The three pillars to fiber network automation success.

Hello everyone and welcome to today's Fiber Broadband Association webinar, Maximizing Fiber Network ROI with Automated Planning and Design presented by IQGeo. I'm Trish Ehlers, Vice President of Membership, Industry Affairs and Operations for FBA. In today's webinar, industry expert Kevin Nguyen explains how fiber organizations can and should automate their fiber network planning and design process. Kevin will explain the three pillars to automation, accurate and usable data, better design rules for network architecture and costs, and a calculation engine that can analyze the complex technical and financial aspects of a new network build. He'll also share an example from the market of how a fiber operator created workflows to automate and optimize the planning process to increase their ROI. Finally, he'll discuss some potential pitfalls as there always are to ensure to avoid planning success with fiber project rollouts. First, FBA is proud to be a part of the group of leading broadband industry associations which have come together to launch a series of monthly educational webinars for state broadband leaders, NTIA, communities, ISPs, and key stakeholders focused on key topics related to the $42.45 billion to the $42.45 billion NTIA BED Broadband Infrastructure Funding Program implementation and deployment. On the next webinar scheduled for tomorrow, Wednesday, December 14th at 1 p.m. Eastern, CCA, NRECA, and NRTC will discuss some of the things states will need to consider when crafting their state broadband plan. Given the NTIA BED rules require states to address potential barriers to potential barriers to potential barriers to deployment. Join us for a discussion of permitting and ROW issues and gain valuable resources and tips for making your broadband deployment projects go as smoothly as possible. Fiber Broadband Association is also excited to announce that registration is open for the first of our 2023 Regional Fiber Connect workshops to be held at the Raleigh North Hills Hilton in Raleigh, North Carolina, on Tuesday, February 7th, February 7th, 2023. You can find out more about this event and the other events scheduled for next year on the FBA website under the events slash regional Fiber Connect page. And finally, before we get started, just a few housekeeping details. All attendees are in listen mode only. If you have a question, you can submit it at any time in the questions box on your control panel. We'll have time at any time in the comments box on your screen. We'll have time at the end for a Q&A session with Kevin. Today's webinar will be available on the FBA website for members only under the events slash webinars tab within 24 hours. And also, don't forget to join us tomorrow at 10 a.m. Eastern for Fiber for Breakfast for Strategy for Winners. It's all about experience. This will be with Matt Collins, Executive VP of Commercial Operations and CMO at Calix. To hear insights about new managed service models broadband service providers should consider that will provide exceptional results. And finally, I would like to introduce our presenter today, Kevin Nguyen. Kevin is head of ComSoft Americas, an IQ Geo business. He currently leads ComSoft's business in the North American market, working with network operators and engineering firms to help them improve their design process. This includes working with planning and engineering firms to help them incorporate ComSoft fiber into their workflows, enabling them to save time and money. Prior to ComSoft, Kevin worked in technology-focused strategic roles, helping drive business and go-to-market strategy. Kevin has a bachelor's degree in electrical engineering from Western University, a master's of business administration, excuse me, administration from HEC Montreal, and is a professional engineer in the province of Ontario. He is active in the broadband industry, speaking at industry events and sitting on the FBA's technology committee. Welcome, Kevin. It's great to have you here today. And now I'm going to go ahead and turn the slides over to you so you can start your presentation. Okay, thank you so much. So, yeah, I'm Kevin and I'm with IQ Geo. I'm responsible for ComSoft's business in the Americas. So what are we going to talk about today? One second here. Always get to start like this. So what are we going to talk about? Well, after the intro, we're going to talk about why you want to use automation in your planning and design process. After that, we're going to get into what we're calling the three pillars of automated design, which is essentially the three main components. And we're going to spend the most amount of time on a case study we have here looking at how we can maximize our investment and get a beneficial ROI. And then we're going to finalize with some of the potential pitfalls of using automated design platforms. So before we get into it, let's talk about ComSoft. So ComSoft, we're a GIS-based solution for automating and optimizing fiber network designs, estimates, and plans. And using our software, we see our users, they get up to 90% savings on their design time. And that's through the automation that we have. But in addition to automation, we also have optimization. So things like optimizing our routes, equipment placement, equipment selection for your outside plant design. And we actually see our users saving up to 10 % of the capital spend on a given deployment. Earlier this year, we joined IQgeo. And that allowed us to expand our business in telecom, but also in other utility markets. So as a company now, we have over 150,000 users, well over 200 staff. And I don't even know the number for customers now. But we also have a nice global network of resellers, partners, and of course, customers. So typically, ComSoft was focused in planning and design. But now it's nice as a company being part of IQgeo, we can also have offerings in the construction management, the sales and marketing, and of course, the system of record. So we're really able to offer that end-to-end platform. So let's get into it. Why use design automation? Well, the traditional way of designing a fiber to the home or a small cell or any type of fiber network has been using CAD or maybe another GIS platform. And manually clicking and adding attributes and doing it in a, I don't want to say slow manner, but a time-consuming manner. Now, of course, this also requires someone who's skilled and understands how to build these networks. Now, doing it in its manual way, it's not wrong, but of course, it means you're going to have errors. It's very repetitive. And it takes quite a bit of time. Now, you combine this with what we see in the industry or just the general workforce overall. So the lack of skilled workers and the increased rate of wages. And it really lends itself to the idea of using automation. So using automation to design your fiber network, it means that you're going to be able to a lot faster get an output for your planning team, for your engineering team, and allow them to go and make decisions on that network deployment. It also means you can look at a given area, and we'll see this later in our case study, and compare different architectures, compare different costs. And this is something that you couldn't traditionally do with manual design because it was just too time-consuming. So what are the three pillars of automated design? Now, these three components we thought would be interesting to talk about because many of you may not have familiar experiences with automated design platforms. For some of you who do have these experiences, you might find this quite intuitive. But we thought for when we're looking at the case study and just in general, it would be interesting to learn more about these. So the three pillars. Well, the first is data. The second is design rules. And the third is your calculation engine. So let's look and see what each of these mean. So data. So data, we're talking about GIS data here specifically, is extremely important when you're trying to do an automated design on a given area. Now, the more data you have, the more representative your design is going to be of that area. So things, you know, like where you have existing assets. But that's only if your data is good, right? Because it's a garbage in, garbage out scenario. So if you have data, but it's not reflective of what's actually in the field, then it's only going to make your automated design poor in the end. So you can actually start with very simple data. Things like your streets and your demand points that you're serving. And augment a lot of this. And still get out a fairly decent automated design. But you can also start with very complex and detailed data. So things like your access structures, your handles, your manholes, any of the existing assets you have, whether it's cables or pipes or any equipment. And you can even put in things like where you have right away, railroads and interstates. Use these as things you might want to avoid or do horizontal directional drilling on. So it really is dependent on what kind of data you have. And then also, as we'll see, how I would say how complex the calculation engine is that's able to use this data. So we start with our input data. And then next, we apply what at least at Comsoft we call our design rules. So what are these design rules? Well, the design rules are your architecture, your materials, and your material and labor costs for your network that you're going to design. So when we say architecture, we mean things like, is it going to be a centralized split, a distributed split? Are we going to be doing splicing in the cabinet versus out of it? Are we going to be using slack loops at certain low distances? All of that can be put into your design rules if you have a robust design engine. In terms of materials, things like on a given layer, do we want to have a 96 fiber cable versus a 48 fiber cable? Do we want to have the options for both? Do we want to use pre-connectorized terminals versus fusion splice? Really just all the different materials that are going to be put into this automated design. And then, of course, finally, the actual costs. So the cost for those materials, the cost for the cables, for the cabinet, for the splitters, for the ODN. But then also the actual cost for labor. So the cost to install a given piece of equipment, the cost for maybe a new pole, the cost to do underground trenching, the cost to splice. All of that can be put into a complex design engine and ultimately make your design rules for that given deployment. So then the third pillar, as we're calling it today, but the third component of an automated design is the calculation engine. And what this calculation engine does is it applies those design rules, so the architecture and costs, to the input data we have. Now, these calculation engines are really what the IP of a given design engine is. So it's the secret sauce. Now, we can have design engines or calculation engines of super basic complexity. So they're just running down. Your output will look like a cable right down the center of the street with just a straight line drop to a house. Or it could be super complicated and then have things like a design that's very close to what you're actually going to be building in the field. This is all dependent on, again, the robustness of the solution. And it's ultimately coming down to the algorithms that are being put into the software. A good platform can also be tailored to your specific use case. So if you have a certain type of input data that you want to be using and maybe some architecture that's specific to you and your design rules, then a good calculation engine will allow you to actually design given your specific use case. So there's an important relationship between these three pillars. So if you have very basic data and very basic design rules, then maybe you don't need the most robust calculation or design engine. However, if you're dealing with very complicated data and a lot of data and you have complicated or very exhaustive design rules, then you're going to need a very powerful calculation engine. But what also comes into play is the actual time required to do this design. So we have users who are trying to do 100,000 or even more homes at a time in a given calculation. They don't want to wait five days to get the answer. So, again, this plays a factor in terms of how these all relate with each other. So you can throw a certain amount of compute power at these calculations, but it doesn't multithread. There's only so much that you can do in terms of compute power. So you also need to make tradeoffs. Do I include less input data or maybe make my rules simpler to get that calculation time down? All of that needs to be considered when you're doing these automated designs. So let's look at a super basic example here. So in terms of our input data, we have what looks like some trenches and crossings. And we're trying to serve these two homes here, this home one, this home two from our cabinet. In terms of our design rules, we have one cable size that's running underground. And there's going to be a uniform cable cost for it. So our calculation engine here is super simple. We're just doing a shortest path algorithm. So when we complete our automated design, boom, boom, we have our design output. Super simple, but that's just kind of giving an example of how these three things play into each other. Now, again, using this, maybe you could do something for an RFP or feasibility study. But this isn't the level of detail you want if you're really trying to go out there and build a network. So now we're going to look at a case study. So this is for a rural deployment, rural fibers of the home deployment. So in this rural area, there is about 700 buildings or 700 homes. It's 200 square kilometers in terms of size, about 79 square miles. And the area that we're going to choose out of this is about 180 of those homes or buildings. So what we want to do to maximize our investment or ROI is to determine the optimal architecture to be used. So we're going to compare four different options or four different set of design rules and see what this does in terms of the costs built into this area. So the first design, which we're going to use as our reference, is going to be an underground deployment. So we're going to be using one fiber per home, and that's going to be throughout these four scenarios. But in this one, we're going to be using a 132 centralized split. So it's going to be a passive architecture. And we're going to have a cost of $40 per meter to do the underground trenching. And we're going to use micro ducts. Now, there's a lot more in terms of the costing and the design rules that go into this. And we're happy to, you know, after this webinar, show you or talk through this. And, of course, these are just our set of unit costs. But what we're trying to do here is be demonstrative of what this means. So in terms of the cost per home pass, so that's the cost, not including the activation cost for a given customer, we're at $9,304. And the cost per home activated, so the cost to activate a given customer if they want to subscribe, is about $1,452. So we're looking at just over $10,000, well, closer to $11,000 of the cost of home connected. So we look here. This is just showing some of the actual physical output of our design. We see the majority of the costs here are the civil works, which is essentially the underground trenching, which isn't that much of a surprise. So the second scenario, we're going to use the same design rules overall, the 1 to 32 centralized split. But we're going to do it with an aerial deployment. So in this scenario, we do have to put in new poles, which has the costs. We're not able to reuse any existing assets. But still, we're able to see that without trenching, we have a much, much lower cost per home connected. So our civil works are less than 50% of our total cost. We're also able to go in on a more granular detail on the output. So if we're looking at this highlighted section in red, we're able to see that this is a feeder and distribution cable section here. And there's 120 kilometers of total length. But on that 120 kilometers, we actually have 176 kilometers of cables. So that means when we look at it, there's actually two 24 fiber cables on the distribution layer on top of each other. So maybe there's another way for us to use a larger cable in our design rules, or we can use another architecture to ensure we don't have these cables overlapping on each other. Again, this is just something we can look at to see if there is a more optimal design rule. So some of this gets covered in our next scenario. So again, an aerial design, but we're going to be using a task gated split or distributed split. So on this one, we're going to be using a 1 to 8 in the distribution closure or cabinet, and a 1 to 4 splitter on the drop closure. Now, of course, this will give us higher costs for our splitters overall, because there's more of them. But it'll also mean our distribution fiber counts will be lower. So when we're looking at this output, and I guess going back to the cost, the cost was slightly lower in the previous example of the decentralized split. Now, when we look at the output on that same section, it's still 120 kilometers total length, but there's less total cable there. So interesting to look at that, because there's only one 24 fiber cable. And it's obviously because we don't need that much distribution cable, because we're doing one set of splitting at the distribution layer. It's also important when we're doing designs like this, we want to make sure that we're not having too many drop clusters of one lone home, because then we're going to have those one to four splitters with a lot of ports wasted. So our final design that we're going to look at for this area. Is it one to 32 centralized? Again, but this time we're going to reuse existing poles. So this would be a scenario where we're going to the region, and then we're talking to the pole owners, and we have the ability to go in and use some of those poles. So, of course, there won't be a capital cost to go in and install our new poles. And looking at this, of course, that reduces the cost to deploy those poles, and it'll be the cheapest options out of all of them, which is pretty intuitive. So, again, this isn't an example to show you that this is the absolute best way to deploy a network. What we're trying to show you is by the automated design, we're able to take our data, apply these design rules, and come up with, using the calculation engine, come up with our output to make some decisions. Now, of course, that's just on the CapEx side, but we also have to look at what is going to be happening in terms of take rate, R2, all of that to determine what really makes the business case for the ROI. So I'm taking the example here where we're doing the aerial design, and it's the cascaded or distributed split. So the one to eight, one to four splitter settings. So if we look at this, if we only have a 30% adoption rate or take rate, which is quite low for a rural area, our actual cost for home connected is $9,649. So quite expensive. If you put the take rate up to 60%, we get that down almost half to $5,199. And finally, if we get it up to 90%, which isn't impossible in some of these rural areas, although it is quite high, it's $3,716. So what we're able to do here is to quickly analyze how the take rate combined with an optimal deployment strategy really affects our business case. We see that it will drastically impact whether we get 30% or 90%. There's quite a delta there. That's really going to change our cost for home connecting. Now, this was just looking at 180 of those 800 or 769 homes. But what we could also do is we could look at this larger area here. And again, it's only 800 homes. But we could say, if we want to go in with that one to 32 aerial deployment, what would be the cost for each one of those homes to get connected? So it's simple for us to look via, this is a heat map, but essentially on a per home basis, determine which homes are optimal to connect. So if we look here, and again, this is pretty intuitive, but the ones that are more centralized and clustered together are a lot less expensive to connect. So we can do this exercise and determine, okay, maybe we want to connect down that kind of street there on the left. But on the bottom left, those homes, well, you know what, unless we get some grants or some sort of subsidies, we're not going to be connecting those homes. So again, we're able to use this automated design to quickly put out iterations and determine what's the best way to deploy our network. And again, when we're looking at it here, the dense area versus what we're calling the sparse area of those 800 homes, you can really see that the average cost per home connected changes from $1,765 all the way up to $6,037. So even in a given small area of 800 homes, there's a huge delta in terms of the cost for each one of those homes to connect. So what are we trying to take away from all this? Well, what the main takeaway should be is that each one of these deployment scenarios is unique. So we're just showing you here on one set of input data with a few set of design rules, what the output could be. Now, your labor and material costs are going to be specific to you. Your data is going to be specific to you. And what you're ultimately trying to do is going to be specific to you. So all we're trying to show here is that using automated design platforms, you can quickly do some iterations and determine what's the way to deploy. Each one of those calculations took a couple of minutes. And of course, this is much quicker than doing it manually. The second thing, which I sort of spoke to, is that using automation, we can have a quick almost it's almost like a sensitivity scenario analysis. We can say, OK, if we change our architectures this way or if our labor rates change this way, we can see how that impacts the total design in terms of a physical layout, but then also a costing layout. And then finally, those design rules, they should be well thought through. Often we see operators or engineering firms recommend one set of uniform design rules or architecture for a large area. But you can actually go through and find different pockets where maybe a distributed split versus centralized split is more optimal or using pre-connectorized terminals versus like a distributed tap versus fusion splice. It might be more more optimal on a given area. And finally, I just wanted to end with some of the pitfalls that we see when we're working with our users in an automated design platform. So the first one is that not all data is created equal. What we mean by this is that just because you have data and lots of it doesn't mean that it's necessarily useful. So we'll be working with a given user and they'll have a whole bunch of data in either their inventory platform or that they received from from a third party. And they'll think because they have this that we should be putting it into our calculation engine. Well, this is where things like field survey solutions come in handy or looking at the attribution on this data is also valuable because just just because we know we have a fiber cable out in the field. If we don't know what the available strands are, we don't know, you know, the current status of it. And if it's still active, all this is important information to know. So just because we have this data, it also needs to be validated to make sure it's good data. So the second pitfall that we that we want to say is the term that perfect is the enemy of good. So also work with some operators and they want to make sure that everything is just perfect. So we did say that it's important to be thorough, but we also don't want you to obsess and obsess and obsess before getting anything out the door. So we see where there'll be almost this paralysis analysis of saying, OK, well, what happens if I tweak maybe the cost of this four fiber drop cable? OK, maybe what happens if we don't don't include this existing cabinet and then they're not able to get anything out the door? Planning and designing shouldn't be the bottleneck for the the whole deployment. And we know right now that time to market is extremely important. So, you know, it's the 80 20 rule and try to get something out the door. And finally, and maybe this is most important, is that automation does not mean it's autonomous. So often when we'll first be engaging with a user, they'll hear that it's an automated design and they'll think it's all AI and that their job is going to be lost. And that's just going to be a one click button and that's it. This this is not the reality. The automation is there to take away those error prone, repetitive, you know, boring tasks that that users often make mistakes on. We still need engineers to be QAing that automated design to make changes, to say this doesn't make sense. I want to come move this cabinet over here. I want to actually change the size of this cable. Oh, wait, I did a field survey and we don't have right away here. We really just want to empower our users and our planners, our engineers to spend their time on more value added, important decision making. And yeah, that's it. There was there was quite a bit that we touched on, but please feel free to reach out. We're happy to go through that case study and look at some of the specific numbers. And yeah, feel free to reach out with any questions you might have. Hey, Kevin, it's Trish with FBA. We have a couple of poll questions. Would you like to run those before we do the Q&A? Yeah, sure. All right. Let's do that. So I'm going to launch the first one. So for everyone attending right now, how soon would you want to introduce automation into your planning and design process? So we've got five options here as soon as possible, three to six months, six to 12, 12 to 24 or not interested in automating. So I'm going to let this run for just a few more seconds here and then we'll pull up the results. This is looking interesting. All right. I'm going to go ahead and close the poll and then I will share the results with everyone. So we've got 40 percent say they would like to introduce it as soon as possible, which is a great number. 21 percent, three to six months, 19 percent, six to 12, 16 percent, 12 to 24 and 4 percent are not interested. So it's interesting how that kind of went along with the time frame. So that resonates with what we're seeing in the industry. It's not really a matter of if it's a matter of when. What it was the workforce trends I kind of spoke to. And then also just this time to market as the next couple of years are extremely important. Exactly. Exactly. So let's pull up that second poll question for you all still on. And what do you see as the biggest obstacle in your in your company to adopting more automated solutions? So we've got some interesting things. Company culture, reception and employees, kind of the autonomous issue that Kevin brought up. Data issues, budgets or all of the above. So let's give this one a few a few more seconds here and then we'll pull up the answers. Interesting. OK, so we're going to close this one out, Kevin, and then I'm going to pull up the answers and share that. So biggest one is data issues, 44 percent. Then we go down to all of the above, everything at 20 percent. We've got budgets at 17 percent, I think can be an issue for everyone. Ten percent employee reception and eight percent company culture. So there you go. And I think Crawford from your team is going to come on and handle some Q&A. Yeah, great. Crawford. Yeah, thanks, Trish. We have had several questions come in. So, Kevin, our first one is, do you optimize the number of connections in a distribution point and in a street cabinet? I mean, we optimize the entire network. So in terms of a distribution point, we'll optimize the location of it. We'll optimize the feeder rail coming to it. The number of homes will you can determine the size of that distribution cabinet that you want. It will make sure that it has the optimal number of homes connected to it. And this is all done based on not only your architecture, but also an efficient and minimal cost basis. Great. Thanks. Next, we have this ComSoft integrate with other systems or software. Yeah, that's a good question. So we try to integrate with our customer solutions, whatever they are. Of course, being a GIS platform is ideal. Now, the tightest integration we have is with the IQ Geo Suite because we're native to it now. So that is beneficial, but we can also work with other platforms as well. Great. Is it possible to integrate with CAD and produce construction and or permitting drawings? Yeah, that's a great question. So the thing is, it's easy to go, well, relatively easy to go from GIS to CAD, but you can't go from CAD, unless it's map 3D, but you can't go from a flat CAD to GIS because there's no geo reference. So with our solutions, we can start with a high level design, go from a more detailed design and ultimately put out your permits and construction drawings and even export that to a DWG or DXF if required. But if we're just starting with a kind of CAD flat file with no geo reference, and that makes it extremely difficult to work in these types of ecosystems. Okay. Next we have, how does the automation process impact quality and accuracy and specific with reference to those things usually being insured with walkouts and on-site human involvement? Yeah. Yeah, that's a great question. So we're seeing more and more elements like LIDAR being introduced or other ways to get that field data captured in a up-to-date and 3D manner. That being said, at some point in your process, you're going to have to do that construction right out for that field survey. So it's really just having a discussion about your workflow and whether that's earlier on or later on in your process. Because at some point, you're going to have to be in the field constructing that network. So even if it's later on when you're actually going out to build it, there is going to be some field element to it. Sure. Next, I think this will be an easy one for you. Can I see a demo of the software after the webinar? Of course. Yes. Just please feel free to reach out. We're more than happy to. And will you also cover fiber indexing? Oh, fiber indexing. That's a good question. So for those who don't know, fiber indexing is a commscope technology. And we're seeing it get adopted more and more. And so we now have the ability to design in an automated manner fiber indexing. So that's fiber indexing terminals. So that's something we can do if that's required by you or your client. Got it. And so this one is about power utilities and how they can benefit from the solution. Yeah. So that's actually a nice time on that question. So traditionally, Comsoft was in the telco or fiber automation business. And a few years ago, we started an R&D project on automating, planning and designing for power networks. So the changes that we're seeing in the grid with the modernization of the grid, distributed energy resources, et cetera. That now, with being part of IQgeo, has made us accelerate that product. And so we do have some early solutions in the power utility market. So that's something we can do there as well. Great. And is IQgeo suitable to operate and maintain a metropolitan-sized optical fiber network? Yes, definitely. So IQgeo not only can operate a metropolitan network, but a nationwide network has customers who are doing that. So that's not an issue. Okay. And looks like one more we have right now. What is the importance of fiber planning and designing tools for automation in underdeveloped regions to have better time to market in ROI? Yeah. So underdeveloped regions, I assume we mean, in this instance, like U.S. rural regions that don't have access. So, of course, with the Infrastructure and Jobs Act now, everyone is seeing this influx of money. And that's interesting. But what we're able to do for some of these jurisdictions that are looking to determine if the grant's enough to make these projects viable, they can use this automation to say, okay, it will cost this much to serve these homes with the funding, because there's no business case without this funding. They're able to say, okay, with the funding, if I determine that I'm going to use this architecture and build out here, maybe I can actually make these networks possible. Got it. Thanks. And, Trish, I believe that's the last one. That's great. Thank you both, Crawford, and many thanks to you, Kevin, for putting on a great webinar today. A reminder, this will be available on our website within 24 hours. But if you have any questions, you can contact Kevin directly at the email address shown on your screen. Thank you very much to IQGO for being a great partner with FBA. If you have some time tomorrow, please tune in to one of our two webinars. And we will see you again shortly on our next FBA webinar. Thanks, Kevin. Have a great rest of your week, everyone. Bye.