you really want to understand the whole domain I didn't take that on inside the book so the book is a good it's not the 101 course of making electronics but it might be the 103 or something like that

if we're talking in college terms or maybe a 200 level I guess it's intended not to be intimidating and I try to take people through with very clear explanations. And it is very tactical. It's step one, you do this. Step two, you do that. Yeah. And they were clear steps to me.

Oh, good. I probably do have, you know... So... I may not be a beginner. Near, you may not be. Well, we're all beginners in some way. So the book was also, I tried to write it so that it was possible for a beginner to actually go through, do all the things, an adventurous beginner to go through and do all the things in the book that they need to do. I don't explain how to solder in detail. I give two tips and a link where you can learn more.

But if you're an independent thinker, you'd have no problem with that. But I also wanted it to have enough information density so it would be useful, I think, for people like you who understand something about embedded engineering, but had never done any radio frequency stuff to get going. So it was a pretty gentle introduction as far as that went. Good. Okay. So I tried to walk that line so that there was enough information that a professional could make use of the book, but that it would not be intimidating to what I describe, I think, as an adventurous beginner. Fair enough. And we had a show recently about where we talked about my experiences writing a book.

And apparently, from the email I got from the listeners, I made it sound really hard. What did you think of writing a book? And you're not for O'Reilly, so there are some similarities here. Had your listeners written a book? No, well, I mean, some of them think about writing a book in a way that many people think about writing a book. I may have been just a little too hard. Yeah, so, no, I don't think you were. Not having heard it.

It didn't seem so hard when I started, although I had heard some things. Both my sister and brother-in-law are professional journalists and have written books. And so I know a little bit about how the sausage is made, even before I tried making my own sausage. It's a struggle. It's more work than you think it's going to be. It does not pay extremely well. So you're, you're, you feel burdened by that. But sometimes you get hired in jobs you might never even have considered existed. Oh, having written a book, I recommend writing. I'm I'm not sure. And it takes much longer than you think it's going to take. There are people who set a schedule for themselves and write a book on that schedule. I think it's a set of – probably Stephen King is the only person who does this, but he exists,, it's going to take longer, it's going to take more soul searching, and late nights, and doing things that you don't like doing, and, you know, interviews with despair.

At some point, I went and visited- Interviews with despair. Yes. Long, long interviews with despair and self-doubt. I went to, about midway through the book writing process, up to see my sister and brother-in-law and they asked how the book is going and I said, oh, well, you know, not so good. It's taking real longer than I thought

and it's really a lot of work. My brother-in-law said, so, do you feel like you're not worthy of being a human being? And I said, no, of course I am. He's like, yeah, you're not a writer yet. That's how you tell

your professional anything. You hate yourself enough to question every move you make. Now, having said all that, by the time I actually figured out how to do it, how to kind of schedule myself, how to keep working when I didn't feel like working, got the whole thing done and had a book party and all my friends came and talked about how great I was.

And then the book came out and I was offered jobs and I get to be on podcasts. And, you know, this is terrific. So, I definitely recommend having been on the other side of it. I'm often asked if I'll do it again. That was actually my next question. I'm reluctant because of the interviews with despair. But on the other hand, I made it to the other side once. So perhaps again. Would you update your current book? Do you have ideas already?

Oh, I have a lot of ideas. I think that's where things start going downhill is when you start getting these ideas and you're like, oh, well, I mean, that'd be cool. So what ideas? Well, I think there's a need for – and there are some people who have written books for the Internet of Things. This book obviously has a bunch of stuff in it that applies. I think actually the technical kind of erosion is the thing that probably would need the most attention first. The book has stood up pretty well. It was written in 2010. It's 2014 now.

That's like 100 years in internet time. I feel lucky. The radios are still made. Most of the components that I used in the book are still available. Arduino is still running and you can buy similar boards. The part numbers for things have changed, but eventually that'll get to the point where some of the code you have to go back to earlier versions of processing to make it run because they've taken useful things out. That's the sort of thing that's going to happen. So there might be a need at some point for an update just for that. I'd love to write a book, however, about techniques for better user observation or just observation in general. So I wonder if I'll update this or if I'll just go in a whole different direction. Is this where you censor up your

users and figure out what they really do like and what they really don't like about your systems or something else? No, see, the problem is it's really worse than that. It has nothing to do with technology at all. It's all about techniques that you can use to become a better observer of the world. Just another one of my many interests. I've had this idea, and I really have been meaning to write a blog post about it. I've actually been meaning to write a blog post that I'd get paid about. So, you know, we'll just ruin that right now.

I have this idea of putting a BeagleBone or a Raspberry Pi and then a couple of cameras. And a couple is important here because one of them would be like a no IR filter camera so that you could see the IR. And you'd put the regular photo, regular image on the screen in black and white and then diff it. Just take the difference between the no IR and the regular photo, regular image on the screen in black and white, and then diff it.

Just take the difference between the no IR and the regular, and you'd put that difference in color. And then you could see what animals could see, the animals that can see IR. Because right now when you have an IR camera, you have an IR LED and you shine stuff, and that's not what animals see. And then I want to do one that has UV with the same sort of setup,

where the UV is shown as a difference between what the animals would see and what we would see. And then I go on to all of the other sensors that I can consider. And I want to do, you know, electric fields and magnetic fields. And no, I don't really want to talk about the physics and the possibility of that. But I have all of these how do animals see observable things. Is that where you're headed? Because if so, I have a lot of ideas.

Well, it's... Or were you going somewhere entirely else? I was going somewhere else, but that may be changing now. Yeah. Oh, okay. I'll tell you the background of it. The background is that I teach and people have to come up with project ideas. And it's not infrequent that someone comes up with a project idea and they are going to invent some sort of better dog walking service and a dog walking device that makes the whole dog walking experience so much better in some way.

And they start talking about the technology for it and how it's going to change the world. And then I usually find a breaking point in their explanation and ask, well, have you watched people walk dogs or walked a bunch of different dogs yourself? And so often the answer is, well, no, I haven't yet. And so- I've walked my personal dog. I've walked my personal dog. I've walked my personal dog. And there's so much you can learn just by watching the world and by watching people do the thing that you're interested in that you'll come up with better ideas. You're going to make us all into Sherlock Holmes.

A little bit, yeah, I think. I think you really need to be to be a good applied technologist. You have to be able to watch people, watch what they do, and really be able to see it and kind of take your own filters away. And you need techniques for that. So I've tried this as kind of an experimental class once already at ITP. It's something I'd like to teach again. And it incorporated techniques like drawing as a way of getting closer observation, data collection as a way, a different way of observing a situation in the world, looking at people, ethnography, market research tools like

focus groups are a really good way of doing a different kind of observation and putting all those things together before you design your project so that you really have an understanding that goes beyond your internal observation of the world to one that's a lot broader and hopefully more applicable and making better projects that way. So that's my goal with the observation thing. Now, if I was actually making something for dogs, I think being able to see as a dog sees would be essential. Well, and it's easy because they do seem different colors.

But yeah, okay. And I understand where you're coming from. I was reading Ed Catmull's Creativity, Inc. He's the CEO of Pixar. He was talking about how they ask everybody who joins, I think it was everybody who joins, to take a drawing class to help them understand that the chair, when you say draw this chair, everybody just draws their

canonical mind chair and not what they see. And it's to show the difference between what you see and what you think you see. And it sounds like you are exploring how to get people to observe instead of just thinking about observing yeah that's exactly right and and also other techniques ways of ways of exploring the world that way one thing that we did is we went on a uh a random uh a random walk uh through the city uh so that uh i have a little technique where uh it it works well in new york city where there's pedestrian signs but when you come to each corner you can use the signs to tell you which way to turn next or whether to go straight. I've written that up.

It's on my blog somewhere. And the idea is just to take yourself to new neighborhoods and walk down streets that you haven't walked down before as a way of just seeing new things. Because exposing yourself to the world in a way surprising yourself a little is a great way to get new ideas. It's a great way to understand things. Because exposing yourself to the world in a way, surprising yourself a little, is a great way to get new ideas. It's a great way to understand things and just put you into a more creative space. So there's lots of good that can come out of that. And it's excellent exercise. Well, Chris and I have this thing where every weekend

we have to do something we've never done before. And whether it's going to a museum we've never been to or just going to take a walk in a new place. It's cool. Travel broadens the mind. Yes. Yes, yes. Oh, and speaking of that, you are normally based in New York.

I am. But you are in Silicon Valley today, this week. Yes, I am. I was actually here. I spoke about the Internet of Things at a conference that Boston Financial had.es Research Center just before coming here, talking to some people who are putting XBs in space. They are sending them up on... Wait a minute, wait a minute.

You really have to do the voice. XBs in space. I think that's you. You can do that, right? No. All right, okay, so XBs in space. I'm sorry. XBs in space, space, space.

I think maybe it's something you can add it in post. Right. So, um, uh, right. So they are sending XB's on a sounding rocket initially. And I didn't really understand the project till I went down and talked to the people doing it. And it's, uh, I hope I summarize it well, because it just happened. But essentially, they work for NASA. NASA has the ability to send radios or anything into space, but they have different ways of doing it. And one way is this sort of hardened, guaranteed to work,

tested in all the different pressure chambers methods. And those are called low-risk missions. They've optimized everything. Anything with human spaceflight is done in this low-risk mode because you're not allowed to just blow people up. You have to have a very low failure rate. But on the sounding rocket, they can do a little more what they want. So they spend less money up front, understanding that there's higher risk. So they try to mitigate the risk, but you don't have to have everything tested perfectly. Any failures of the experiment, they're undesirable, but they won't be critical to the mission. You basically have to rule out any failures in your experiment from endangering the launch vehicle or anything with safety um but uh what that allows you to do is to try new things without spending an arm and a leg on it which is very important because uh nasa is limited by uh grants by funding and so they can get the most done if they're willing to take on a little more risk in the experiments and one of the ways they try to do that is by – there's a mandate to try to use as much commercially available technology as possible.

So they're always trying to incorporate commercially available technology so that everything done for spaceflight doesn't have to be invented at NASA. The group that I was talking to is actually doing that. So they have limited funding. They're putting something together for one of these higher risk missions where you don't have to do – everything doesn't have to prototype things quickly. Code is written by people at NASA who are interns. Again, not their star coders one would hope are on the manned missions, but they want to get more people involved, and this is a way of doing it. And so they decided to work with maker parts. And I sat in their office and watched as they showed me what they were sending up. So the idea is they want to put a Zigbee network on a manned spaceflight, unmanned in this case,

sorry, spaceflight mission. They're going to be deploying a parachute, a new kind of parachute. They want to take data off the parachute, but they don't want to run cables down the wires. That adds weight. So they think wireless is a good way to go. There are two schools of thought in NASA. There's some schools of thought that say, you know, you'll never have external wireless wired networks are always going to be better. taking data from different parts of the launch vehicle and then transmitting that over Zigbee radios, specifically over XBs that they're putting in space, space, space. And then they're taking all that data from it, passing it to a radio that's connected to the Iridium satellite network. And that's how they're going to get the data off. The sounding vehicle goes up in the air over the Atlantic Ocean, and then it will go up about 200 miles. So it's in space, but it's ballistic. So then it turns around and heads back down into the ocean. They're testing this new parachute, but the

parachute is one that's intended to operate in low air density, so in the upper atmosphere, to break the vehicle. They'll do their experiment, but then the whole thing ends up in the ocean, so they won't recover the rocket itself. For their experiments, they are actually plugging XBs into Arduino shields and into Arduinos and sending the Arduinos into space, space, space as well. The whole thing is maker parts. I was blown away because I think it is such an amazing story that this turns out to be a useful way to do prototyping in the space program

as well as in people's homes. What about cosmic rays and space hardening and all of the reasons that satellites are truly, truly god-awful expensive? Well, those things will all have to happen if this technology is adopted for low-risk missions. But in this case, what they're trying to do is learn. It's proof of concept. This thing is only going to be in space for a limited amount of time. I don't know.

It's something like 10 minutes or 5 minutes or something like that, and then it's going to come back. Enough to hit the top of the arc. So there'll be some shielding and there's shielding on the radios already but not necessarily for outer space where you have a ton more radiation. We talked a lot about that. I now know a lot more things about

extra geo radiation. Space radiation. There's a lot of it. Yeah, there's, there's a lot of it and it's terrible to electronics. Oh yeah. And, uh, but, uh, this is something they, they think they can deal with. Um, and like I said, it's a high risk mission. So if they learn that the shielding is just no good at all, and they only get a little

bit of data, well, that's still value to the mission because they, now they understand something more about it. Uh, but they are hoping that it works well. And they think that it will, and they're obviously running the numbers on it to make sure that they have an excellent chance at it. But I was just fascinated to see maker parts going into space. Yeah, it's amazing. And it lines up with the whole CubeSat thing that's happening, too, with the low-cost satellites that are just being basically hand-tossed off the

ISS. Yeah, and the CubeSats are exactly the same idea that if you go smaller and faster and cheaper, you can cover a lot more ground, a lot of ground very quickly. It's a system that works very well, I think, in

low-Earth orbit, on sounding rockets, that sort of thing. If you're going to send something to Mars to do a multi-year mission and you're going to spend a few billion dollars on it, it better work. I think that... And if it's safety critical, I don't really want it to be bought from SparkFun. Sorry, SparkFun.

I do love you. And we just lost a potential sponsor. Well, I think the people at SparkFun would agree with you that if you're going to use it for something that's safety critical you really have to go through the whole process of and their their parts uh some of their parts may be perfect for that but they're not certified uh and haven't been tested for that sort of thing so uh i love them too and they make excellent stuff and uh i will encourage them to sponsor your show um next I talk to them. However,

if they designed their products to be totally hardened for space radiation, they would cost a lot more and it would have no purpose on Earth. Fair enough. Yes. And they do currently have quite a bit of purpose. I think we are not only out of time, we are almost bumping into my next appointment. It's been really good to talk to you. Is there any last thoughts you'd like to leave us with? Well, I always like to encourage people to make things because I have to say that

I have seen so many amazing projects come out of the living rooms of people who are just trying electronics for the first time. I think the message I'd like to get out there is it's much easier than you think it's going to be. The price has come down. The amount of information available has gone way up. And if you've got what you feel is a brilliant idea, go hands-on as quickly as possible. That is great advice. Thank you for being here. Thank you for having me brilliant idea, go hands-on as quickly as possible. That is great advice. Thank you for being here.

Thank you for having me. Thanks, Rob. My guest has been Rob Faludi, Chief Innovator at Digi International, Faculty of Interactive Design at the School of Visual Arts in New York, and author of O'Reilly's Building Wireless Sensor Networks. You can contact him on Twitter at Faludi, or check out his blog on http://faludi.com.

That will be in the show notes. Thank you also to Christopher White for producing and co-hosting. And thank you once again to Punky for suggesting a guest. Finally, thank you for listening. If you've got questions, comments, or suggestions, email us, show at embedded.fm or hit the contact link on embedded.fm. The final thought for this week, I think Tim O'Reilly, because, you know, we do have O'Reilly authors here. And he said, the network is opening up some amazing possibilities for us to reinvent content, reinvent collaboration.