Long Live Small Cells and LTE…LTE-U is a Terrible Idea at Best

Posted · Add Comment

Quick refresher: LTE-U (LTE Unlicensed) is a scheme to aggregate the “free” 5GHz spectrum typically used in WiFi with LTE at lower “Wide Area Network” (WAN) governmentally auctioned frequencies such as 600-700MHz, 1.8-1.9GHz, 2.3GHz etc etc…

lte-u-img-Image from Qualcomm

Image from Qualcomm

The business drivers are Qualcomm (more chip$) and the network carriers which are looking for relief from network congestion and ability to sell more throughput to the end users. To increase profitability, they are using ‘free’ spectrum and selling it back to LTE end users.

I have always thought this to be a dumb idea and expected it to die many times, but $ can move mountains. Top of head, here are some reasons why it’s not a good idea…

c7859b564a73c4dd70031f19259d5b03

1.  Devices need to be engineered to:

(A) support those bands, which band adding real estate is already at a premium

(B) manage intermodulation

2. The system needs to manage interference to WiFi. Not fair if your free WiFi service is severely limited by companies selling the free airwaves.

3. The leading proposed methods of interference management, are not perfect by a long shot, all have some Listen Before Transmit (LBT) principles… specifically:

(A) Carrier Sense Adaptive Transmission (CSAT) – basically turning the transmitter OFF/ON in timed cycles to explicitly control the interference time.

(B) Licensed Assisted Access (LAA) – ugh, basically making the Licensed eNB (basestation) the anchor.. primary carrier licensed and secondary free…duh.

(C) Qualcomm’s proprietary MuLTEfire technology – which just runs LTE over WiFi spectrum with no outside spectrum.

(D) LTE-WLAN Radio Level Integration and Interworking Enhancement (LWA) – a little better… operates a WiFi channel but does an amazing amount of network accounting and management tricks to make the WiFi channel an exact proxy for an LTE channel. Pave that cowpath!

Let’s be clear, these things cost money and 3/4 of them degrade WiFi.

Maybe I have an axe to grind. My mindset years ago at the introduction of LTE and thinking about using them with small cells was, hey we could integrate WiFi into a small cell and aggregate (logically) and have more spectrum. So an small cell would control device assignments over Wifi channels. Not at huge amount of stuff to engineer. But this LTE-U thing is a different deal- it’s an LTE Physical layer (PHY) over free spectrum. That’s like engineering a special Redbox (Video rental) machine that holds specially engineered DVD form factor media that’s actually a USB to deliver your (normally) streamed Netflix over DVDs. Oh yeah, don’t forget you need a special machine at home to watch these too, but hey, it’s all video over DVD right?

Chart warfare

Qualcomm advocates the benefits:

qcom_LTEU_gainSeems to indicate WiFi throughput gains from not having WiFi neighbors.

qcom_LTEU_conclAnd everybody wins with this idea.

Meanwhile Cable Labs testing indicated something less beneficial.

Duty-Cycle-Fig2

Duty-Cycle-Fig3

wi-fi-duty-cycle-lte_fig31-2

 

This says that more LTE = worse WiFi (as expected) but even the minimum level has a negative impact. Ouch…

Business Case?

Gods+huh+as+in+plural+sounds+like+heresy+to+me+_15d7d1c2e6b51e230c244f54e423b763

Looking at it from a different dimension, remember, unless the spectrum rules are changed, the LTE eNB (base station) won’t be able to transmit it’s typical 20W+ for this spectrum in the majority of the universe so this limits the use to small coverage areas/indoor situations.

So were looking at an effort focused on small cells/WiFi Access Points (APs) and new mobile devices.

The cost of developing a new standard (LTE-U) is uncertain as most of it is private, so let’s make some assumptions. Firstly aggregate all the duplicate efforts at the different OEMs and network operators and assume it conservatively at $100M (20 firms worldwide @ $5M USD each) for design, meetings, marketing, development, prototyping, testing, interoperability  etc…

Additionally, QCOM, Broadcom,  GCT, Altair, Sequans etc… will need to spend another $100M (10 firms worldwide @ $10M USD) to create chipsets for mobile devices and basestations.

The small cells market in 2018 is roughly $3B USD according to Infonetics. So let’s say they used a unit value of $300/ea, that would suggest about 10M units. Let’s add in the $100M of basestation costs so now we have to buy $3.1B of small cells. Doesn’t sound like much but take it from me, trying to advocate the first small cells to carriers, $10 dollars extra per unit was a huge deal. Remember, these are the best case numbers and assumes massive quantity discounts. First movers will pay much much more for this luxury.

This goes on and on as there are additional labs, deployment expenses, test and measurement tools, and additional device expenses….

This is just a tax, and has to be paid by someone. The business case appears to be to shake mobile users down to pay for more LTE data per month that happens to be running over free spectrum at the expense of Wifi while sort of blocking WiFi operators.

Strategically maybe a pre-emptive strike against the Internet of Things/Internet of Everything market.

So…

This approach is a VERY BAD IDEA, my MAIN fundamental issue is the end user doesn’t need to be charged to use free spectrum and this SURELY shouldn’t happen at the EXPENSE of innocent WiFi users. It is costing and will cost a lot of money to overcome political, legal and engineering issues and the end users will get the bill for this. The benefit to the end user is questionable, any gain in LTE speed is offset by losses by Wifi and additional costs of devices, small cells, service increases etc….

Now the second major issue is this is not necessary (end users are going to be throwing money at carriers to get the opportunity to spend more money per month for faster LTE while indoors, spending more on devices, more on access points vs just WiFi 802.11 say AC.) Seriously, mobile devices normally search for WiFi channels for data transmission by default. This is the experience everyone is familiar with. Network carriers haven’t seemed to crack the “who pays for the small cell?” problem. This seems like some serious running in place.

I’m a huge LTE and small cells advocate and have been since pre-commercial launches. This idea is 6 years TOO LATE! It WAS reasonable BEFORE WiFi took the mantle as data throughput problem solution. The standards war is over at home, the carriers moved too slow on small cells and also thanks to large OEMs egging them on, fearing infrastructure revenue leakage.

Looking at the next near future blips on respective roadmaps, PHY layer of LTE-Advanced with OFDMA (8×8 peak throughout roughly 1.6Gbs) is not better in the small coverage environment than the PHY OFDM of 802.11 AC WiFi (peak ~5.3Gbs.) It is not a meaningful replacement for access point sized coverage areas. For example if a home user happens to have 1Gbps (!!) broadband to use as backhaul, there is no noticeable difference to Facebook, Youtube etc…when attached to either.  Small venues may get a 10Gbs backhaul and AC can outperform LTE-Advanced in this case. There is just not a strong technical merit that you can point to and say, “hey this makes sense.”

The LTE-U approach strikes me as a denial of service (DoS) attack on Wifi from LTE. Commercially, it’s just network operators reducing competition by eating the resources from independent WiFi operators. It doesn’t make the world a better place.

As a side note, I would be curious as to how Cisco plans to deal with this long term because it is an attack on a market they dominate and their market position on the attacking side is currently negligible at best (not including core network products such as routing etc…) There is a blog entry by Cisco’s Paul Mankiewich in 2014 that is mostly neutral on this topic. They are participating in both sides of standards development but undoubtedly they are motivated by their network operator customers to participate with LTE-U. They are trying to grow their small cell business after all. The highly competitive Asian companies/GovCos will happily export boxes so there will be product availability to be sure.

One last point… This is a LTE Release 13 technology while most infrastructure is between Release 9 and 10 so there is at least a 2 year window before we see widespread availability on the ground. If 802.11 stood still there would be no meaningful advantage at that time (2019ish) however there are 2 evolutionary paths for 802.11, AC has an evolution path and AX looks to be ratified 2018-9ish and is taping out to be more efficient with throughput peaks a bit higher than LTE-Advanced Release 13. 802.11AX is much like an evolved WiMAX which was effectively the precursor to LTE.  Best case the timing doesn’t favor LTE-U at this point.

One more final itty bitty musing….I remember 1xEV-DV had this same timing issue and was bypassed by 1xEV-DO thanks to Qualcomm’s revenue focus… 1xEV-DO became irrelevant very quickly in terms of global network technology adoption. The best technology doesn’t automatically gain market leadership but lesser ones get thrown out or disrupted, most of the time quickly.

Alternatives? 

  1. Increasing throughput such as antenna and SON improvements as it’s less costly and requires no changes of user behavior. (Get more out of what you already have anyone?) 
  2. Invest in improving and deploying more small cells (This is close to the ultimate solution thanks to the laws of Physics.) Example? Release 10 LTE small cell using licensed spectrum (in home, SOHO, small venue etc…) TDD, with 8x MIMO is very competitive on a throughput basis with 802.11 AC and it has been on the roadmap for ages. No pixie dust required to make this happen!
  3. Create more interesting IEEE 802.11 standards that further improves interoperability with WiFi. 
  4. My gut tells me that it’s probably a stronger case to use 802.11ac as a small cell backhaul, and there are several significant scenarios where using 802.11AC or AD as an ‘in the field’ X2 mesh likely blows away the business cases compared to LTE-U small cells. Yes, LTE-U Interference will limit the benefits of 802.11 backhaul.

In the end, I’m anticipating some technical push back with a reciprocal proposal to allow WiFi over (under used) licensed spectrums. What’s the difference? If we can duty cycle free spectrum, why not duty cycle paid spectrum? If LBT is so bulletproof it should work both ways, right?

Network operators, no!

The business case for providing networks is to make the network available and handle as many users as possible with few resources…. similar business case as an airplane/train/bus vs a car. However in this case the network operators are striving to be the car and WiFi is the bus. LTE operators are defending against WiFi encroachment vs growing their capability and becoming more compelling for end users.

Misguided this is.

Yoda_SWSB

Additional information:

FCC Letter.

LTE-U industry forum.

This is Qualcomm’s presentation submitted to the US Government… shows all the happiness with both…

LTE-U Forum interoperability proposal.

3G4G blog..

Cable labs testing blog.

WiFi Forum page on interoperability.

3GPP Release 13