Modernization Hub

Modernization and Improvement
Cox Cable Technology Advancements Winter Conference CableLabs

Cox Cable Technology Advancements Winter Conference CableLabs


Happy Friday everyone today is Februari 19th, and this is episode 19 of our Google hangouts podcast and all things doctors I’m Brady volt founder of the [volk] firm and nimble this Today for our show We have two great guests the first is our tried and true John Downey cmt s technical lead at cisco systems Hi, John great to have you [back] Good to be back. Hey Brady piano Also, we have with us our special Guest Jeff Finkelstein Executive director [of] Network strategy at Cox communications, [Jeff]. It’s great to have you with us today Thanks, Brady good to be here. Hey John. Hey So today, we’re going to cover a number of topics including cablelabs winter conference it happened last week the NCT sees winter Educational conference sats chapter p.m.. Training at John, and I did this past week and So first we’re going to cover some things in the news cable apps has announced a newly unveiled Project at cable, abs illustrates, how docsis three dot one technology? Can can be evolved to basically support? bi-directional Gigabit traffic over the Network And I’m really kind of interested in so it looks like we’re going to be evolving the cable. Tech spec or the cable, abs Docsis three dot one specification that gives us sort of gigabit down traffic, but not symmetrical gigabit upstream traffic to allow us to [have] symmetrical Gigabit down gigabit up I’m really interested in like how does this work and would it work you know in in a non? Passive HFC system or just only going to work in the Passive HSC System um Jeff have [you] heard more about this Or is this is this kind of just like brand new news out on the Street? so we’ve been working on it probably about three years now and And the idea [was] that the challenge we threw out and what created ducks is 31 in the end was how do you do gigabit services over a Coax network and we we took very the original docs [of] spec and and in working with a new newer technology ofdm LDPC ETC came up with Docsis 31 the next challenge was now that we did that as We as we do the traditional interpolation of What is what’s going to be needed in years to come we know that it takes typically five years or so from starting a spec to where you actually Have something that is as a product But it’s usually another year or two before that when you’re when you’re in an investigative phrase her days so we Started thinking along those lines and said what can we do with it going forward? And what came out of it was some work that was done with Stanford university bell labs and other folks and said how can we take some newer technologies like echo cancellation and Having integrated Transceivers to be able to simultaneously Transmit and receive signals on the same cable And where that ended up was a technology very similar to what is being done in the mobile space? with using a form of digital subtraction Where if the device knows what it’s transmitting and it receives a signal? If it subtracts the transmitted signal from what it received what it’s left with is. What was transmitted from the other side and That’s what’s become the next generation of Docsis spec The number still is stolen up in the air whether it actually goes from a science project? to an actual product still remains to be seen but [it] does have the potential for providing a 10 gigabit symmetrical capability in a 1 Gigahertz plant and Potentially can work even higher Yeah, so so what I saw and kate anyone can can look at this cable abs has this posted on video kind of like the constructs of or sort of an outline and diagram of their specification on their website so can check that at cablelabs calm And had some unique graphics, so they showed. How we’re currently doing TDMa in the upstream with our existing docsis Specification where you know multiple modems are using time division multiple access In the upStream Channel kind of looks like they’re saying this new specification concept would allow all Communications and docsis to do this not just from five to forty two Megahertz, but we basically do this with the entire frequency spectrum So to me it sounds like we would have in order to do this. We’d have to eliminate dialects filters from the plant is And and so if we eliminate die Plex filters we would have a this would be a passive plant technology for Docsis am I correct in a son and assuming that or is there some other way that we would use this technology with A bi-directional HFC plant So in its purest form you would want to be in a diplexer lyst plant That there have been some work done in mocha in particular with which is a similar technology Where you could potentially have an amplifier, but it’s got to be a very intelligent expensive amplifier The way that it works most efficiently is in a passive Coax or node plus [zero] type of plant Now you can make it work in a diplexer plant But you don’t get to use the whole spectrum, so think of it if you have for example a traditional 52 4254 2002 cable plant You could take a portion of that Downstream so you might use 5 242 for your uPstream or it could be 6585? Whatever you were going to use You could then take a hundred 200 or whatever to 500 for your legacy single carrier qualms 500 227 54-Year Docsis 31 or or whatever spectrum here? We’re going to use but then you take that upper spectrum, and you can make that your full duplex Dots is 31 or Docsis three point 1x Upstream Downstream Scenario, and then you could grow that over time as you got away from single carrier qualms, you can move 31 down or you could put the next gen Docsis in that space you Could also have it coexist because it’s using 31 carriers so that you could [actually] share the downstream or theoretically share the upStream and Let the scheduler deal with making [sure] that there’s no collisions in that in that Spectrum that the legacy devices are working in but if I have this full duplex and from my house I decide that I’m going to have this new device Dude is full duplex at say 500 to 700 Megahertz whatever happens to be and I [translate] from my house How does my uPstream from 500 700 acres? get backwards Through my fiber node, how does it transmitted so the fiber note is not just a node plus zero the fiber node has to be Some hardware has to change there as well Well, you would have to be in a distributed architecture for it to work so that When you’re in a when you move the cmt SC cap into a remote phy device kind of scenario you would be handling it at that point because that’s where the the either it would be justify or if you’re in a modular what everybody so the optics really is an Ethernet or digital optics no more analog Amplitude modulated optics that’s correct So it’s very interesting yeah, I think it’s a it’s a unique expansion of Docsis it definitely Allows us to extend we get to say a node plus zero Technology where we have that passive opportunity? And then it changes the dynamics of say you know when we may move to a pawn solution You could extend for the lifespan of [Docsis] a little bit further So yeah, and and I don’t think there’s any doubt that HFC in General has got a long useful life ahead of it, and it’s just how do we how do we take? as much advantage of its inherent capabilities as we possibly can without doing anything crazy and Being able to leverage the investments that we’ve made over time push fiber deeper going to that fiber deep architecture getting Node housings in that will be upgradable from Potentially a legacy node to adding in Docsis three one remote phy a modulation and then be able to Replace that over time with a next-gen docsis which will also be backwards compatible And that way, we’re leveraging all of these investments getting fiber closer, and then you can adapt your your Deployment scenarios into a very micro segmentation View and do demand based bandwidth so that as you you grow above the Gigabit Services will have with Docsis 31 You’ll be able to go into many multi-gigabit Services with next-gen docsis and still be able to retain some of those sunk assets We’ve been putting it out in the outside plant for all these years Right awesome awesome, thanks Oh, thanks for the the background on that jeff that sheds a whole lot more light on the information That is available in [cablelabs] right now, so speaking of cable, abs they had their winter conference last week I was not able to tend to atTend It wasn’t in Orlando, so I I think the weather was a little colder than people were Expecting to have Jeff. I think you were there John I don’t think you could make it but Jeffa any any other exciting news during the winter conference Well, I think that the big news next to next-gen Docsis was [cablelabs] to point out that Cable Abs is in the midst of reinventing itself into into a number of very distinct Organizations one is the cable, abs we’ve known all these years the testing labs for providing certification and qualification of Docsis equipment The other two that they’ve created are an RD Branch which is more focused on the zero to three year the eighty ninety percent success rate for new technologies The new one is an innovation Branch. Which is more focused on three to eight years and coming up with ideas for technologies that we can potentially leverage as an industry in the future so in cablelabs [tudo] is there’s I mean there’s still a nonprofit organization [there] just expanding their Basically [their] [RD] there in two different realms is that the [the] focus of it? correct yeah, they’re they’re shifting the financial capital that they have available, and they’re they’re structuring it so that as opportunities arise whether it’s in an RD or an innovation branch there they can be very malleable and take advantage of the The money that they have available and move it into the areas that that they can have the most impact Now they did create a for-profit Branch Was Network Fx originally, and I think [they’ve] rebranded it curio. Which is where they’re going to? Make things available on a for-profit basis for testing and analysis ETC Right and I saw there was an announcement the cable app is going to do some they’re going to have like a small incubator or something like that [where] they’re going to work with some companies that may be doing like startups and Some innovative technologies that benefit the cable industry and they’re going to help help guide those organizations in one way or another 22 as they grow and develop to Develop technologies to benefit the Gabe organizations. Yeah, that’s correct They made a lot of investments in developing a world-class Innovation Innovation Organization they brought in a lot of outside talent the ceo film McKinney Has got a very good background and a good track record of success doing this at a number of firms So it’s really a pretty exciting time for the cable industry and for cable, abs itself yeah, it’s great to see them expand and change so That’s awesome Um so uh last are actually it was this week. It’s been a long week for me So you mentioned? Cablelabs test labs and stuff, I think that the head of their test labs Matt schmidt he and I were at the NCTC NCTC. If you’re not aware of that is the national cable television cooperative they have a really big focus on tier 2 and tier 3 cable operators and put that pull all of them together and focus a lot of training and Do different things with them one of the things that they have is a winner education conference Which was in Phoenix, Arizona this past Monday and Tuesday. I had the opportunity along with mesh myth. We did a panel on Docsis [31] and It was fiber to the curb or fiber to the home but it was really focusing on epon versus Jeep on we talked about the different technologies, but We we found there was actually some really differing opinions on when it makes sense to Deploy docsis or Docsis three dot one versus epon and Jeep on and And so I mean I I personally see a future where Docsis three dot one has quite a long life span in it as well as I see that epon and Jeep on have their own rightful solutions in different places I kind of wanted to open up a dialogue as to what you guys see as you know where docsis is what its lifespan is and also if you know what you see as far as Operators deploying Epon Jeep on and another you know which one is one technology. Epon orgy pawn better than the other So Jeff come on and let me let me jump in you know coming [from] a vendor side and seeing multiple vendors or Msos And seeing what other people are doing in the us and internationally It’s like the epon [Jeep] on debate was more for cherry picking meaning. It’s the prices have come down Fibers, not that expensive right aways are usually expensive the equipment prices come down And then you could offer Potentially gigabit symmetrical now with cheap on or epon may we went so far that you look at cablelabs. They made a spec Dpoae and then they expanded it to Dp 0 g right boxes provisioning over a pond or docsis provisioning over coupon so the docsis Provisioning side of things with the class of service and service classes and all that People are very used to and they like that idea The conducting median in between whether it be cheap on a pond or docsis width modulation all that It didn’t really matter if you were maybe doing classes of service and whatever I saw the deep on Jeep on being like cherry picking meaning Five customers in the node want One gig service and willing to pay for well I’m not going to upgrade my entire cable [plan] to Docsis three one today If I can just maybe cherry pick those five customers give them fiber and fiber to the home and be done with it Or I maybe look at our fog, but if I’m running fiber, I must will do cheap on anyway But if I need video well, I still get the video out there. So maybe our fog For video and a Jeep on overlay together sometimes works So there are a lot of solutions out there, but when I saw the box was 31 with the potential speeds that are offered We could offer much higher speeds gigabit on a Downstream Upstream is going to be a different story for a while But we could utilize an exploit our existing HFC plants with not much change and still offer Giggy down Ubiquitously meaning to almost everyone in might give my cable plant There’s still capacity planning concerns and stuff like that, but on the upstream I was going to ask Jeff Earlier you know are you guys looking at 85 megahertz up streams. Are you looking at 204? I’m sure you’re looking at you’re going to say yes to everything [so] I’m just wondering like how far along. Are you guys it was interesting cox when they went private? They’re no longer. You know [holden] to wall Street They went private got rid of the public sector side of it. Whatever you know being publicly traded Once they went private a few years back. They upgraded their entire plant the one gigahertz on the downstream now You don’t assume any msS doing that because they always got you know slapped on the hand from wall street by saying a we needed A billion dollars a upgraded our cable plan so cox was in a very unique position to upgrade before anybody else without ever seen Be able to go to [1] gigahertz I’m just wondering if that same thing might happen on the upstream for these guys or Is there other msS brady you might know other msS. I have a few of down Latin America Look at 85 Megahertz upstream one gig down to me. I would go to 1.2. 18 Gigahertz Downstream at this point I wouldn’t do a 1 gigahertz downstream. You know [I] want to upgrade my plant one time and that would do pluggable dialects voters with potential remote phy upgrades to my note if I could That’s just my philosophy at this point. What do you think Jeff? That’s a there’s some pretty loaded questions so so I’ll take it and sort of [uh] From it from the first questions brady asked through to You know you’re you’re very very stupid comments there so II pongee pond is It I agree with you. That’s a red herring that the reason We’re part of f cyan and as part of that, we pushed very Heavily to have this spec created called x gs. Pond and This is where we really love kind of level the playing field because both in today’s scenario the Differences between the epon Makka Jeep on Mac are very very small As a matter of fact many of the chipsets can be booted in Annie pond mode or a jeep on mode? The real differences came in the optics with x gs pond and 10 gige pond they both use the same optics now So both get the advantage of having a very large installed base Over time where everybody’s using the same optics, so everybody is a winner is sort of in both sin areas So it’s not that you have to do one or the other you can do both you could there’s many options there So that that’s sort of a non discussion point anymore, and people will choose what they want It more comes down to if you want to do docsis provisioning like like you said You would go [with] the DP oe spec if like us you do flow through provisioning, then you provision directly out of your oss and And you can go whichever path you choose and then you just have these modules that are unique to whatever vendor. You’re using So it makes it simpler Depending on on whichever path you choose to go with a with a pond technology So when it comes to the split and the uPStream? We’re thinking of that. We would do in 85 split But it really comes down to a much more granular scenario where it’s really a greenfield Brownfield discussion In a [greenfield] your your most often required by the homeowners associations or the builders to pull fiber So we do fiber and and for us we do Jeep on Others do [a] pond like I said it’s it’s a it’s whatever is most appropriate for the operator deploying it So that you really there’s no Coax that’s on the table in that scenario How do you do video for those customers? Well initially you have to do our fog? But over time as you deploy a video And IP IP based video product you’ll be moving to Ap video And then you’ll be taking the our fog out of the equation yep, but you have no choice in the short term excuse me unless you have a video product today so so now you have to deal with Brownfield and And one thing is is [is] just pretty much guaranteed Growth continues Whether it’s a 50-percent long year-over-year compound annual growth rate Maybe because of people putting more and more on a mobile. Are you know going from fixed wire? to to Mobile scenario then you you would maybe see that rate come down a little bit, but the difference from 52-47 to 45 is [the] matter of a couple of months So you need to figure out? How do you leverage it the way we’ve traditionally leveraged it um Thanks in large part to the effort that cisco is done is we’ve been adding more channels But you eventually run out of channels that you can put in we only have 32 channel Modems So now you have to look at what are your alternatives? You can add another 32 channel bonding group. You can upgrade to a 2 a.c. Captive ice and then you could add some 31 channels, but you’ve got all these legacy devices we can continue to split nodes because every time you split a node you pull more fiber and A good rule of thumb is you can never have too much fiber in your outside plant Your diet or in your diet. Maybe maybe I don’t know you would go there so so as you get this fiber out now you you create a whole new series of possibilities for yourself and in every operator system We have got parts where we have a very highly competitive scenario you have somebody over building you and somebody doing fiber and And one one of our rules is the only way you really compete with fibers with fiber and in those scenarios we would give very strong attention to the analysis of pulling fiber and and being very hyper competitive in that scenario But that’s a small part of many operators footprint So now you’ve got the rest of your footprint well You’re also going to have a part of your footprint where customers are very happy with with a gigabit Docsis 30 they have the services they need the plants in good shape and There’s a possibility that we would continue using 30 for many years to come because there’s still a lot of inherent capabilities in 30 but then you’ve got this middle portion you know the middle of the bell curve and Here’s where 31 gives us a very strong play So we would start start by deploying some amount of spectrum for 31 you would increase that over time and We would be able to leverage See caps ability to do both 3-1 and 30 in those and then follow sort of our our Traditional five seven percent nine percent a year upgrades of Cpe and Be [able] to move into having more and more dots as 31 which gives us higher and higher Bandwidth When we do touch the plant as you said if we’re going to change the split and we’re going to go to 85 which is probably the more economically feasible because It’s going to impact very small percentages of existing cpe as opposed to 204 which is a complete cpe Replacement if we go to 85 if we’re going to touch the plant like you said We’re probably going to go to 1.2. And some may want to go to 1.8 1.8 is a lot more problematic, and you know you guys know way more about this than I do but We’re thinking that 1.2 Would be where we would end up in everything that we’re going to start we’ve been deploying for a while And we’ll continue deploying are going to be 1.2. Capable passes as we upGrade to 285 We in turn would start replacing all of the trunk with 85 diplexer and then we would start working into the home and we would terminate the plant at the Gateway device, which would be an 85 in this way we [avoid] the whole home amplifier scenario, so you can mix and match with Legacy upstream with with the 85 upstream put 31 in the you know 42 or 45 285 Start your down streams of 108. Yeah, there’s going to be a small percentage of set-Top boxes you have to replace because of telemetry But but those are just things those are just be a you to us. We live that world all the time So it like I said it becomes a very demand based response To what what others are doing that are in your footprint? so in the 1.2, gigahertz [Jeff] one of the topics that came up when we were at the NCTC web panel it was disgusted 1.2. Gig in the docsis three dot one specification That was actually put in there for plants that are already upgraded to one gigahertz That you know they [had] done some testing and they found out 14 1 Gigahertz Plants would roll off out to about 1.2 Gigahertz, so You know apparently when they were just developing the docsis three. I want three dot one specification They felt a 1 gigahertz plant wouldn’t necessarily need upgraded to 1.2 Gigahertz because we could we could use that roll off section above one gigahertz out all the way to 1.2 Gigahertz and put Docsis three dot one ofdm carriers out there is that is that something that you see is actually true in You know in your one gigahertz plants? Or do you think you would have to upgrade? One gig plans to 1.2 gig at some point in a future to actually take advantage of that roll off, band so [I’ll] answer it based on what I’ve seen and let John get into the Into the nitty-Gritty because he understands that I mean I’ve learned a lot from him in this regard so Depending on when you did the upgrade What we deployed is we did our upgrade from 2006 approximately 2 2010 The roll-off isn’t quite that good because the technology has advanced a lot in the past 10 years so in implants that were upgraded 10 years ago You can’t really get pic advantage of that roll off unless You go from a very high order modulation to a much lower order modulation The people in in you know in an 870 that take advantage of the roll-off are using like 64 qualms in that role off region I’ve expect the same thing even though it so at the end because while the carriers are smaller You still have the problems with the modulation? It’s like we’re taking a step for it one step forward and two steps back or two steps forward one step back like we want To go to ofdm Dr. You want to get higher speeds, but now we’re just going [to] run it in some roll off and get lower speeds again So it’s the kind of shooting ourselves in the foot here plus the 870 roll off You’re not running into the roll off of the taps Where at so maybe you could get away with a no running it above an 860 in the roll off but for a one gig plant your taps Could be roll out pretty hard at one gig and the fact that we have a lot more attenuation and Coax right at higher frequencies So I that one’s a tough one to Really roll with because if you’re already going to make a change out to your amplifier Modules you [maza] just change it all out once anyway, right? I mean, that’s the whole point and then having gann technology gallium nitride I could change out modules in a housing without reese pacing my Ma housings because I I’m going higher frequency But I have more gain now in those amplifiers, so I don’t have to reach space and splice cables and all that I hope that’s what we’re looking for right? Yeah, but if you’re telling me [that] I can only run 64 qualms Above 1 Gigahertz in the you know maybe to the 200 Megahertz 200 Megahertz more well back you a wet That’s pretty good because I’m you know that’s bandwidth I can’t use and in many cases. You know. Maybe I’m you know today We’re running to 256 qualms But I couldn’t even think of running 256 qualms above a gig so I Don’t know that sounds like a pretty great thing without having to do anything. I could run 64 qualms up there Right I see you’re saying but yeah I think we say like if say cox is all right at one gigahertz on the downstream And they don’t want to touch your upstream. They’re not going to change anything, right? I’m No money invested no die flex builder change you’re still five to 42 on the uPstream like to be able to Expand maybe another ofdm block out to 1.2 gig and maybe they could but maybe it snowed plus 1 or no plus 2 I Don’t know for if it doesn’t work for that last customer how do you actually offer it if you’ve got to be able to offer to everybody or Maybe you just push those devices to the first ofdm block, and I don’t know that is Really an option it’s a demand based scenario So that even today every operator has challenges in the outside plant and in the drop and in the customer premise and We do what we do best. We we have some very talented technicians they go out they Find the problems they they correct them they replace cables they replace AmPs Passives Etc but it gets more complex as you go to these much higher order modulations and traditionally We’ve gone from the days of this 704 to digital spectrum analyzers to vector signal analyzers and The one thing that we are still working through as part of this in se te has a has a big Effort going on in this space as well is that? you can sweep and that gets you so much information, but you really have to generate a mod high order modulated carrier and send it through and Demodulate it to really understand. What’s going on [in] the plant? Proactive Network maintenance helps a lot because you can turn cable Modems into full spectrum school full band spectrum analyzers but in the end It’s really going to be that the talented technicians who are going to be able to really find those problems? Yeah, we can get we can get a lot of information from looking at linear impairments Nonlinear you deal with as as they crop up? But we have to really undertake the training of the technicians as soon as you possibly can and Get them the tools that they need To be able to really make these technologies work as we go forward Yeah Yeah, yeah, I think you’re spot on there even even with p.m. Pm helps, you find the problems but it’s it’s the technicians that are that have to be out there solving it and And I know this is gonna be hard for you to believe Jeff, but John Downey and I spoke for four hours On low is that Wednesday John would join in situ go yea after in West Virginia on nothing but p.m. You know we had prequel John did some like free Equalization how it works how to turn it on in the cmt s and he gave a live demo But that was pretty much the the focus on it. So it was a I don’t like John would you think about that training? I mean a lot of people myself included Always thought p.m.. Was pulling the Modems for pre equalization Information but p.m. Now is much more than that right full bandwidth capturing the downstream Potentially upstream spectrum analysis at the cmt s and the cable modem Utilizing a not a lot more functionality like Kaiba lab showed it at Ct. Last year not just looking at say the preparation taps but how they react to themselves over time the fluctuations impulse noise, or Sporadic or what would we call in [a] connection that was? Intermittent Connection emit an onion so to be able to take that information and reevaluate it and try to come up with you know troubleshooting Targeted troubleshooting right so I mean it’s it’s come a long way and with Docsis 3-1 and 30 it’s you know adding even more Functionality, I always thought it was crazy to say we’d have a spectrum analyzer and modem But here we are right Spectrum analyzer in [a] Modem so it’s pretty impressive and Jeff You’ve been on Pnm for as long as it’s been around, so I’m sure you have an opinion on it Well just so when I came up with the idea originally in 2005 I think of from looking at equalization coefficients in Working with how roberts who at the time was with Big band To reverse-engineer them and figure out. What was going on in there We were looking at it from a very simplistic perspective just wanted to know what was going on and then as We really started to understand it and we’re able to Able to look at the taps in time to figure out where the the linearity. ‘he’s were occurring That that was a big big deal But the real Magic has been from from what cable abs did what you’ve done with what John has done? Where you’ve turned it from a science experiment from when when we did it? to an art form and ways in which it can impact the daily life of The cable operators by taking advantage of the information, that’s just inherent in the network It’s it’s come so much further than then I think either power. I ever thought and it’s great to see [I] Even thought when it was first deployed. I had the same simplistic approach was I Have pre equalization masking HFC problems My end customers happy, but it is masking it and the Rf guys are just sitting around not fixing anything So a lot of msS like. I’m afraid to use Preq because it is masking I want my rf guys to fix things and I said you know from the beginning it would have been nice to be able to have a way to show [the] emmy R of that modem before and after pre equalization kicks in and that’s exactly what you’re sort of talking about right going to the modem and How hard is this Modem working? I need to know how hard is it working and am I near like that needle? That or that straw that broke the camel’s back am I close to that point? How do you know unless you talk to the Modem to say how hard are you working? Because I really don’t know how much oomph is left in that modem Yeah agreed All right Comcast has taken that way out of the stratosphere, right? I mean, they’ve done larry walcott and those guys have taken a lot of effort and put a lot of effort into the p.m. and and it’s gone is Farther than I ever thought it would yeah, larry. I mean it’s now that were mentioning names let Larry has a good team, so they have a very nice solution, but they have been working on it for many many years and They have a lot to be proud of so hats off to Larry Now back to your docsis three one example of running above 1 gigahertz. I personally think what we’re going to have to do You know there was some diagrams shown from Comcast showing kind of like a standard deviation curve of this many Modems with this dallas really are and this many modems with this down from when we are and knowing this standard deviation curve I could probably um up with four docsis three on modulation Profiles of robust one a Really fast one a mediocre one, but where was that emmy are taken from one downstream frequency who? Validated that downstream. Mmer is it going to be the frequencies. We’re actually using for my ofdm carrier You know I personally feel what we’re going to have to do is Deployed ox is 314 with a very robust profile Record all the information back that way we know we’re using a frequency that’s intended to use Monitor all the Modems over time and then we pick out from there, which modems are viable for I’m faster speed And then then we start divvying at this point. This is going to have to be an Sdn app right something that’s Processing a lot of information potentially what [4000] subcarriers for one modem? Over and over and over and then have to go to thousands of Modems So we need to get all those mu our readings of all these sub carriers or even the pilots or whatever? And then decide a this modem is a candidate for 4K qualms, but this modem over here. He’s off a 5 aMp cascade He’s better off for 256 one. You know so that’s where the power of three one’s going to come into play But that’s further down the road. That’s all we need your see mt. To do John is create an individual modulation profile or every single Modem, and I think that would make Jeff really happy Yeah, I don’t know how happy you would you? know so one of the The beauties of 31 is its ability to do just that but You know there’s there’s a basic rule to a lot of things we do that We sometimes lose track of and that is the juice worth the squeezing so that There are a lot of things we can do But as John said there are a lot of subcarriers this is a big data application on a grand scale and And how much are you really going to squeeze out? by Doing all of these computes just to get a couple extra bits per second through one or more carriers if there’s a there’s a big gain that’s going to justify the cost and the computations Then it’s there’s an intrinsic value to that But to put in all these servers with all this memory and all the data storage and run all of these real-time models you [have] to there’s a trade-off involved so you know I like the idea of what John was saying and When we were originally sort of working through 31 in the very early days They were originally only two profiles there was a 3-1 profile and a 30 profile, and that was it and Then we started thinking about well, we can have we can have my permit improv I’ll a lot of complexity we could have maybe 16 Profiles or 8 or 4 I think As John said starting out with something very simplistic Gathering information applying a very simple sort of linear regression to the correlation between Any we’re collecting two things going on Will tell [us] just how much juice we can really squeeze out of it and and if it’s not worth it Even just going to 31 and a simplistic works everywhere profile simplifies plant management the one thing that is a lot of time is being spent on right now is How do you measure the utilization of these? Ofdm blocks when you have so many different profiles. I was going to bring that up. Yeah the big one. It’s a big one and It’s going to take the very smart folks that are writing the Ems and the NMS is To really understand the implication of it You know maybe maybe we’re overcomplicating it. Maybe there’s some more simplistic solutions But but that’s a key element, and that’s how we’ve always managed. Ro Sp is by looking at utilization Those rules are going to change and we have to thank that through before we make it so complex that That we’ve just made it almost unmanageable. I almost think when we graph this out One we have to decide. What is my rolling time window that I’m grabbing the data right is it 30 seconds? Let’s say it is 30 seconds for every 30 seconds as that windows rolling I’m going to have to have a graph that points to the throughput in that 30 seconds and the usable throughput in that 30 seconds So I’m going [to] have to have two graphs overlaying each other to say all right? I had 100 megabits per second, but the pipe at that 30 seconds was really only worth 200 megabits seven fifty percent So because the usability of that pipe is going to change right how many modems are transmitting what modulation are using? You name it so we’re going to have to have something that kind of tracks to say all right? Here’s my speed at this time But I don’t know the utilization because I don’t know the size of the pipe so we have to be able to take the and Look at what’s been scheduled in the modulation of the Modems and over a rolling window And that window has to be determined It should be we think user-configurable, but like how much processing do we want to do with this as well But you get the idea right yep? all right, gentlemen we are We have run out of time but Thank you for [your] for your time today We’ve covered a lot of great topics for all of our listeners If you’ve liked like the video make sure and you’re watching this on YouTube make sure you hit the subscribe button for anyone offline this will be posted to Itunes to now Android has their own podcast so this be made available also as an audio only video Audio only so thanks everyone for listening and have a great day jeff. Thank you for your time, John Thank you for your time, and happy Friday. Have a great weekend all everyone. Bye. Bye. Thanks guys [cheers]

Leave a Reply

Your email address will not be published. Required fields are marked *