A discussion with Gabriel Desjardins – Head of Product for Wi-Fi and Bluetooth in Broadcom’s Mobile Wireless Division.
But before we jump into the full interview… Keith and Sam make a little announcement about a new series they will be introducing in some upcoming episodes (you’ll hear that at the beginning of the show).
In this episode Gabe shares a brief overview of Broadcom’s involvement in the history and development of Wi-Fi chipsets over the years and when did the industry identify the critical need for new spectrum?
He also discusses what we can expect from Broadcom in regards to Wi-Fi 6E.
I was a statistical analyst (and product manager) in pro sports: I created behindthenet.ca, the first advanced statistical hockey website, put together a broad media strategy to build a community around the site, and leveraged this into a statistical consulting business with a client base of NHL front offices. Like a lot of startups, I was at least five years too early for the market.
Listen to the Show
Read the Transcript
197 WLAN Pros Podcast Broadcom WiFi Chipsets Past Present Future.mp3 transcript powered by Sonix—easily convert your audio to text with Sonix.
197 WLAN Pros Podcast Broadcom WiFi Chipsets Past Present Future.mp3 was automatically transcribed by Sonix with the latest audio-to-text algorithms. This transcript may contain errors. Sonix is the best audio automated transcription service in 2020. Our automated transcription algorithms works with many of the popular audio file formats.
Wireless LAN professionals podcast, episode one ninety seven. Welcome back to wireless LAN professionals podcast. Today I’m here with Sam Clemens. Sam, how are you? I’m doing well yourself. Great in the middle of this coveted lockdown thing. Are you and your family okay?
Yes. Yes, we are. Everybody is safe and sound, if not a little on the stir crazy side.
Well, you do have a huge backyard. That’s true. Yeah, that’s true. And in white for like puttering in the garden. So you can least go out there and play. Yes. And I hear you also have a bounce house in your house.
Yeah. That makes it that makes an appearance every once in a while when the kids get particularly rambunctious and the and the weather is bad. It’s we have just enough space, although it was the cause of quite the goose egg on the little girl’s four head the other day she was bouncing a little too far obviously and ended up hitting some furniture.
So is about hours outside now?
No, it’s still inside.
We don’t learn our lessons very well the second time she gets hurt. That will be the. That will be it.
I thought I’d have you join me today on this podcast to announce a new series that we’re going to do together about rules to follow when designing wireless lands and when to break them. So what are we doing in this series is I’m gonna come up with a rule, some sort of best practice rule. I like all rules. They’re made to be broken.
I’ll go through the part of why I think it’s a rule that we should follow, and then I will talk about why they should be broken or when it should be broken or something like that.
And if we we don’t always agree on everything, which is totally fine. But I’ll allow our listeners to have a little back and forth and a little understanding of that. There is a rule, but sometimes it’s OK to break those in an upcoming episode. You’ll have an actual rule and when to break it. Thanks for your time today, Zampa. Likewise.
Wireless LAN professionals is a place to educate, inform, encourage and entertain those involved in wireless LANs. This wireless LAN professionals podcast is an audio manifestation of these goals. Our host is a wireless land veteran consultant, designer and teacher Keith Parsons. And now the podcast for wireless LAN professionals by wireless LAN professionals.
Did you have with me Gabriel Danger? Dan, how do I do on that last name there?
Well, so it’s it’s in French, it’s it’s dead of them.
Well, that sounds way better. So we’ll just go with that. Well, Gabriel, glad you could join us. And you’re with Broadcom scene. Can you tell us a little bit about what you do at Broadcom? Sure. Absolutely.
I work for one of broad comes free Wi-Fi divisions. This is the mobile wireless division. We make Wi-Fi and Bluetooth products for primarily smartphones to make low power Wi-Fi and Bluetooth for battery powered applications. I’m the head of product for Wi-Fi and Bluetooth in that division are focused primarily on Wi-Fi and Bluetooth innovation, trying to figure out our roadmap for the next few years.
What are the other two divisions that do with Wi-Fi?
We have an access division that’s a broadband cable, access to business. Let’s focus on access points. So if you buy DSL router or Hohn router or just a standalone Wi-Fi access point, that’s what that division builds. And then we have a set top box division and they have their own Wi-Fi parts that attacks the cable set top boxes.
Do you have separate designs for each of those divisions or how much of that do you share between each of you?
Historically, it all came from the same core, the same IP. And over time as the divisions have grown, they’re a bit too big for everything to all been bundled under a single division. So the divisions have been split up a little bit, but the core IP came from the same place. All the people who work on the Wi-Fi designs have work together for going on 20 years. Fundamentally, it’s all the same thing, even if there’s some financial delineation between the groups that are working on things.
And how long have you been working in this mobile division? I have been there for 10 years. You’ve seen lots of changes.
I have. I came in and we had not yet contemplated our first ATO to eliminate CE products. My first big assignment was to take on the ATO 2.0 and they see mobile product line where I worked for for many years as the product manager for that product line. And gradually that morphed it to Wi-Fi six. Now Wi-Fi Saxenian Wi-Fi, some about how far out.
I’d want to ask anything that you can’t share, but about how far out is your work before it actually hits the consumers that they can put it in their hand?
Sort of a mix. I work on products that are shipping today, probably extending out to about a five year time horizon at this point. I’d be looking at 20, 20 through 20, 25.
So that’s a pretty long time horizon to be contemplating. And yet, if you don’t start early, there’s a lot of work to get things done.
Absolutely start to finish. I think it probably takes at least 36 months to productize a new chip. So you need to spend quite some time prior to that thirty six month time period figuring out what it is that you want to.
Well, all of the previous well, at least since you’ve been there, we’ve had five gigahertz and two point four. The chip manufacturers knew that the spectrum was from two point four one two of the whole range that was available. You made the radios work in that one. Did you as a broader team think of we’re going to need to go to more spectrum?
Well, I think there’s been a longstanding push for that for a long time. We tried to figure out ways to enable more spectrum in five gigahertz, like some of the you need to be or something.
We thought, oh, look at that, smack in the middle of what we want. That didn’t work out so well.
Yeah, well, it’s sort of like what are some of the things that you can do to open up some of the radar channels? Can you add another radio that sits there and performs all of the listening tasks required to get a DFS channel allocation? Can you use an adjacent device in the same room to get you your first allocation? Can we get DFS to run our mobile? There was a lot of work where we were kind of twisting ourselves in knots trying to figure out how to get access to the spectrum in five gigahertz. That can be a bit of a challenge to get to. So that’s been going on for years and years and years and years. And that’s basically been happening almost as long as I can remember at Broadcom. I don’t really know the exact timing or genesis of at some point this whole idea of having devices operate in the six gigahertz spectrum really, really took shape from my standpoint.
We’re trying to use contiguous band, which would made life much easier. So that was the first pass and that didn’t quite go through. So we went looking for more spectrum. And amazingly, the six gig worked after years of practice. We’re now recording this about a week or so after the FCC announced that in the US we’ll have twelve hundred megahertz of new frequency up and six gig. How long have you been working in that? I mean, it was it was predicted it was going to happen. So I’m sure you’ve been at this for quite some time.
You’ve had a sense the six gigahertz band will it will eventually open up. And a lot of people working on that process at Broadcom. I would say that we probably identified six gigahertz as a capability in our products back. Twenty, seventeen, some time we started talking to customers about how it’s going to be an eventuality and you pick the winning horse and it is.
So what does that mean from a hardware standpoint, though? Five gig and six gig are close. Can you reuse a lot of the same parts that you had before? You know, what are the things you had to think about as you went and put both those two together?
One of the interesting things that we’ve done as we’ve moved from altitude to let me see to Wi-Fi six and now to Wi-Fi Cixi, we’ve actually done a process node transition in each of those steps. So we did most of our itoh to the Lebanese sea work in a 40 nanometer process. We did our initial Wi-Fi six work and a 28 nanometer process and Wi-Fi sixteen sixteen engineer. And so the good thing about making that transition, which obviously takes a lot of investment and effort and work to get the product station, is that it’s a very good time to go back to the drawing board and re architect for an optimal radio performance.
If we were sitting in the same process, you know, yeah, there’d be a challenge. You have to go back and do a lot of work to add these extra bands. But here, because we were opening everything up and starting from scratch, it made it a lot easier to go and add a bunch of different spectrum because that was on the drawing board from from day one.
Does that mean that these same radio can do everything from the low end of five gig all the way up to the low end of seven?
Correct, probably. There’s some capability. The lowest frequency is the Japan band. Probably still support for that band in there. And then all the way up to seven point two gigahertz.
Impressive. So when people say, is this a single radio to dual radio or tri radio, really the five gig to seven is just one radio threat. And since you’re on the mobile side, on the AP side, we have a whole bunch of different designs, depending on who the market is, where there’s this Soho market enterprise and if it’s a transition in the mobile space. And then I’ll just pick an iPhone. I’m sure there’s Android phones. We’ll be the same. Well, we have phones of the future that can do everything from the low end cellular depending where they are in the world, to two point four CPRS in the middle, five and six. All on the same phone.
Yes, yes, absolutely. There are obviously multiple different architectures that you could come up with, multiple different radio pictures that you’ve come up with. But cell phones in particular are extremely size constrained and cost constrained. And so the optimal solution for that space is to have a dedicated single radio that operates in five and six gigahertz because of it being contiguous spectrum that turns out to be the main driver behind the architecture in that space. In access points where you have a little bit more space on your platform, you have a little bit more freedom in terms of cost, where Wi-Fi is really the key feature of those platforms. There can be a little bit more flexibility in terms of architecture, and you already have tri band routers. So sort of a trivial modification to make those tribe and routers into two point four, five and six on each of those three bits, thinking about the the insides of a phone of a mobile phone.
There’s radios and radios have associated and tennis. Is that something that Broadcom does? Do you recommend certain antenna pairings with your radios? Or is it something the vendors do on their own? And you just pick it up from the antenna side.
There was a time where we drove a little bit of the antenna design, but that’s so long ago, it’s not really something we do anymore. Once you got to a point where you needed for cellular antennas inside of phone industrial design, antenna design became an incredibly complex process that really needed a lot of specialization and any OEM that’s going to build a phone. So it’s not the kind of thing that you could ever really leave up to any of your vendors because you have this need to balance out intendant gain and antenna isolation between each of the different subsystem. That’s something that’s become very, very challenging design task and a real area of specialization for for smartphone apps.
It is quite a difficult thing. Every one of them has a different frequency, which means they have a different optimum size and they’re not the same. Yeah, absolutely.
And you have to figure out how you can combine different subsystems to use the same antenna. If you put you WB in your phone, is that gonna be combined with six gigahertz Wi-Fi? How will you combine L.A. with five gigahertz Wi-Fi? You know, you have a ton of questions of that nature. So there’s a lot of complexity in terms of industrial design and tenant design, front and design, filter design. It’s really a huge challenge. It’s a massive, massive solution space. I guess you would say come and figure this out. And I think I knew it was probably.
Maybe eight or nine years ago, I went to talk to a friend of mine who was a cellular and RF architect, and he had this sheet of paper sitting down in front of him that had what looked to me like fifty six different bands with tons of filters, amplifiers, antennas.
I was just blown away by the complexity of what he was imagining. So this is just what a standard phone is going to look like in a couple of years. I think at that point I knew that whatever we could contribute on the Wi-Fi side, we were going to have to cut it off at the output of our PS and hand it off to put it.
Let them deal with that on the other side. That’s. And two years ago, we used to be able to get a cell phone that could do F.M. as well as Wi-Fi and G.P.S.. How many of those those other features? Because you just mentioned you WB. That’s just one of many other radio things are happening inside these little handheld computers. Where do you do the handoff within your chip design from what’s Wi-Fi? What’s cellular? What’s neither of those is it or is that all inside of a single chip?
So we have had a lot of different technologies at Broadcom over the years and we’ve tried to do all kinds of different combinations. We sell a lot of Genesys products. We’ve tried to do combo devices with D.a.’s. We had a famine, a lot of products with NFC. So we pulled all of those subsystems into an overall combo chip, quite frankly, too. But ultimately, what we found is the smartphone. Williams wants a little bit more flexibility in terms of how they build Jenice s and NFC, where they place those within the phone and does obviously not as much uptake on UNEF m anymore as there used to be.
Things have kind of been narrowed down to Wi-Fi plus Bluetooth. Cellular has always been a separate subsystem.
I’ve seen lots and lots and lots of proposals to go combine the radios, combine the base bands, and nobody has ever really done this in a meaningful way. Some other vendors may do a different partition between radio and baseband than we do, but ultimately Wi-Fi and Bluetooth have kind of found a space where they are most naturally isolated together away from the other wireless products.
That does allow you to focus at least on only those two you have to deal with in your world, with mobile world. Battery life is paramount. Unlike the enterprise access point, I know they’re always plugged in. What kind of tradeoffs you have to make in design?
When you when you look at that battering usage, I think one of the biggest things that we introduced many, many years ago was the concept of offloads. There used to be a notion that every function needed to take up some MIPS on main applications processor SAAC. And we basically developed the software architecture where a lot of activities scanning, for example, or white listing a long, long list of things were completely offloaded to our device. And so the whole rest of the phone was off then, only our devices on. You know, obviously our power is going to be significantly less in a lower and low clock speed single core arm versus some for a twelve core Soucy somewhere. So that’s that’s a first question. The second thing is we go and we design around the modes that our customers end up using most of the time that we optimize for how their products are actually used. If you’re building an access point, you spend a lot of time transmitting. You spent some amount of time receiving. You really need to spend time optimizing your piece. But, you know, a PDA is basically trying to drive power into a 50 onload. There’s a lot of people say you’re limited by physics. There is only so much you can do on the phone side. You spend most of your time listening. We brought in a lot of innovations in our receivers to lower the power consumption that we use while we’re listening to the media. We really driven that power down. So that really gives our customers the longest possible battery life while using Wi-Fi. The critical observation was that’s the main mode that phone operates in. And the critical design decision is to go and optimize that power.
Since your chips are doing Wi-Fi and Bluetooth and both of those happen simultaneously. How do they go about sharing the antenna that listens to build Wi-Fi and Bluetooth at the same time? You can be obviously on the phone and listening to music, to your headset on all the same time.
This is something we’ve been working on for over a decade, for the most part there. You’re not operating simultaneously. You give some amount of time to Bluetooth and then you give some amount of time to Wi-Fi and you go back and forth. And unless there’s some pressing operation that one or the other subsystem needs to complete, you simply cycle back and forth. That ends up behaving pretty well. I think the key observation there is just that you need to have a different set of rules for cycling back and forth with. Listening to stereo audio on Bluetooth, you’re doing a phone call or you’re using a keyboard or a mouse or some kind of a Bluetooth controller. You know, we’ve gone down the path of just optimizing for all those different configurations with the way that you want to use Wi-Fi as well.
And does that change dynamically? If I have the same phone, but I connect up a mouse to it. Is it now adjust that algorithm accordingly?
Obviously, there’s a different perception of latency when you’re using a mouse or a keyboard versus when you’re listening to audio. Similarly, Bluetooth phone calls, for example, are extremely unforgiving in terms of losing time slot. And when the medium must be available.
Compare that to human interface device, which gives you a why greater bag in terms of when you actually have to respond to or comparing to an email download that you don’t have to worry about it.
And so, you know, I mean, Wi-Fi as well as is very is very forgiving. Right. You have a lot of situations where you’ll tolerate very large fluctuations in your instantaneous Wi-Fi throughput.
We can make a lot of intelligent decisions around what kind of throughput we’re going to allow, given the priority of the Bluetooth packets and how the user is going to perceive that Bluetooth.
I’m just intrigued with this time slicing. You’re actually popping back and forth between Bluetooth and Wi-Fi on the same infrastructure in the way you discuss it. It’s like it’s a really long time. But these are very, very short slices that you’re talking about.
Yes, they’re extremely, extremely short slices. If you look at a phone call, you’ve got extremely short periods of time, extremely short slots, hundreds of milliseconds to switch back and forth. You’re looking at a keyboard. You may have a little bit a little bit more time. It’s time slicing very, very small periods.
And I can see you sitting around in a conference room talking to people about these like to actually exist in human time when as a human, you could do all of that, switching back and forth before we even notice. But then again, I see that also in your chip manufacturer, that when you were talking about the nanometer size, we can’t even imagine how small that is. And yet you went, oh, yeah, we got this much smaller. Yeah, absolutely.
And certainly not something that’s visible to the to the naked eye or really meaningfully perceptible to people. The only thing that probably makes sense in a in a macro sense is how big the chip is. And so we started out with functionality that might require 40 square millimeters. We were able to drive it down to 20 square millimeters and then add a bunch of functionality and still end up with a smaller dioxides. And that’s the kind of thing you can see with the naked eye. And I think below that is microscopic.
And yet we talk about it like like we talk about how it goes this and it slices over there and it goes down this little path when they’re just invisible in an unprocessed tool to humans. We have to do that in order to make them and do what they’re doing. You’ve been obviously working for years now on six gig, didn’t know exactly when it was going to be approved, both in the U.S. and worldwide. When do we start expecting to see chips that actually end up in product that we can go and buy and talk six gig or the way the Wi-Fi Lance calls it sexy.
Broadcom has announced a broad portfolio of Wi-Fi sexy products. Other companies have as well. We ran an Interop trial until a couple of months ago. We all have these Wi-Fi, 60 products in the pipe. I think you’ll start to see the first products in the markets in Q4 of this year. Access points, they were faster time to market. And so they can pick up Wi-Fi, 60 chips today and start shipping them. I think in terms of seeing a phone, you’ll be waiting until early 2021, but you’ll see a full ecosystem in less than a year. You’ll be able to have a full Wi-Fi 16 ecosystem in your home. Very good.
That’s pretty impressive. I’ve seen presentation by some from Broadcom last year about how they’re going to use the different bands in houses. We normally in the wireless LAN professionals in our audience, deal with more of the enterprise class. But there is a huge market out there for the Soho class that could use maybe six gig is the mesh or as a backhaul and then still leaves the clients on the different side. So I think we’re gonna see a lot of different options coming in the next year.
Well, you know, it’s funny, I never really thought about this all that much until I ended up working home for the last couple months while my wife and I have a number of different networks set up in my home. You know, I moved her her laptop and all of her devices onto one one network.
And I put mine on her on a different channel through running into significant bandwidth limitations. When you’re running on the same AP or even the same mission.
I think that’s going to change for a lot of us as we move into six gig. Well, Gabe, thanks for your time. I appreciate you sharing a little of your information and knowledge about how the chips are done and especially from the mobile client side.
Sure. Absolutely. And you or anyone else feel free to reach out to me on Twitter. All there is behind the net, and I’m happy to answer any questions anybody has about why fire Bluetooth or really any other wireless technology or professional sports. Any time anybody has a question.
That’s great. We’ll add your link in the show notes and hope to talk to you soon.
Fantastic. Thanks for having me on. Great talking to you guys. Great to meet you guys.
Thank you for joining us for another episode of the wireless LAN professionals podcast. The podcast for wireless LAN professionals by wireless LAN professionals. Be sure to follow us on Twitter at Wireless LAN Prose. For all the latest news and updates and also connect directly with Keith on Twitter at Keith Parsons.
Head over to w w w w LAN prose dot com. For this episode show notes as well as the latest in all things Wi-Fi.
Automatically convert your audio files to text with Sonix. Sonix is the best online, automated transcription service.
Sonix uses cutting-edge artificial intelligence to convert your mp3 files to text.
Manual audio transcription is tedious and expensive. Are you a radio station? Better transcribe your radio shows with Sonix. Sonix takes transcription to a whole new level. Sometimes you don’t have super fancy audio recording equipment around; here’s how you can record better audio on your phone. Automated transcription is getting more accurate with each passing day. Lawyers need to transcribe their interviews, phone calls, and video recordings. Most choose Sonix as their speech-to-text technology. Better audio means a higher transcript accuracy rate. Do you have a podcast? Here’s how to automatically transcribe your podcasts with Sonix.
Sonix uses cutting-edge artificial intelligence to convert your mp3 files to text.
Sonix is the best online audio transcription software in 2020—it’s fast, easy, and affordable.
If you are looking for a great way to convert your audio to text, try Sonix today.