Mobile Broadband Working Group Report



Mobile Broadband Working Group ReportThe goal of the group is to generate a white paper.The mobile broadband ecosystem depends on the facilities owners continuing to invest in upgrades for capacity and capability. Investment must be justified by a return on investment and overall profitability. ISPs are asking, what happens if our investments make someone else like Netflix or Microsoft more money?Could broadband access stagnate? That question motivated us.There are important differences between wireline and wireless. There’s been lots of wireline investment in the U.S., driven by competition. The cable guys might just be winning. (Fios is suspended, U-Verse has suspended) Might this stimulus fade?In Europe, there’s facilities unbundling – those with copper pairs forced to sell to competitors -- has driven retail costs down, but reduced the incentive to invest. Shared infra might be the answer (voluntary unbundling?)In mobile world, there is both competition and higher costs. Hypothesis: the mobile world may be a more challenging case to study. (Although a lot of this applies to wireline.)Three related issuesGrowth in demand (is raw capacity the only issue?)Cost of usage Sources of revenueGrowth – Cisco’s predictions are looked at by device. There are a few interesting things about this picture.This is a picture drawn by someone who wants to sell equipment. It’s a prediction not a fact.The graph shows exponential growth. Andy said scalability is not a point, it’s a curve. If you’re facing exponential growth, regardless of the multiplier, you need an exponential response. Growth is due to both new users and increasing usageSmart phones will continue to dominate – the puzzle is whether we’ll see a bigger adoption of bigger form factor displays. I’ll come back to thatWhat are people doing?Data from Sandvine and Ericsson shows that most mobile traffic is YouTube (not Netflix) –people are watching short form video, but they’re watching long form content on their tablets. This picture could change if there were more tablets in the ecosystem. But if smart phones are the dominant device and people are watching videos on them, then they will be short form. I’d like to get some of the media and content people to react. Seems that the platform or the device will shape the content experience.In the developing world, the Internet is accessed primarily by a mobile device, and Facebook is trying to define that experience by making sure that when you look at the Internet, you come in via Facebook. There are some implications there regarding what kind of content will be made and what people will be looking at. Lots of radio consumption.Facebook has been isolated from HTTP – maybe with an app like zero.facebook they can redefine the Internet experience as the Facebook experience.Sandvine growth predictions are slightly lower than Cisco, and their data is based on class of behavior (by 2018, real time entertainment will be 2/3 of the behavior (in the mobile world).Another set of data by Sandvine compares mobile vs fixed behavior – mobile download is 1/3 of wireline. Mobile usage is 1% on a monthly basis. Relates to what Hannu was saying – we’d like to get cellular to 1 G a day. Unless you’re going to hand off 90% of your traffic to WiFi, wireless will never match. Except in a world where there is no wireline, wireless is more of a complement than a substitute. But that means we’re going to break the world up into 2 parts – mobile and mixed. Compare US to EU – notice that mobile isn’t that different but in wireline they are very different. Why? Could be Netflix – up to half the traffic in the U.S. It could be that EU is in for a big jolt. They haven’t had the sudden transition to all this online video. My conclusionsGrowth is not exogenous. These graphs are part of an ecosystem.What’s going to shape growth? Netflix is important. Re QoE, people don’t use things that don’t work well. In the innovation space, people introduce apps when the network works well enough. The growth will be gated by how fast the network improves.Example: Dina Katabi’s work -- I love what Dina is doing. With MegaMIMO we can get a factor of 10, which is 5 years according to Moore’s Law. She’s basically saying, I can get you a one time, 5-year excuse… that’s not to deprecate the idea it just means it reduces the rate at which you need to respond to the exponential demand. But in the end you need an exponential response. In the case of mobile, spectrum efficiency doesn’t scale exponentially. The only exponential response is smaller cells. This is an intriguing space for 2 reasons. 1) when the cells get small enough what you’ve built or depended upon is a wireline infrastructure 2) it’s going to change the character of a cell. Could this be a Clay Christensen event, similar to the disc drive story: smaller drives, people brought new products to market based on smaller disks – I see this as a challenge to our supplier friends. I was talking to Van Bose, he sells small cell base stations built out of sw radios, screws them onto telephone polls in Vermont to fill in wireless gaps, $2-300, plugs it into the Comcast cable, buys a biz service from Comcast. That’s a challenging cost point for someone whose model is that a base station is a brick building. Is this an opp for new entrants, but not in the operator business, rather in the equip business, to come up with a new model for what a cell looks like. Comcast could buy these.Next thing is the use of WiFi – putting traffic on the wireline network.At some point, will we saturate the market. In the developed world, the rate of new consumers has leveled off – we’re stuck with 20-22% of the population not interested in the Internet.Re cost – it’s more complicated in the mobile space than wireline. Not sure we know enough yet. It’s clear that upgrades to LTE generate one kind of cost structure; you’ve already got the tower, you’ve already got the generators. Capacity upgrades to existing cells – you need more backhaul capacity, that’s another cost structure. And if you’re putting in new base stations, new cells, that’s another kind of cost structure. And someone pointed out that increasing demands has two effects. One is that your popular cells get overloaded, but the less loaded cells become more efficient. Wireline is engineered for usage, but mobile is engineered for coverage, so a lot of cells may be less loaded – what is the diversity of the load on existing cells? It’s complicated. What does the cost model look like? Models are hard to come by, they are influenced by accounting, they are proprietary. We hunted around the literature for cost models, they all take usage and think about it, as, not peak capacity, but behavior measured in gigabytes and what does it cost for a G of usage. Here are 3 quotes: it’s possible to get down toIt is possible to get down to a fully loaded network cost of less than EUR 1/GB at a level of around 15% average network utilization. Looking at capacity upgrades, it is possible to easily double capacity at a cost per GB of EUR 0.1 to 0.2.”–Greger Blennerud, Mobile broadband – busting the myth of the scissor effect, Ericsson white paper EBR #2 2010Operators can (and must, to remain profitable) reduce costs to $1/GB by 2013–0.1 cent per MB: ensuring future data profitability in emerging markets, McKinsey,RECALL_No_17, 2011”Monthly network Capital Expenditure (CAPEX) and Operational Expenditure (OPEX) can be kept below 3 EUR per subscriber over an eight-year depreciation period. This is true if the average mobile broadband penetration is at least 500 subscribers per site, and if subscribers use less than 2 GB per month.” “If total data use is high, …, the cost per GB can be below 1 EUR.”–Mobile Broadband with HSPA and LTE—capacity and cost aspects, Nokia Siemens Networks, White PaperThere are more dimensions to the cost issues around mobile than around wireline.But mobile depends on wireline.Estimates of total cost for wireline are around $0.10/GB—one tenth of mobile.But incremental might be closer to wireline costs.RevenuesA recurring conversation centers on the sources of revenues for ISPs.Today the consumer pays essentially all the costs of the access provider.Is there an alternative frame in which other parts of the ecosystem contribute to the cost of access?Advertising?Paid content?Other sources?See previous talk.Usage-based billing.Creates positive incentive for ISP, negative incentive for user.Zero-rating?Charging for priority?A way to benefit from ecommerce.Cost vs RevenueCurrent pricing for additional usage:Verizon: $5/GBAT&T $10/GBLooks like plenty of incentive for U.S. providers to facilitate usage.But how much is user discouraged by price point?Audience: you made the hypothesis that the usage rates in the EU are lower than US. It’s been my experience that they’re higherDave Clark, MIT: I got that data from Sandvine. If you have another data set, I’d love to see itAna Serrano, CFC: In terms of the mobile usage, the data is per household, but it doesn’t break it down per user inside the household. I’m curious to know what effect the usage of kids under 8 for example is, given that the smart phone is being used in unanticipated ways. It’s now the new babysitter, etc.Dave Clark, MIT: I’m curious too, but the data comes from DPI, they don’t know anything about demographics. You’d have to cross reference with demographic data. They don’t have access to that. It’s a complicated but intriguing question to answer.Dave Oran, Cisco: I saw a usage graph years ago that showed P2P was going to drive growth, but now it looks like it’s almost gone. How can we understand these phase changes in traffic and usage patterns? Are we destined to see streaming video follow through its trajectory without another phase change?Dave Clark, MIT: I would be highly surprised if the future didn’t surprise us. Somebody could come up with a different way to delivery a video experience that would change the profileBill Lehr, MIT: There’s still a lot of P2P traffic, but relative to the growth of vide it drops out. Dave Clark, MIT: According to this, Bittorrent is about 5% of fixed traffic.Dave Oran, Cisco: I suspect it’s going down because, all the arguments about P2P being about technology and not about piracy turned out to be false – it was all about piracy not technology. The minute all the content became available legally for a reasonable price, the piracy dropped dramatically.Audience: My experience was you couldn’t get it any other way. But now with legal options, that’s what people choose.Dave Oran, Cisco: Or the piracy has shifted to streaming sites because the enforcement was directed at P2P where as now, piracy has shifted to the streaming sites.Dave Clark, MIT: There is a certain amount of password sharing on Netflix. You can have 5 devices foo one account.Bill Lehr, MIT: Re what’s going on in the household, the focus is on mobile, not mobile devices. A lot more of the fixed traffic comes from mobile devices within the home. When you think about those devices, how much of it is stuff in the house, which doesn’t show up in the mobile traffic – that effects what you want to do about spectrum and the device economics.Sharon Gillette, Microsoft: I’m curious about the effects of cloud services – does it have an effect on traffic distribution?Dave Clark, MIT: this data has upload data on it. Sandvine has upload data. But notice, it’s tiny. In the mobile space, it’s 15% of download. Sandvine breaks that out – they think its XXX.Sharon Gillette, Microsoft: The question is whether it’s changing (the percentage of download vs upload). Download has been In Europe there’s a discussion as to whether repurposing broadcasters should be big download pipes – does this make any sense at all?Sharon Gillette, Microsoft: My other question builds on your observation re two divergent futures – one where wireless is complementary to wireline and the other, in the developing world, where it is the primary technology because there is no wireline. I’m interested in the second – how do you get Internet to people living on less than $10 a day, it seems like mobile is the more expensive and the more limited technology, so those seem like two contradictory ideas. What are your thoughts?Dave Clark, MIT: In the limit, a cellular network is a wireline network with radios on the end -- fiber to the mobile – so you could ask why is it that the cost of usage is so high for mobile? That’s the question. It could be because base stations are expensive without a lot of capacity per base station, you have to put up a lot – which is why I said there may be a Clay Christensen event, where the way you drive the cost down is do fiber to the mobile but the base station costs $1000 and maybe with abundant backhaul (so you’re not buying DS3s or something like that—and clearly in the developing world, most people are giving up and hauling the fiber so t hey get infinite backhaul… I have a feeling there could be a disruptive cost event. But I agree there’s a disconnect that needs to be unpacked. Audience: what about WiFi?Dave Clark, MIT: WiFi is the extreme of putting a radio on the end of the wireline network, and we know what the cost point for that is. You can buy a base station for $150. The consumer is capitalizing that infrastructureTony Tauber, Comcast -- You can get them for $15 now.Sharon Gillett, Microsoft – but it still has to hang off of something. I should have said fiber to the mobile and fiber to the wifi – but what does the wifi hang off of in a world where there’s no wireline infrastructure?Andy: If you populated the developing world with hot spots you could balance the load between carrier-based wireless and hot-spot based wireless in a way that might shift the economics. Is that reasonable?Sharon Gillett, Microsoft: It is, but I have a different scenario in mind, e.g., remote villages where there’s nothing there. Not an urban area.Andy Lippman, MIT: right. Dave Clark, MIT: Clearly in the dev world backhaul is an issue, but the one advantage they have is cheap labor.Audience: One quick point to add, I’ve heard that one problem in the developing world is that deploying wires, physical infra is more difficult to secure because XXX get stolen as you deploy them, so wireless is easier to deployDavid Clark, MIT: I just talked to someone who is part of a public/private partnership to develop fiber infra in the poorer parts of working in Johannesburg and they’re putting them all underground – they rip them off before they realize there’s no copper, so it’s better to just put it underground. He said, they steal the manholes then. It’s worth the cost of ditching to prevent stealing. ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download